diff --git a/CHANGELOG.md b/CHANGELOG.md
index a2841373..252cb4b5 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,6 +8,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [Unreleased]
 
 ### Added
+- **StackedDiD `vcov_type` parameter (Phase 1b PR 2/8).** `StackedDiD(vcov_type=...)` now accepts `{"hc1","hc2_bm"}` (defaults to `"hc1"`, preserves prior behavior at machine precision — WLS-CR1 sandwich is algebraically invariant between the prior bake-Q-into-X pattern and the new `solve_ols(weights=composed_weights, vcov_type=...)` path, differing only by float64 multiplication ordering at ~2 ULPs at SE scale; pinned by `test_hc1_se_bit_equal_to_pre_pr_baseline` at `atol=1e-13`). `classical` and `hc2` rejected at `__init__` with cluster-incompatibility `ValueError` — StackedDiD clusters intrinsically at `'unit'` or `'unit_subexp'` (no `cluster=None` opt-out), so one-way families are not composable with the linalg validator's cluster_ids check. `hc2_bm` routes CR2 Bell-McCaffrey through the clubSandwich WLS-CR2 port (PR #475) — matches `clubSandwich::vcovCR(lm(weights=Q,...), cluster=~unit|unit_subexp, type="CR2") + coef_test()$df_Satt` at `atol=1e-10` on the new `benchmarks/data/stacked_did_golden.json` fixture (6 R-parity tests in `tests/test_methodology_stacked_did.py`). HC1 + cluster matches `clubSandwich::vcovCR(..., type="CR1S")` (Stata-style `G/(G-1) * (n-1)/(n-p)` correction; plain `type="CR1"` omits the `(n-1)/(n-p)` term and would diverge by ~1.4% on the test fixture). **Bell-McCaffrey Satterthwaite DOF is threaded into the user-facing aggregated inference for hc2_bm**: per-event-time `event_study_effects[h]['p_value']/['conf_int']` use the per-coefficient contrast DOF from `_compute_cr2_bm_contrast_dof`; `overall_p_value`/`overall_conf_int` use the post-period-average contrast DOF, matching R `Wald_test(test="HTZ")$df_denom` at atol=1e-10. Without this threading the small-sample inference would silently fall back to normal-theory (df=None) — mirrors the SunAbraham aggregated-inference pattern from PR #472 and addresses the R1 CI codex P0 caught at submission time. `vcov_type` propagated to `StackedDiDResults.vcov_type`. `SurveyDesign` combined with `vcov_type != "hc1"` raises `NotImplementedError`: survey TSL/replicate-refit overrides analytical sandwich. Reject order locked: fweight/aweight check at `stacked_did.py:309` fires before the survey + non-hc1 check at `stacked_did.py:~325` (pinned by `test_aweight_plus_hc2_bm_rejected_by_stacked_did_level_guard`). `conley` rejected with a deferral message. `weight_type ∈ {"aweight","fweight"}` + `hc2_bm` continues to raise per the linalg validator's pweight-only restriction in PR #475 (`vcov_type="hc1"` supports all three weight types via the existing StackedDiD Q-weight semantics reject). Second PR of the Phase 1b standalone-estimator threading initiative (6 PRs to follow: WooldridgeDiD-OLS, CallawaySantAnna, ImputationDiD, TripleDifference, TwoStageDiD, EfficientDiD).
 - **ContinuousDiD methodology-review-tracker promotion.** Tracker row flipped **In Progress** → **Complete** with full Verified Components / Test Coverage / Corrections Made / Deviations / Outstanding Concerns structure mirroring the HAD precedent (PR #473). REGISTRY `## ContinuousDiD` gains a formal Deviations block consolidating the boundary-knots deviation from R `contdid` v0.1.0 (`range(dose)` vs `range(dvals)` — library avoids extrapolation), the `bspline_derivative` derivative-failure `UserWarning` (Phase 2 axis-C #12), the `+inf` → `0` never-treated recoding warning, and the zero-`first_treat`+nonzero-`dose` force-zeroing warning (both axis-E silent-coercion fixes) into a single AI-review-recognized labeled surface. R cross-language coverage for ContinuousDiD runs at relative tolerance across two surfaces: (a) **scalar parity with raw R `cont_did` / `pte_default`** at 1% on overall ATT for all 6 benchmarks and on overall ACRT for benchmarks 4-5 (benchmark 6 is event-study, scalar `overall_att` only); (b) **harmonized boundary-knot-normalized curve parity** with R-side ATT(d) / ACRT(d) reconstructed under `Boundary.knots = range(treated_doses)` (matching the library) on benchmarks 1-3 via the benchmark harness — `_run_r_contdid` does the R-side rebuild at `tests/test_methodology_continuous_did.py:333-367`, and `_compare_with_r` orchestrates the Python-vs-R comparison at `:395-459` — max ATT(d) at 1% and max ACRT(d) at 2%. NOT bit-exact (`atol=1e-8`) like HAD — the boundary-knots deviation precludes algorithmic bit-equality on aggregated dose-response curves. Surface (a) is direct raw-package parity; surface (b) is reconstructed-basis parity because raw `contdid` curves use `range(dvals)`. No source code changes, no new tests, no new docstrings — consolidation only against the existing 15 methodology tests (`tests/test_methodology_continuous_did.py`), 80 unit tests (`tests/test_continuous_did.py`), and `docs/methodology/continuous-did.md` theory note. `METHODOLOGY_REVIEW.md` ContinuousDiD row promoted **In Progress** → **Complete**.
 - **`SpilloverDiD(vcov_type="conley", survey_design=...)` integration via stratified-Conley sandwich on PSU totals (Wave E.2).** Lifts the Wave E.1 `NotImplementedError` (`spillover.py:2201` upfront, `two_stage.py:217` helper-level) and adds spatial-HAC + design-based variance for the previously deferred composition. **Documented synthesis** of Conley (1999) spatial-HAC × Gerber (2026, arXiv:2605.04124) Proposition 1 Binder TSL (the Wave E.1 foundation) × Wave D Gardner GMM first-stage uncertainty correction (Butts 2021 §3.1 + Gardner 2022 §4) applied to SpilloverDiD's ring-indicator stage-2 design. No reference software combines all three ingredients on a two-stage influence function. **Mechanical composition (panel-aware):** preserves the library's existing `conley_lag_cutoff = 0` semantic at `diff_diff.conley._compute_conley_meat` ("within-period spatial only — exclude cross-period spatial pairs") by looping over periods. For each period `t`, SpilloverDiD's per-obs Hájek-weighted Wave D IF `psi_i` is aggregated to per-period PSU totals `S_psu_t[g] = sum_{i in PSU g, time t} psi_i` (via `np.add.at`); per-PSU spatial centroids are panel-constant (mean of per-observation `conley_coords` within each PSU, vectorized `np.add.at` sums / `np.bincount` counts); for each stratum the within-stratum sandwich is `M_h_t = (1 - f_h) * n_h/(n_h-1) * sum_{j,k in PSUs_h} K(d(centroid_j, centroid_k) / conley_cutoff_km) * (S_psu_t[j] - S_bar_h_t)(S_psu_t[k] - S_bar_h_t)'`, where K is the Bartlett kernel (SpilloverDiD currently exposes Bartlett only and hardcodes it; the survey helper accepts `"uniform"` too but exposing that on the SpilloverDiD constructor is a separate follow-up) and `d` is haversine / euclidean / callable per `ConleyMetric`. Cross-stratum kernel weights are exactly zero by sampling design (strata are independence partitions). Total meat is `sum_t sum_h M_h_t`. Cross-period spatial pairs are excluded by construction — the per-period loop matches the library's panel Conley contract exactly. **Reduction semantics (load-bearing for tests):** the orchestrator's panel-aware meat equals `sum_t` of per-period within-stratum stratified-Conley sandwiches on per-period PSU totals (pinned at `tests/test_spillover.py::TestSpilloverDiDWaveE2ConleySurveyDesign::test_b_panel_aware_per_period_sum_invariant`); single stratum (H = 1, FPC = inf) reduces to `sum_t` plain Conley sandwich on per-period PSU totals (NOT on time-collapsed totals). **Implementation:** new `_compute_stratified_conley_meat_from_psu_scores` helper in `diff_diff/survey.py` (parallel to existing `_compute_stratified_meat_from_psu_scores` 3-tuple `(meat, variance_computed, legitimate_zero_count)` contract; per-stratum loop replaces the inner `centered.T @ centered` with `_compute_conley_meat(scores=centered, coords=psu_coords_h, ...)` in cross-sectional mode); new dispatch wrapper `_compute_stratified_conley_meat` in `diff_diff/two_stage.py` (parallel to existing `_compute_binder_tsl_meat`, performs per-obs Psi → PSU aggregation + centroid derivation + dispatch to survey helper, intentionally drops `cluster_ids` at the dispatch boundary — see Restrictions). `_compute_gmm_corrected_meat` conley branch extended with `if resolved_survey is not None` routing to the new wrapper; the `resolved_survey is None` branch is bit-identical to Wave D. **Singleton-stratum `lonely_psu="adjust"` parity:** the survey helper mirrors the Binder helper's `continue` to skip the FPC scale on singleton strata (with `n_h = 1` the scale `n_h / (n_h - 1)` would divide by zero); the degenerate one-PSU kernel `K = [[K(0)]] = [[1.0]]` reduces to `centered.T @ centered`, matching Binder's singleton-adjust output. **Saturated `df_survey = 0` NaN-fail:** mirrors Wave E.1 (`_compute_stratified_conley_meat` returns NaN meat with `UserWarning` template "Wave E.2 stratified-Conley sandwich: df_survey = 0..." so callers can `pytest.warns(UserWarning, match="Wave E.2 stratified-Conley")`). **Public surface restrictions:** replicate-weight variance (BRR / Fay / JK1 / JKn / SDR) raises `NotImplementedError` (inherits Wave E.1 gate; per-replicate full refit is separate follow-up scope); `cluster=<col> + survey_design.psu + vcov_type="conley"` coerces `cluster=<col>` to PSU per Wave E.1's warn-and-use-PSU pattern (the Conley cluster product kernel becomes a no-op after PSU aggregation, so `cluster_ids` is intentionally not threaded into the inner Conley kernel call — every PSU is its own cluster post-aggregation, which would zero all cross-PSU pairs); LinearRegression-side `vcov_type="conley" + survey_design=` gate at `diff_diff/linalg.py:2853` remains (separate Bertanha-Imbens 2014 weighted-Conley "Phase 5" roadmap, not Wave E); DiagnosticReport routing for `SpilloverDiDResults(vcov_type="conley", survey_design=)` requires `_APPLICABILITY` / `_PT_METHOD` registration (separate Wave F PR). **Tests:** new `TestSpilloverDiDWaveE2ConleySurveyDesign` and `TestSpilloverDiDWaveE2ConleySurveyDesignEventStudy` classes in `tests/test_spillover.py` (bit-identical no-survey fallback; panel-aware per-period sum invariant on the orchestrator + helper composition; hand-computation methodology anchor; single-stratum ≡ plain Conley on PSU totals; cross-stratum independence as a unit test on the survey helper with interleaved cross-stratum centroids; Binder vs Conley singleton-adjust FPC skip parity; lonely-PSU sensitivity across three modes; FPC large ≡ no-FPC and FPC = n_h zeros stratum; saturated NaN-fail with `pytest.warns(match="Wave E.2 stratified-Conley")`; replicate-weight + non-pweight rejections; cluster warn-and-use-PSU; fit idempotency; `finite_mask` survey-array subsetting; no-PSU coverage — weights-only `SurveyDesign(weights=...)`, strata-only `SurveyDesign(weights=..., strata=...)`, and a per-period re-index unit invariant pinning that no cross-period spatial pairs leak into the meat on implicit-PSU layouts; event-study path on both `is_staggered=True`/`False` branches per `feedback_cohort_loop_trigger_cache_both_branches`; drift goldens at `rtol=1e-12 / atol=1e-14`). The pre-existing `tests/test_spillover.py::test_fit_conley_plus_survey_design_not_implemented` Wave E.1-era gate-assertion test is removed (replaced by the positive-path tests above). Wave E.1 entry's "Public surface restrictions" bullet updated to past-tense the conley+survey gate reference.
 - **`SpilloverDiD(vcov_type="conley", conley_lag_cutoff > 0, survey_design=...)` panel-block composition via spatial + serial Bartlett HAC (Wave E.2 follow-up).** Lifts the Wave E.2 upfront `NotImplementedError` at `spillover.py:2210` and extends the panel-aware stratified-Conley sandwich (cross-sectional `lag=0` shipped in Wave E.2) with a within-PSU serial Bartlett HAC over time (Newey-West 1987 form). **Documented synthesis** of Wave E.2's panel-aware Conley × Binder TSL × Wave D Gardner GMM composition with Newey-West (1987) serial Bartlett HAC, matching the no-survey panel-block decomposition at `diff_diff.conley._compute_conley_meat` (Conley 1999 + Newey-West 1987 separable form, NOT Driscoll-Kraay 2D-HAC). The composition is `meat = meat_spatial + meat_serial` with disjoint index sets: spatial is the shipped Wave E.2 per-period within-stratum sandwich on PSU totals; serial is the new within-PSU sum `meat_serial_h = FPC_h_panel * sum_{g in stratum h} sum_{|t-s| <= L, t != s, both periods present for PSU g} (1 - |t-s|/(L+1)) * S_centered_t[g] @ S_centered_s[g]'` where `S_centered_t[g] = S_psu_t[g] - S_bar_h(g)_t` is per-period within-stratum centered (Binder TSL form — matches the spatial helper's centering exactly), and `|t-s|` uses panel-wide dense time codes (mirrors `conley.py:940` R deviation that matches R `conleyreg::time_dist`). Serial Bartlett kernel is hardcoded regardless of `conley_kernel` (the user-selected kernel governs the spatial term only). **FPC convention (panel-wide per-stratum):** standalone Newey-West composition on stratified clusters — the serial sum is a PANEL-level construct, so the cluster set is the panel-wide PSU set in stratum h; FPC denominator uses `n_h_panel = |unique PSUs in stratum h across active sample|`, NOT per-period `n_h_t`. The spatial term keeps its existing per-period FPC unchanged. For balanced panels the two FPC denominators converge; the difference surfaces under unbalanced panels. Citation chain: Binder (1983) for the FPC factor form, Gerber (2026) Prop 1 for the Binder TSL composition with two-stage IF, Newey-West (1987) for the serial Bartlett kernel weights, Conley (1999) for the spatial kernel and panel-block decomposition (deliberately NOT by analogy to the Binder helper's cross-sectional per-stratum FPC convention). **Centering asymmetry vs no-survey reference:** the no-survey panel-block path at `conley.py:949-965` uses RAW scores for the serial term because it assumes `E[scores] = 0` under correct specification; the survey-weighted Binder TSL form centers explicitly (textbook stratified-cluster sandwich). Using raw scores in the survey case would inflate variance by twice the squared per-period stratum mean and would NOT reduce to the cross-sectional Wave E.2 form at lag=0. **Reduction semantics (load-bearing for tests):** `conley_lag_cutoff = 0` or `None` produces bit-identical ATT and scalar SE to shipped Wave E.2 (orchestrator skips the serial helper invocation; the spatial loop + saturation guard + new PSD/finite guard still run on the spatial-only meat). `assert_array_equal` regression pin at `test_a` covers user-visible ATT + scalar SE; `test_a2` mock-spy independently asserts the serial helper is NOT invoked at `lag=0`. The meat matrix itself is not exposed on `SpilloverDiDResults`, so full meat-matrix equality is implied (not asserted); `conley_time is None` or `T = 1` short-circuits the serial helper to zero meat (degenerate panel-block, not a saturation diagnostic); single-stratum H=1 with FPC=inf reduces to Newey-West Bartlett HAC on per-period within-stratum-CENTERED per-PSU score sequences (NOT raw scores — Binder TSL centering is retained at H=1; the panel-wide `G/(G-1)` survey factor replaces FPC); bandwidth → 0 with L > 0 reduces the spatial term to per-period within-stratum HC sandwich while leaving the serial term unchanged (separable form). **Singleton-stratum `lonely_psu="adjust"` panel-wide mean asymmetry:** for the serial helper, `_global_psu_mean` is the panel-wide mean of per-period PSU totals (averaged over all `(g, t)` with `present[g, t]`), NOT the per-period within-stratum mean used by the spatial helper. The `continue`-skip-FPC pattern matches the spatial helper at `survey.py:2007-2017` to avoid divide-by-zero on `n_h_panel = 1`. **Restrictions inherited:** replicate-weight variance still raises `NotImplementedError`; DiagnosticReport routing for the panel-block case is queued for the same Wave F follow-up as the cross-sectional case. **Implementation:** new sibling helper `_compute_stratified_serial_bartlett_meat` in `diff_diff/two_stage.py` (parallel to the Wave E.2 spatial orchestrator; ~200 LoC). Orchestrator `_compute_stratified_conley_meat` signature extended with `conley_lag_cutoff: Optional[int] = None`; spatial loop unchanged; serial helper called after spatial loop when `conley_lag_cutoff > 0`; saturation NaN-fail accounting merges both terms' `(variance_computed, legitimate_zero)` flags into the same template (`UserWarning` "Wave E.2 stratified-Conley sandwich: df_survey = 0..." covers both spatial-only and panel-block cases). Dispatch in `_compute_gmm_corrected_meat` conley branch threads `conley_lag_cutoff` through; the `cluster_ids` non-threading rationale (post-PSU-aggregation every PSU is its own cluster) still applies to the new serial branch. Spillover-side gate at `spillover.py:2210-2230` (Wave E.2-era `NotImplementedError` for `lag > 0 + survey`) deleted; gate rationale comment replaced with shipped-behaviour note. Stage-1 weighted FE solver, `finite_mask` survey-array subsetting, `df_survey` threading, bread weighting, and `SpilloverDiDResults` survey metadata are all inherited UNCHANGED — Psi construction is bit-identical regardless of vcov_type or lag. **Tests:** new `TestSpilloverDiDWaveE2FollowupConleySurveyLagCutoff` and `TestSpilloverDiDWaveE2FollowupConleySurveyLagCutoffEventStudy` classes in `tests/test_spillover.py`; existing `test_j0_panel_conley_lag_cutoff_rejected_under_survey` (Wave E.2-era gate-assertion) DELETED.
diff --git a/TODO.md b/TODO.md
index 6ca61bc0..a642394e 100644
--- a/TODO.md
+++ b/TODO.md
@@ -99,8 +99,9 @@ Deferred items from PR reviews that were not addressed before merge.
 | PreTrendsPower: CS/SA `anticipation=1` R-parity fixture. The PR-C R-parity goldens cover NIS power + γ_p MDV at `atol=1e-4` on four shifted-grid / regular / irregular / K=1 fixtures, but R `pretrends` has no anticipation parameter so the Python-side `_extract_pre_period_params` anticipation filter (`if t < _pre_cutoff` in `pretrends.py` lines 1138-1150 for CS; mirror in SA branch) is not R-parity-locked. Build a synthetic `CallawaySantAnnaResults` (or `SunAbrahamResults`) with `anticipation=1` and a t=-1 event-study entry that should be filtered before reaching `_compute_power_nis`, then assert the resulting γ_p matches R's `slope_for_power()` on the K=4 shifted-grid fixture. Existing PR-B MC-based tests (`TestPretrendsPropositions`) and full-VCV tests (`TestPretrendsCovarianceSource`) already cover the filter mechanically; this would close the loop against R. | `tests/test_methodology_pretrends.py::TestPretrendsParityR`, `benchmarks/R/generate_pretrends_golden.R` | PR-C follow-up | Low |
 
 
-| Thread `vcov_type` (classical / hc1 / hc2 / hc2_bm) through the 7 standalone estimators that expose `cluster=` but not yet `vcov_type=`: `CallawaySantAnna`, `ImputationDiD`, `TwoStageDiD`, `TripleDifference`, `StackedDiD`, `WooldridgeDiD`, `EfficientDiD`. Phase 1a added the chain to DiD/MPD/TWFE; Phase 1b PR 1/8 added `SunAbraham` (this row tracks the remaining 7). | multiple | Phase 1b | Medium |
+| Thread `vcov_type` (classical / hc1 / hc2 / hc2_bm) through the standalone estimators that expose `cluster=` but not yet `vcov_type=`: `CallawaySantAnna`, `ImputationDiD`, `TwoStageDiD`, `TripleDifference`, `WooldridgeDiD`, `EfficientDiD`. Phase 1a added the chain to DiD/MPD/TWFE; Phase 1b PR 1/8 added `SunAbraham`; Phase 1b PR 2/8 added `StackedDiD` (this row tracks the remaining 6). | multiple | Phase 1b | Medium |
 | Extend `SunAbraham` with `vcov_type="conley"` (Conley spatial-HAC) as a first-class feature: thread `conley_coords` / `conley_cutoff_km` / `conley_metric` / `conley_kernel` / `conley_time` / `conley_unit` / `conley_lag_cutoff` through `_fit_saturated_regression`. Phase 1b PR 1/8 deferred this; SA currently rejects `vcov_type="conley"` at `__init__` with a deferral message. | `diff_diff/sun_abraham.py` | follow-up | Medium |
+| Extend `StackedDiD` with `vcov_type="conley"` (Conley spatial-HAC) — thread the six `conley_*` params through `solve_ols` at `stacked_did.py:419` (and the `_refit_stacked` closure at `:444`). Phase 1b PR 2/8 deferred this; StackedDiD currently rejects `vcov_type="conley"` at `__init__` with a deferral message. Same shape as the SunAbraham conley follow-up. | `diff_diff/stacked_did.py` | follow-up | Medium |
 | Harmonize SunAbraham's HC1 within-transform finite-sample correction with `fixest::sunab()`. SA's `solve_ols` applies `n / (n - k_dm)` (within-transform columns only); fixest applies `n / (n - k_total)` (counts absorbed FE). SE values differ by ~1-2% on typical panel sizes (documented in REGISTRY.md "Deviation from R"; pinned at `atol=5e-3` in `tests/test_methodology_sun_abraham.py`). Either thread `df_adjustment` into the vcov scaling or document as an intentional difference. | `diff_diff/sun_abraham.py`, `diff_diff/linalg.py::compute_robust_vcov` | follow-up | Low |
 <!-- Rows 104-105 LIFTED 2026-05-20 via the clubSandwich WLS-CR2 port. The diff-diff
      form matches clubSandwich's specific algebra (W not sqrt(W), W^2 in bias term,
diff --git a/benchmarks/R/generate_stacked_did_golden.R b/benchmarks/R/generate_stacked_did_golden.R
new file mode 100644
index 00000000..9f80ded4
--- /dev/null
+++ b/benchmarks/R/generate_stacked_did_golden.R
@@ -0,0 +1,231 @@
+# Generate StackedDiD CR1 + CR2-BM golden values via R clubSandwich.
+#
+# This script is the parity source for Phase 1b 2/8 (StackedDiD vcov_type
+# threading). It loads the pre-stacked panel produced by the Python helper
+# at `benchmarks/data/stacked_did_test_panel.csv`, fits a single WLS regression
+# with weights=Q (matching the StackedDiD post-PR pattern in `stacked_did.py`
+# which calls `solve_ols(weights=composed_weights, vcov_type=...)`), and
+# computes both CR1 and CR2 Bell-McCaffrey goldens against which Python's
+# output is pinned at atol=1e-10.
+#
+# The pre-stacked CSV is committed (not regenerated by this script) so the
+# R-side regression is independent of any Python-side stacking-logic changes.
+# To regenerate the CSV, run:
+#     python3 benchmarks/python_helpers/dump_stacked_did_panel.py
+# (which uses StackedDiD.fit to extract the internal stacked dataset).
+#
+# Usage:
+#   Rscript benchmarks/R/generate_stacked_did_golden.R
+#
+# Requirements:
+#   clubSandwich (CRAN) >= 0.7.0
+#   jsonlite
+#
+# Output:
+#   benchmarks/data/stacked_did_golden.json
+
+suppressPackageStartupMessages({
+  library(clubSandwich)
+  library(jsonlite)
+})
+
+stopifnot(packageVersion("clubSandwich") >= "0.7.0")
+
+# --- Variant configurations -------------------------------------------------
+# Each variant has its own pre-stacked CSV + reference period + non_ref vector.
+# Default aggregate (kappa_pre=2, kappa_post=2, anticipation=0, weighting='aggregate')
+# is the primary case; non-default variants pin the methodology-relevant
+# parameter interactions (per local codex R8 P1).
+variants <- list(
+  default = list(
+    csv = "benchmarks/data/stacked_did_test_panel.csv",
+    ref_period = -1L,
+    out_key = NULL  # written to top-level keys: unit, unit_subexp
+  ),
+  population = list(
+    csv = "benchmarks/data/stacked_did_population_panel.csv",
+    ref_period = -1L,
+    out_key = "population_unit"  # heterogeneous-population Q-weights
+  ),
+  anticipation1 = list(
+    csv = "benchmarks/data/stacked_did_anticipation1_panel.csv",
+    ref_period = -2L,  # shifted: e = -1 - anticipation
+    out_key = "anticipation1_unit"
+  ),
+  sample_share = list(
+    csv = "benchmarks/data/stacked_did_sample_share_panel.csv",
+    ref_period = -1L,
+    out_key = "sample_share_unit"  # third Q-weight formula
+  )
+)
+
+
+process_variant <- function(csv_path, ref_period) {
+  stopifnot(file.exists(csv_path))
+  df <- read.csv(csv_path)
+
+# Sanity check expected columns
+required_cols <- c("unit", "period", "outcome", "_sub_exp", "_event_time",
+                   "_D_sa", "_Q_weight", "unit_subexp")
+# read.csv mangles underscore-prefix columns to X_sub_exp etc; normalize:
+colnames(df) <- gsub("^X_", "_", colnames(df))
+for (col in required_cols) {
+  stopifnot(col %in% colnames(df))
+}
+
+  # --- Build the design matrix ----------------------------------------------
+  # StackedDiD's regression form (Equation 3 in Wing-Freedman-Hollingsworth 2024):
+  #   y ~ intercept + D_sa + sum_h lambda_h * I(e=h) + sum_h delta_h * D_sa * I(e=h)
+  # with the reference period dropped.
+  event_times <- sort(unique(df$`_event_time`))
+  non_ref <- event_times[event_times != ref_period]
+# Rename underscore-prefix columns to R-friendly names for the formula.
+# (lm formulas can't reference backtick-quoted names with leading underscore
+# reliably across R versions.)
+df$D_sa <- df$`_D_sa`
+df$Q_weight <- df$`_Q_weight`
+df$event_time <- df$`_event_time`
+
+# Build dummies + interactions with R-friendly names. Use _neg suffix for
+# negative event times so column names are valid identifiers.
+ev_name <- function(h) {
+  if (h < 0) paste0("_neg", abs(h)) else paste0("_", h)
+}
+  for (h in non_ref) {
+    lname <- paste0("lambda", ev_name(h))
+    dname <- paste0("delta", ev_name(h))
+    df[[lname]] <- as.integer(df$event_time == h)
+    df[[dname]] <- df[[lname]] * df$D_sa
+  }
+
+  # Construct formula explicitly
+  lambda_cols <- paste0("lambda", sapply(non_ref, ev_name))
+  delta_cols <- paste0("delta", sapply(non_ref, ev_name))
+  lambda_terms <- paste(lambda_cols, collapse = " + ")
+  delta_terms <- paste(delta_cols, collapse = " + ")
+  form_str <- paste("outcome ~", "D_sa", "+", lambda_terms, "+", delta_terms)
+  form <- as.formula(form_str)
+
+  # Fit WLS with weights=Q
+  fit <- lm(form, data = df, weights = Q_weight)
+
+  coef_names <- names(coef(fit))
+  es_coef_names <- delta_cols
+
+  # Anticipation: under non-default anticipation>0, the "post-period" for
+  # the overall ATT contrast is `h >= -anticipation`, which depends on the
+  # reference period offset. anticipation = -1 - ref_period (since
+  # ref_period = -1 - anticipation in Python).
+  anticipation <- -1L - ref_period
+  post_threshold <- -anticipation  # h >= -anticipation enters overall ATT
+
+  clusters_to_test <- list(unit = df$unit, unit_subexp = df$unit_subexp)
+  variant_result <- list(
+    meta = list(
+      event_times_non_ref = non_ref,
+      es_coef_names = es_coef_names,
+      ref_period = ref_period,
+      anticipation = anticipation
+    )
+  )
+
+  for (cl_name in names(clusters_to_test)) {
+  cl <- clusters_to_test[[cl_name]]
+
+  # Use CR1S (Stata-style: G/(G-1) * (n-1)/(n-p) finite-sample correction)
+  # to match diff-diff's HC1 + cluster output. clubSandwich's plain "CR1"
+  # uses only the G/(G-1) factor and would diverge by ~(n-1)/(n-p) (~1.4%
+  # on this 325-row, 10-col stacked design).
+  vcov_cr1 <- vcovCR(fit, cluster = cl, type = "CR1S")
+  se_cr1_all <- sqrt(diag(vcov_cr1))
+  se_cr1_es <- se_cr1_all[es_coef_names]
+
+  vcov_cr2 <- vcovCR(fit, cluster = cl, type = "CR2")
+  ct_cr2 <- coef_test(fit, vcov = vcov_cr2)
+  se_cr2_all <- sqrt(diag(vcov_cr2))
+  # coef_test() returns rows in the order of names(coef(fit)); index by name.
+  rownames(ct_cr2) <- coef_names
+  se_cr2_es <- se_cr2_all[es_coef_names]
+  dof_bm_es <- ct_cr2[es_coef_names, "df_Satt"]
+
+    # Overall ATT contrast: post-period delta average. Under anticipation>0
+    # the post-period starts at h=-anticipation, NOT h=0 (mirrors Python's
+    # `_post_event_times_preview` filter at stacked_did.py:546-548).
+    post_delta_names <- delta_cols[non_ref >= post_threshold]
+    K_post <- length(post_delta_names)
+    c_overall <- matrix(0, nrow = 1, ncol = length(coef_names))
+    colnames(c_overall) <- coef_names
+    for (nm in post_delta_names) {
+      c_overall[1, nm] <- 1.0 / K_post
+    }
+    wt_overall <- Wald_test(fit, constraints = c_overall, vcov = vcov_cr2, test = "HTZ")
+    dof_bm_overall <- as.numeric(wt_overall$df_denom)
+    overall_var_cr2 <- as.numeric(c_overall %*% vcov_cr2 %*% t(c_overall))
+    se_overall_cr2 <- sqrt(overall_var_cr2)
+    overall_var_cr1 <- as.numeric(c_overall %*% vcov_cr1 %*% t(c_overall))
+    se_overall_cr1 <- sqrt(overall_var_cr1)
+
+    variant_result[[cl_name]] <- list(
+      se_cr1_es = as.numeric(se_cr1_es),
+      se_cr2_es = as.numeric(se_cr2_es),
+      dof_bm_es = as.numeric(dof_bm_es),
+      dof_bm_overall = dof_bm_overall,
+      se_overall_cr1 = se_overall_cr1,
+      se_overall_cr2 = se_overall_cr2,
+      es_coef_names = es_coef_names,
+      se_cr1_intercept = as.numeric(se_cr1_all["(Intercept)"]),
+      se_cr2_intercept = as.numeric(se_cr2_all["(Intercept)"]),
+      coef_es = as.numeric(coef(fit)[es_coef_names])
+    )
+  }
+  return(variant_result)
+}
+
+
+# --- Drive all variants -----------------------------------------------------
+output <- list(
+  meta = list(
+    clubSandwich_version = as.character(packageVersion("clubSandwich")),
+    R_version = R.version.string,
+    panel_descriptor = list(
+      n_units = 50L, n_periods = 8L,
+      cohort_periods = c(3L, 5L, 7L),
+      never_treated_frac = 0.3, treatment_effect = 2.0,
+      dynamic_effects = FALSE, seed = 20260521L
+    ),
+    estimator_descriptor = list(kappa_pre = 2L, kappa_post = 2L)
+  )
+)
+
+for (variant_name in names(variants)) {
+  v <- variants[[variant_name]]
+  cat("Processing variant:", variant_name, "(csv=", v$csv, ", ref=", v$ref_period, ")\n")
+  result <- process_variant(v$csv, v$ref_period)
+  if (variant_name == "default") {
+    # Default variant: write per-cluster keys at top level (back-compat)
+    output$unit <- result$unit
+    output$unit_subexp <- result$unit_subexp
+    output$meta$event_times_non_ref <- result$meta$event_times_non_ref
+    output$meta$es_coef_names <- result$meta$es_coef_names
+  } else {
+    # Non-default variants: write under variant-specific key
+    output[[variant_name]] <- result
+  }
+}
+
+# --- Write JSON -------------------------------------------------------------
+out_path <- "benchmarks/data/stacked_did_golden.json"
+write_json(output, out_path, digits = 15, auto_unbox = TRUE, pretty = TRUE)
+cat("Wrote", out_path, "\n")
+cat("Default CR1 unit-cluster SE for delta_0:",
+    output$unit$se_cr1_es[which(output$meta$event_times_non_ref == 0)], "\n")
+cat("Default CR2-BM unit-cluster SE for delta_0:",
+    output$unit$se_cr2_es[which(output$meta$event_times_non_ref == 0)], "\n")
+if (!is.null(output$population)) {
+  cat("Population CR2-BM unit SE for delta_0:",
+      output$population$unit$se_cr2_es[which(output$population$unit$es_coef_names == "delta_0")], "\n")
+}
+if (!is.null(output$anticipation1)) {
+  cat("Anticipation1 CR2-BM unit overall_se:",
+      output$anticipation1$unit$se_overall_cr2, "\n")
+}
diff --git a/benchmarks/data/stacked_did_anticipation1_panel.csv b/benchmarks/data/stacked_did_anticipation1_panel.csv
new file mode 100644
index 00000000..b0dc1f37
--- /dev/null
+++ b/benchmarks/data/stacked_did_anticipation1_panel.csv
@@ -0,0 +1,391 @@
+unit,period,outcome,first_treat,treated,treat,true_effect,_sub_exp,_event_time,_D_sa,_Q_weight,unit_subexp
+0,0,7.406759096457782,inf,0,0,0.0,3.0,-3,0,0.768,0_3.0
+0,1,7.483662635760572,inf,0,0,0.0,3.0,-2,0,0.768,0_3.0
+0,2,6.841502772126143,inf,0,0,0.0,3.0,-1,0,0.768,0_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,3.0,0,0,0.768,0_3.0
+0,4,7.947042738617837,inf,0,0,0.0,3.0,1,0,0.768,0_3.0
+0,5,6.317944717889617,inf,0,0,0.0,3.0,2,0,0.768,0_3.0
+1,0,9.479768948681208,inf,0,0,0.0,3.0,-3,0,0.768,1_3.0
+1,1,9.509053965877307,inf,0,0,0.0,3.0,-2,0,0.768,1_3.0
+1,2,10.614366425343084,inf,0,0,0.0,3.0,-1,0,0.768,1_3.0
+1,3,9.423988866091562,inf,0,0,0.0,3.0,0,0,0.768,1_3.0
+1,4,11.746966365956794,inf,0,0,0.0,3.0,1,0,0.768,1_3.0
+1,5,10.014392475247982,inf,0,0,0.0,3.0,2,0,0.768,1_3.0
+2,0,8.648909561731092,inf,0,0,0.0,3.0,-3,0,0.768,2_3.0
+2,1,8.504571437173666,inf,0,0,0.0,3.0,-2,0,0.768,2_3.0
+2,2,7.968205325235655,inf,0,0,0.0,3.0,-1,0,0.768,2_3.0
+2,3,8.129454052191617,inf,0,0,0.0,3.0,0,0,0.768,2_3.0
+2,4,8.633961259377445,inf,0,0,0.0,3.0,1,0,0.768,2_3.0
+2,5,9.895662186433224,inf,0,0,0.0,3.0,2,0,0.768,2_3.0
+3,0,14.422259200530778,inf,0,0,0.0,3.0,-3,0,0.768,3_3.0
+3,1,13.825385295202855,inf,0,0,0.0,3.0,-2,0,0.768,3_3.0
+3,2,13.829286373980507,inf,0,0,0.0,3.0,-1,0,0.768,3_3.0
+3,3,13.948413480279912,inf,0,0,0.0,3.0,0,0,0.768,3_3.0
+3,4,13.516792177472347,inf,0,0,0.0,3.0,1,0,0.768,3_3.0
+3,5,14.119361417606083,inf,0,0,0.0,3.0,2,0,0.768,3_3.0
+4,0,11.372109139013972,inf,0,0,0.0,3.0,-3,0,0.768,4_3.0
+4,1,11.72507772055074,inf,0,0,0.0,3.0,-2,0,0.768,4_3.0
+4,2,11.912646534344725,inf,0,0,0.0,3.0,-1,0,0.768,4_3.0
+4,3,11.813278538150167,inf,0,0,0.0,3.0,0,0,0.768,4_3.0
+4,4,13.525689949379311,inf,0,0,0.0,3.0,1,0,0.768,4_3.0
+4,5,12.732831972997747,inf,0,0,0.0,3.0,2,0,0.768,4_3.0
+5,0,6.963034690669673,inf,0,0,0.0,3.0,-3,0,0.768,5_3.0
+5,1,6.378890705428073,inf,0,0,0.0,3.0,-2,0,0.768,5_3.0
+5,2,6.440945780933881,inf,0,0,0.0,3.0,-1,0,0.768,5_3.0
+5,3,6.6127868751986805,inf,0,0,0.0,3.0,0,0,0.768,5_3.0
+5,4,6.513422311622866,inf,0,0,0.0,3.0,1,0,0.768,5_3.0
+5,5,6.75979943656341,inf,0,0,0.0,3.0,2,0,0.768,5_3.0
+6,0,8.711680027275019,inf,0,0,0.0,3.0,-3,0,0.768,6_3.0
+6,1,8.633688139950424,inf,0,0,0.0,3.0,-2,0,0.768,6_3.0
+6,2,8.600953374134306,inf,0,0,0.0,3.0,-1,0,0.768,6_3.0
+6,3,8.411617732005928,inf,0,0,0.0,3.0,0,0,0.768,6_3.0
+6,4,8.387370342059572,inf,0,0,0.0,3.0,1,0,0.768,6_3.0
+6,5,8.945688592547397,inf,0,0,0.0,3.0,2,0,0.768,6_3.0
+7,0,10.0495061084566,inf,0,0,0.0,3.0,-3,0,0.768,7_3.0
+7,1,11.334453872912409,inf,0,0,0.0,3.0,-2,0,0.768,7_3.0
+7,2,10.898891009750589,inf,0,0,0.0,3.0,-1,0,0.768,7_3.0
+7,3,10.456288009791654,inf,0,0,0.0,3.0,0,0,0.768,7_3.0
+7,4,11.086419785594533,inf,0,0,0.0,3.0,1,0,0.768,7_3.0
+7,5,11.136486986823087,inf,0,0,0.0,3.0,2,0,0.768,7_3.0
+8,0,13.270274333031267,inf,0,0,0.0,3.0,-3,0,0.768,8_3.0
+8,1,12.363906817019398,inf,0,0,0.0,3.0,-2,0,0.768,8_3.0
+8,2,13.349921038000932,inf,0,0,0.0,3.0,-1,0,0.768,8_3.0
+8,3,11.289014907437469,inf,0,0,0.0,3.0,0,0,0.768,8_3.0
+8,4,12.413821661809754,inf,0,0,0.0,3.0,1,0,0.768,8_3.0
+8,5,12.599831686253742,inf,0,0,0.0,3.0,2,0,0.768,8_3.0
+9,0,7.276206709953215,inf,0,0,0.0,3.0,-3,0,0.768,9_3.0
+9,1,8.226753444350999,inf,0,0,0.0,3.0,-2,0,0.768,9_3.0
+9,2,8.04357609368333,inf,0,0,0.0,3.0,-1,0,0.768,9_3.0
+9,3,7.935929232288304,inf,0,0,0.0,3.0,0,0,0.768,9_3.0
+9,4,8.947274202622067,inf,0,0,0.0,3.0,1,0,0.768,9_3.0
+9,5,8.525815827970705,inf,0,0,0.0,3.0,2,0,0.768,9_3.0
+10,0,7.458142081323446,inf,0,0,0.0,3.0,-3,0,0.768,10_3.0
+10,1,6.616529053324495,inf,0,0,0.0,3.0,-2,0,0.768,10_3.0
+10,2,7.904056043755299,inf,0,0,0.0,3.0,-1,0,0.768,10_3.0
+10,3,7.385638433127546,inf,0,0,0.0,3.0,0,0,0.768,10_3.0
+10,4,7.6853500749042345,inf,0,0,0.0,3.0,1,0,0.768,10_3.0
+10,5,7.841052147984791,inf,0,0,0.0,3.0,2,0,0.768,10_3.0
+11,0,8.533611818996034,inf,0,0,0.0,3.0,-3,0,0.768,11_3.0
+11,1,8.904134028447222,inf,0,0,0.0,3.0,-2,0,0.768,11_3.0
+11,2,9.722452361429282,inf,0,0,0.0,3.0,-1,0,0.768,11_3.0
+11,3,8.983212026944074,inf,0,0,0.0,3.0,0,0,0.768,11_3.0
+11,4,9.774571594385389,inf,0,0,0.0,3.0,1,0,0.768,11_3.0
+11,5,9.962850837041142,inf,0,0,0.0,3.0,2,0,0.768,11_3.0
+12,0,13.559439661765467,inf,0,0,0.0,3.0,-3,0,0.768,12_3.0
+12,1,14.247804715186929,inf,0,0,0.0,3.0,-2,0,0.768,12_3.0
+12,2,13.59781501202838,inf,0,0,0.0,3.0,-1,0,0.768,12_3.0
+12,3,14.268684067203575,inf,0,0,0.0,3.0,0,0,0.768,12_3.0
+12,4,13.089643664151836,inf,0,0,0.0,3.0,1,0,0.768,12_3.0
+12,5,13.739823106373846,inf,0,0,0.0,3.0,2,0,0.768,12_3.0
+13,0,10.925098782191794,inf,0,0,0.0,3.0,-3,0,0.768,13_3.0
+13,1,12.466946753384763,inf,0,0,0.0,3.0,-2,0,0.768,13_3.0
+13,2,12.577723255899597,inf,0,0,0.0,3.0,-1,0,0.768,13_3.0
+13,3,12.989441559722076,inf,0,0,0.0,3.0,0,0,0.768,13_3.0
+13,4,12.274600589012058,inf,0,0,0.0,3.0,1,0,0.768,13_3.0
+13,5,12.427220506745886,inf,0,0,0.0,3.0,2,0,0.768,13_3.0
+14,0,9.149296607907232,inf,0,0,0.0,3.0,-3,0,0.768,14_3.0
+14,1,9.321384661956385,inf,0,0,0.0,3.0,-2,0,0.768,14_3.0
+14,2,8.537850792031616,inf,0,0,0.0,3.0,-1,0,0.768,14_3.0
+14,3,9.118405772601042,inf,0,0,0.0,3.0,0,0,0.768,14_3.0
+14,4,9.74251766740271,inf,0,0,0.0,3.0,1,0,0.768,14_3.0
+14,5,9.254403864326273,inf,0,0,0.0,3.0,2,0,0.768,14_3.0
+15,0,8.008987192084824,7.0,0,1,0.0,3.0,-3,0,0.768,15_3.0
+15,1,7.441009679583563,7.0,0,1,0.0,3.0,-2,0,0.768,15_3.0
+15,2,8.485304881474772,7.0,0,1,0.0,3.0,-1,0,0.768,15_3.0
+15,3,8.564340753151164,7.0,0,1,0.0,3.0,0,0,0.768,15_3.0
+15,4,8.1815349724801,7.0,0,1,0.0,3.0,1,0,0.768,15_3.0
+15,5,9.311637156674976,7.0,0,1,0.0,3.0,2,0,0.768,15_3.0
+16,0,7.904371486673272,3.0,0,1,0.0,3.0,-3,1,1.0,16_3.0
+16,1,7.6836207307333035,3.0,0,1,0.0,3.0,-2,1,1.0,16_3.0
+16,2,8.357654569597338,3.0,0,1,0.0,3.0,-1,1,1.0,16_3.0
+16,3,10.056550820184588,3.0,1,1,2.0,3.0,0,1,1.0,16_3.0
+16,4,10.207378999928212,3.0,1,1,2.0,3.0,1,1,1.0,16_3.0
+16,5,10.096882823598278,3.0,1,1,2.0,3.0,2,1,1.0,16_3.0
+17,0,9.305982495584836,3.0,0,1,0.0,3.0,-3,1,1.0,17_3.0
+17,1,8.175606640239206,3.0,0,1,0.0,3.0,-2,1,1.0,17_3.0
+17,2,8.600350852687033,3.0,0,1,0.0,3.0,-1,1,1.0,17_3.0
+17,3,9.60085063995874,3.0,1,1,2.0,3.0,0,1,1.0,17_3.0
+17,4,11.265857113284081,3.0,1,1,2.0,3.0,1,1,1.0,17_3.0
+17,5,10.805731826455423,3.0,1,1,2.0,3.0,2,1,1.0,17_3.0
+19,0,8.68828511075003,3.0,0,1,0.0,3.0,-3,1,1.0,19_3.0
+19,1,8.287479893276272,3.0,0,1,0.0,3.0,-2,1,1.0,19_3.0
+19,2,8.838449778710965,3.0,0,1,0.0,3.0,-1,1,1.0,19_3.0
+19,3,10.301088160788524,3.0,1,1,2.0,3.0,0,1,1.0,19_3.0
+19,4,10.193467173412117,3.0,1,1,2.0,3.0,1,1,1.0,19_3.0
+19,5,10.93570451644154,3.0,1,1,2.0,3.0,2,1,1.0,19_3.0
+20,0,7.605010847122977,3.0,0,1,0.0,3.0,-3,1,1.0,20_3.0
+20,1,6.608825849966525,3.0,0,1,0.0,3.0,-2,1,1.0,20_3.0
+20,2,6.192567902449463,3.0,0,1,0.0,3.0,-1,1,1.0,20_3.0
+20,3,8.62011884779571,3.0,1,1,2.0,3.0,0,1,1.0,20_3.0
+20,4,9.534699361996061,3.0,1,1,2.0,3.0,1,1,1.0,20_3.0
+20,5,8.972714344039472,3.0,1,1,2.0,3.0,2,1,1.0,20_3.0
+23,0,11.382987102633042,7.0,0,1,0.0,3.0,-3,0,0.768,23_3.0
+23,1,11.395156619865356,7.0,0,1,0.0,3.0,-2,0,0.768,23_3.0
+23,2,12.292733093551572,7.0,0,1,0.0,3.0,-1,0,0.768,23_3.0
+23,3,10.987337838152467,7.0,0,1,0.0,3.0,0,0,0.768,23_3.0
+23,4,12.858675160893522,7.0,0,1,0.0,3.0,1,0,0.768,23_3.0
+23,5,11.476064698198886,7.0,0,1,0.0,3.0,2,0,0.768,23_3.0
+26,0,11.077555559492247,3.0,0,1,0.0,3.0,-3,1,1.0,26_3.0
+26,1,10.668383782601722,3.0,0,1,0.0,3.0,-2,1,1.0,26_3.0
+26,2,10.446234601225571,3.0,0,1,0.0,3.0,-1,1,1.0,26_3.0
+26,3,13.556242815369252,3.0,1,1,2.0,3.0,0,1,1.0,26_3.0
+26,4,13.576729933292084,3.0,1,1,2.0,3.0,1,1,1.0,26_3.0
+26,5,13.922175150825803,3.0,1,1,2.0,3.0,2,1,1.0,26_3.0
+27,0,8.063532168847798,7.0,0,1,0.0,3.0,-3,0,0.768,27_3.0
+27,1,8.580588297539316,7.0,0,1,0.0,3.0,-2,0,0.768,27_3.0
+27,2,9.053926240501355,7.0,0,1,0.0,3.0,-1,0,0.768,27_3.0
+27,3,8.688974179281848,7.0,0,1,0.0,3.0,0,0,0.768,27_3.0
+27,4,7.928575988525038,7.0,0,1,0.0,3.0,1,0,0.768,27_3.0
+27,5,9.170520905605873,7.0,0,1,0.0,3.0,2,0,0.768,27_3.0
+28,0,9.961088788816808,3.0,0,1,0.0,3.0,-3,1,1.0,28_3.0
+28,1,9.371848258038815,3.0,0,1,0.0,3.0,-2,1,1.0,28_3.0
+28,2,9.154422149577181,3.0,0,1,0.0,3.0,-1,1,1.0,28_3.0
+28,3,11.965196547208972,3.0,1,1,2.0,3.0,0,1,1.0,28_3.0
+28,4,11.302065396349953,3.0,1,1,2.0,3.0,1,1,1.0,28_3.0
+28,5,11.589704550706,3.0,1,1,2.0,3.0,2,1,1.0,28_3.0
+29,0,9.099752609079598,3.0,0,1,0.0,3.0,-3,1,1.0,29_3.0
+29,1,8.985010266355241,3.0,0,1,0.0,3.0,-2,1,1.0,29_3.0
+29,2,8.92226436961385,3.0,0,1,0.0,3.0,-1,1,1.0,29_3.0
+29,3,11.587183906367892,3.0,1,1,2.0,3.0,0,1,1.0,29_3.0
+29,4,11.061585148973101,3.0,1,1,2.0,3.0,1,1,1.0,29_3.0
+29,5,11.32705363934025,3.0,1,1,2.0,3.0,2,1,1.0,29_3.0
+30,0,7.717097435194422,3.0,0,1,0.0,3.0,-3,1,1.0,30_3.0
+30,1,8.376287388505098,3.0,0,1,0.0,3.0,-2,1,1.0,30_3.0
+30,2,8.657693030574496,3.0,0,1,0.0,3.0,-1,1,1.0,30_3.0
+30,3,10.18781936742593,3.0,1,1,2.0,3.0,0,1,1.0,30_3.0
+30,4,11.000739934625502,3.0,1,1,2.0,3.0,1,1,1.0,30_3.0
+30,5,10.282065104573295,3.0,1,1,2.0,3.0,2,1,1.0,30_3.0
+33,0,11.87643097539092,7.0,0,1,0.0,3.0,-3,0,0.768,33_3.0
+33,1,11.57454492762582,7.0,0,1,0.0,3.0,-2,0,0.768,33_3.0
+33,2,11.691765125046665,7.0,0,1,0.0,3.0,-1,0,0.768,33_3.0
+33,3,11.84979372696007,7.0,0,1,0.0,3.0,0,0,0.768,33_3.0
+33,4,12.62424994943963,7.0,0,1,0.0,3.0,1,0,0.768,33_3.0
+33,5,12.3254386483586,7.0,0,1,0.0,3.0,2,0,0.768,33_3.0
+34,0,11.753650691987085,7.0,0,1,0.0,3.0,-3,0,0.768,34_3.0
+34,1,12.563769947545987,7.0,0,1,0.0,3.0,-2,0,0.768,34_3.0
+34,2,11.966597306919729,7.0,0,1,0.0,3.0,-1,0,0.768,34_3.0
+34,3,12.803090421682993,7.0,0,1,0.0,3.0,0,0,0.768,34_3.0
+34,4,13.544985407250604,7.0,0,1,0.0,3.0,1,0,0.768,34_3.0
+34,5,13.17341536766152,7.0,0,1,0.0,3.0,2,0,0.768,34_3.0
+35,0,12.933027771034341,3.0,0,1,0.0,3.0,-3,1,1.0,35_3.0
+35,1,13.198034881842501,3.0,0,1,0.0,3.0,-2,1,1.0,35_3.0
+35,2,13.13803314498448,3.0,0,1,0.0,3.0,-1,1,1.0,35_3.0
+35,3,14.775491475016876,3.0,1,1,2.0,3.0,0,1,1.0,35_3.0
+35,4,14.755631298179171,3.0,1,1,2.0,3.0,1,1,1.0,35_3.0
+35,5,15.82726562084955,3.0,1,1,2.0,3.0,2,1,1.0,35_3.0
+36,0,5.516117799035085,3.0,0,1,0.0,3.0,-3,1,1.0,36_3.0
+36,1,7.2152253987939226,3.0,0,1,0.0,3.0,-2,1,1.0,36_3.0
+36,2,6.610853150968622,3.0,0,1,0.0,3.0,-1,1,1.0,36_3.0
+36,3,8.853563997233643,3.0,1,1,2.0,3.0,0,1,1.0,36_3.0
+36,4,9.09329932509169,3.0,1,1,2.0,3.0,1,1,1.0,36_3.0
+36,5,9.298377296285087,3.0,1,1,2.0,3.0,2,1,1.0,36_3.0
+37,0,8.78384655702291,7.0,0,1,0.0,3.0,-3,0,0.768,37_3.0
+37,1,10.146958297869952,7.0,0,1,0.0,3.0,-2,0,0.768,37_3.0
+37,2,9.060497069998855,7.0,0,1,0.0,3.0,-1,0,0.768,37_3.0
+37,3,9.318388049432945,7.0,0,1,0.0,3.0,0,0,0.768,37_3.0
+37,4,9.26097646005531,7.0,0,1,0.0,3.0,1,0,0.768,37_3.0
+37,5,8.919210921226544,7.0,0,1,0.0,3.0,2,0,0.768,37_3.0
+39,0,12.050833460924963,3.0,0,1,0.0,3.0,-3,1,1.0,39_3.0
+39,1,12.353298527293795,3.0,0,1,0.0,3.0,-2,1,1.0,39_3.0
+39,2,12.242756403937744,3.0,0,1,0.0,3.0,-1,1,1.0,39_3.0
+39,3,15.149617439612653,3.0,1,1,2.0,3.0,0,1,1.0,39_3.0
+39,4,14.088452790656742,3.0,1,1,2.0,3.0,1,1,1.0,39_3.0
+39,5,14.060806514376624,3.0,1,1,2.0,3.0,2,1,1.0,39_3.0
+40,0,8.392053799385957,7.0,0,1,0.0,3.0,-3,0,0.768,40_3.0
+40,1,7.342148190809627,7.0,0,1,0.0,3.0,-2,0,0.768,40_3.0
+40,2,8.356636279545437,7.0,0,1,0.0,3.0,-1,0,0.768,40_3.0
+40,3,8.853341199215448,7.0,0,1,0.0,3.0,0,0,0.768,40_3.0
+40,4,7.646273376822638,7.0,0,1,0.0,3.0,1,0,0.768,40_3.0
+40,5,7.959815635944495,7.0,0,1,0.0,3.0,2,0,0.768,40_3.0
+41,0,12.129423882345126,3.0,0,1,0.0,3.0,-3,1,1.0,41_3.0
+41,1,12.529859573199765,3.0,0,1,0.0,3.0,-2,1,1.0,41_3.0
+41,2,12.359862327664574,3.0,0,1,0.0,3.0,-1,1,1.0,41_3.0
+41,3,14.594595526225248,3.0,1,1,2.0,3.0,0,1,1.0,41_3.0
+41,4,14.627150386514797,3.0,1,1,2.0,3.0,1,1,1.0,41_3.0
+41,5,15.438306775111634,3.0,1,1,2.0,3.0,2,1,1.0,41_3.0
+43,0,11.416936766137415,7.0,0,1,0.0,3.0,-3,0,0.768,43_3.0
+43,1,11.589875560059742,7.0,0,1,0.0,3.0,-2,0,0.768,43_3.0
+43,2,11.566062741673996,7.0,0,1,0.0,3.0,-1,0,0.768,43_3.0
+43,3,11.081605645195344,7.0,0,1,0.0,3.0,0,0,0.768,43_3.0
+43,4,11.740097124129901,7.0,0,1,0.0,3.0,1,0,0.768,43_3.0
+43,5,12.466392539854581,7.0,0,1,0.0,3.0,2,0,0.768,43_3.0
+45,0,7.623038515654278,7.0,0,1,0.0,3.0,-3,0,0.768,45_3.0
+45,1,8.77119845157113,7.0,0,1,0.0,3.0,-2,0,0.768,45_3.0
+45,2,8.039535155953265,7.0,0,1,0.0,3.0,-1,0,0.768,45_3.0
+45,3,8.077961859940832,7.0,0,1,0.0,3.0,0,0,0.768,45_3.0
+45,4,8.881163504521098,7.0,0,1,0.0,3.0,1,0,0.768,45_3.0
+45,5,8.557759225694356,7.0,0,1,0.0,3.0,2,0,0.768,45_3.0
+46,0,12.35051638560887,7.0,0,1,0.0,3.0,-3,0,0.768,46_3.0
+46,1,12.148563177783899,7.0,0,1,0.0,3.0,-2,0,0.768,46_3.0
+46,2,12.307754494386236,7.0,0,1,0.0,3.0,-1,0,0.768,46_3.0
+46,3,12.393620432500153,7.0,0,1,0.0,3.0,0,0,0.768,46_3.0
+46,4,13.035568610439492,7.0,0,1,0.0,3.0,1,0,0.768,46_3.0
+46,5,12.995360934519939,7.0,0,1,0.0,3.0,2,0,0.768,46_3.0
+0,2,6.841502772126143,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,0_5.0
+0,3,7.9831226286999195,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,0_5.0
+0,4,7.947042738617837,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,0_5.0
+0,5,6.317944717889617,inf,0,0,0.0,5.0,0,0,1.3866666666666667,0_5.0
+0,6,8.030151283551504,inf,0,0,0.0,5.0,1,0,1.3866666666666667,0_5.0
+0,7,7.857022192736064,inf,0,0,0.0,5.0,2,0,1.3866666666666667,0_5.0
+1,2,10.614366425343084,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,1_5.0
+1,3,9.423988866091562,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,1_5.0
+1,4,11.746966365956794,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,1_5.0
+1,5,10.014392475247982,inf,0,0,0.0,5.0,0,0,1.3866666666666667,1_5.0
+1,6,10.364571813434443,inf,0,0,0.0,5.0,1,0,1.3866666666666667,1_5.0
+1,7,10.370697302329921,inf,0,0,0.0,5.0,2,0,1.3866666666666667,1_5.0
+2,2,7.968205325235655,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,2_5.0
+2,3,8.129454052191617,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,2_5.0
+2,4,8.633961259377445,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,2_5.0
+2,5,9.895662186433224,inf,0,0,0.0,5.0,0,0,1.3866666666666667,2_5.0
+2,6,8.55134266155172,inf,0,0,0.0,5.0,1,0,1.3866666666666667,2_5.0
+2,7,8.517976639648715,inf,0,0,0.0,5.0,2,0,1.3866666666666667,2_5.0
+3,2,13.829286373980507,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,3_5.0
+3,3,13.948413480279912,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,3_5.0
+3,4,13.516792177472347,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,3_5.0
+3,5,14.119361417606083,inf,0,0,0.0,5.0,0,0,1.3866666666666667,3_5.0
+3,6,13.338205224874367,inf,0,0,0.0,5.0,1,0,1.3866666666666667,3_5.0
+3,7,14.4234803289339,inf,0,0,0.0,5.0,2,0,1.3866666666666667,3_5.0
+4,2,11.912646534344725,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,4_5.0
+4,3,11.813278538150167,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,4_5.0
+4,4,13.525689949379311,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,4_5.0
+4,5,12.732831972997747,inf,0,0,0.0,5.0,0,0,1.3866666666666667,4_5.0
+4,6,12.618074611389941,inf,0,0,0.0,5.0,1,0,1.3866666666666667,4_5.0
+4,7,11.481627079417956,inf,0,0,0.0,5.0,2,0,1.3866666666666667,4_5.0
+5,2,6.440945780933881,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,5_5.0
+5,3,6.6127868751986805,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,5_5.0
+5,4,6.513422311622866,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,5_5.0
+5,5,6.75979943656341,inf,0,0,0.0,5.0,0,0,1.3866666666666667,5_5.0
+5,6,6.600117721454566,inf,0,0,0.0,5.0,1,0,1.3866666666666667,5_5.0
+5,7,7.410185509058965,inf,0,0,0.0,5.0,2,0,1.3866666666666667,5_5.0
+6,2,8.600953374134306,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,6_5.0
+6,3,8.411617732005928,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,6_5.0
+6,4,8.387370342059572,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,6_5.0
+6,5,8.945688592547397,inf,0,0,0.0,5.0,0,0,1.3866666666666667,6_5.0
+6,6,9.424533467045443,inf,0,0,0.0,5.0,1,0,1.3866666666666667,6_5.0
+6,7,8.558459954832495,inf,0,0,0.0,5.0,2,0,1.3866666666666667,6_5.0
+7,2,10.898891009750589,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,7_5.0
+7,3,10.456288009791654,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,7_5.0
+7,4,11.086419785594533,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,7_5.0
+7,5,11.136486986823087,inf,0,0,0.0,5.0,0,0,1.3866666666666667,7_5.0
+7,6,10.49707772184382,inf,0,0,0.0,5.0,1,0,1.3866666666666667,7_5.0
+7,7,11.220866186095618,inf,0,0,0.0,5.0,2,0,1.3866666666666667,7_5.0
+8,2,13.349921038000932,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,8_5.0
+8,3,11.289014907437469,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,8_5.0
+8,4,12.413821661809754,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,8_5.0
+8,5,12.599831686253742,inf,0,0,0.0,5.0,0,0,1.3866666666666667,8_5.0
+8,6,12.770672804509399,inf,0,0,0.0,5.0,1,0,1.3866666666666667,8_5.0
+8,7,12.12649797254452,inf,0,0,0.0,5.0,2,0,1.3866666666666667,8_5.0
+9,2,8.04357609368333,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,9_5.0
+9,3,7.935929232288304,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,9_5.0
+9,4,8.947274202622067,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,9_5.0
+9,5,8.525815827970705,inf,0,0,0.0,5.0,0,0,1.3866666666666667,9_5.0
+9,6,8.29259367272137,inf,0,0,0.0,5.0,1,0,1.3866666666666667,9_5.0
+9,7,7.755986984180787,inf,0,0,0.0,5.0,2,0,1.3866666666666667,9_5.0
+10,2,7.904056043755299,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,10_5.0
+10,3,7.385638433127546,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,10_5.0
+10,4,7.6853500749042345,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,10_5.0
+10,5,7.841052147984791,inf,0,0,0.0,5.0,0,0,1.3866666666666667,10_5.0
+10,6,7.779603182233025,inf,0,0,0.0,5.0,1,0,1.3866666666666667,10_5.0
+10,7,9.262810775890333,inf,0,0,0.0,5.0,2,0,1.3866666666666667,10_5.0
+11,2,9.722452361429282,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,11_5.0
+11,3,8.983212026944074,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,11_5.0
+11,4,9.774571594385389,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,11_5.0
+11,5,9.962850837041142,inf,0,0,0.0,5.0,0,0,1.3866666666666667,11_5.0
+11,6,9.456066431566448,inf,0,0,0.0,5.0,1,0,1.3866666666666667,11_5.0
+11,7,9.374591364502958,inf,0,0,0.0,5.0,2,0,1.3866666666666667,11_5.0
+12,2,13.59781501202838,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,12_5.0
+12,3,14.268684067203575,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,12_5.0
+12,4,13.089643664151836,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,12_5.0
+12,5,13.739823106373846,inf,0,0,0.0,5.0,0,0,1.3866666666666667,12_5.0
+12,6,13.669519287347907,inf,0,0,0.0,5.0,1,0,1.3866666666666667,12_5.0
+12,7,13.99578598067457,inf,0,0,0.0,5.0,2,0,1.3866666666666667,12_5.0
+13,2,12.577723255899597,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,13_5.0
+13,3,12.989441559722076,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,13_5.0
+13,4,12.274600589012058,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,13_5.0
+13,5,12.427220506745886,inf,0,0,0.0,5.0,0,0,1.3866666666666667,13_5.0
+13,6,12.754093760275953,inf,0,0,0.0,5.0,1,0,1.3866666666666667,13_5.0
+13,7,13.718433817762524,inf,0,0,0.0,5.0,2,0,1.3866666666666667,13_5.0
+14,2,8.537850792031616,inf,0,0,0.0,5.0,-3,0,1.3866666666666667,14_5.0
+14,3,9.118405772601042,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,14_5.0
+14,4,9.74251766740271,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,14_5.0
+14,5,9.254403864326273,inf,0,0,0.0,5.0,0,0,1.3866666666666667,14_5.0
+14,6,9.330459677777796,inf,0,0,0.0,5.0,1,0,1.3866666666666667,14_5.0
+14,7,9.00828730479836,inf,0,0,0.0,5.0,2,0,1.3866666666666667,14_5.0
+18,2,9.888993399380476,5.0,0,1,0.0,5.0,-3,1,1.0,18_5.0
+18,3,8.750153667767103,5.0,0,1,0.0,5.0,-2,1,1.0,18_5.0
+18,4,10.369418029568877,5.0,0,1,0.0,5.0,-1,1,1.0,18_5.0
+18,5,11.814853522513967,5.0,1,1,2.0,5.0,0,1,1.0,18_5.0
+18,6,11.880991600710768,5.0,1,1,2.0,5.0,1,1,1.0,18_5.0
+18,7,12.084939640054936,5.0,1,1,2.0,5.0,2,1,1.0,18_5.0
+21,2,11.681432246244258,5.0,0,1,0.0,5.0,-3,1,1.0,21_5.0
+21,3,11.130313135937552,5.0,0,1,0.0,5.0,-2,1,1.0,21_5.0
+21,4,11.783557955755562,5.0,0,1,0.0,5.0,-1,1,1.0,21_5.0
+21,5,13.932178697330485,5.0,1,1,2.0,5.0,0,1,1.0,21_5.0
+21,6,13.311045276660398,5.0,1,1,2.0,5.0,1,1,1.0,21_5.0
+21,7,12.672260765381138,5.0,1,1,2.0,5.0,2,1,1.0,21_5.0
+22,2,10.15237018274473,5.0,0,1,0.0,5.0,-3,1,1.0,22_5.0
+22,3,10.370162504733846,5.0,0,1,0.0,5.0,-2,1,1.0,22_5.0
+22,4,10.130795877474743,5.0,0,1,0.0,5.0,-1,1,1.0,22_5.0
+22,5,13.327665186472212,5.0,1,1,2.0,5.0,0,1,1.0,22_5.0
+22,6,14.006133024294815,5.0,1,1,2.0,5.0,1,1,1.0,22_5.0
+22,7,11.997025447630525,5.0,1,1,2.0,5.0,2,1,1.0,22_5.0
+24,2,11.65920009043054,5.0,0,1,0.0,5.0,-3,1,1.0,24_5.0
+24,3,12.216528415765985,5.0,0,1,0.0,5.0,-2,1,1.0,24_5.0
+24,4,11.844123402747586,5.0,0,1,0.0,5.0,-1,1,1.0,24_5.0
+24,5,13.783296021523771,5.0,1,1,2.0,5.0,0,1,1.0,24_5.0
+24,6,13.374452489632073,5.0,1,1,2.0,5.0,1,1,1.0,24_5.0
+24,7,14.257003341404754,5.0,1,1,2.0,5.0,2,1,1.0,24_5.0
+25,2,12.002838899464496,5.0,0,1,0.0,5.0,-3,1,1.0,25_5.0
+25,3,12.376139740663152,5.0,0,1,0.0,5.0,-2,1,1.0,25_5.0
+25,4,12.451932231507831,5.0,0,1,0.0,5.0,-1,1,1.0,25_5.0
+25,5,14.40345513916028,5.0,1,1,2.0,5.0,0,1,1.0,25_5.0
+25,6,15.399316274400507,5.0,1,1,2.0,5.0,1,1,1.0,25_5.0
+25,7,14.121018820768782,5.0,1,1,2.0,5.0,2,1,1.0,25_5.0
+31,2,5.5250809770396385,5.0,0,1,0.0,5.0,-3,1,1.0,31_5.0
+31,3,5.699685628062173,5.0,0,1,0.0,5.0,-2,1,1.0,31_5.0
+31,4,6.640244849690189,5.0,0,1,0.0,5.0,-1,1,1.0,31_5.0
+31,5,8.80287858433971,5.0,1,1,2.0,5.0,0,1,1.0,31_5.0
+31,6,7.478011213888342,5.0,1,1,2.0,5.0,1,1,1.0,31_5.0
+31,7,7.785449789009849,5.0,1,1,2.0,5.0,2,1,1.0,31_5.0
+32,2,13.040290095128224,5.0,0,1,0.0,5.0,-3,1,1.0,32_5.0
+32,3,13.145351442221973,5.0,0,1,0.0,5.0,-2,1,1.0,32_5.0
+32,4,14.213246370804844,5.0,0,1,0.0,5.0,-1,1,1.0,32_5.0
+32,5,15.673222444412447,5.0,1,1,2.0,5.0,0,1,1.0,32_5.0
+32,6,14.660677350520446,5.0,1,1,2.0,5.0,1,1,1.0,32_5.0
+32,7,14.475511437433273,5.0,1,1,2.0,5.0,2,1,1.0,32_5.0
+38,2,8.70755674968225,5.0,0,1,0.0,5.0,-3,1,1.0,38_5.0
+38,3,9.198240092902193,5.0,0,1,0.0,5.0,-2,1,1.0,38_5.0
+38,4,8.489100540648527,5.0,0,1,0.0,5.0,-1,1,1.0,38_5.0
+38,5,11.45417220010028,5.0,1,1,2.0,5.0,0,1,1.0,38_5.0
+38,6,10.634610329560985,5.0,1,1,2.0,5.0,1,1,1.0,38_5.0
+38,7,12.158628732093817,5.0,1,1,2.0,5.0,2,1,1.0,38_5.0
+42,2,9.504174709649446,5.0,0,1,0.0,5.0,-3,1,1.0,42_5.0
+42,3,9.165900951128542,5.0,0,1,0.0,5.0,-2,1,1.0,42_5.0
+42,4,9.836927314238459,5.0,0,1,0.0,5.0,-1,1,1.0,42_5.0
+42,5,11.893273621648287,5.0,1,1,2.0,5.0,0,1,1.0,42_5.0
+42,6,10.811622966738568,5.0,1,1,2.0,5.0,1,1,1.0,42_5.0
+42,7,12.387788948877349,5.0,1,1,2.0,5.0,2,1,1.0,42_5.0
+44,2,12.062723495722159,5.0,0,1,0.0,5.0,-3,1,1.0,44_5.0
+44,3,13.218744977031886,5.0,0,1,0.0,5.0,-2,1,1.0,44_5.0
+44,4,13.584858653200563,5.0,0,1,0.0,5.0,-1,1,1.0,44_5.0
+44,5,15.869766671523484,5.0,1,1,2.0,5.0,0,1,1.0,44_5.0
+44,6,15.606050081127213,5.0,1,1,2.0,5.0,1,1,1.0,44_5.0
+44,7,14.741953786805245,5.0,1,1,2.0,5.0,2,1,1.0,44_5.0
+47,2,10.398401916354672,5.0,0,1,0.0,5.0,-3,1,1.0,47_5.0
+47,3,11.405672205690639,5.0,0,1,0.0,5.0,-2,1,1.0,47_5.0
+47,4,10.327840224798175,5.0,0,1,0.0,5.0,-1,1,1.0,47_5.0
+47,5,12.972063268138358,5.0,1,1,2.0,5.0,0,1,1.0,47_5.0
+47,6,13.3375192672357,5.0,1,1,2.0,5.0,1,1,1.0,47_5.0
+47,7,13.949251964590784,5.0,1,1,2.0,5.0,2,1,1.0,47_5.0
+48,2,8.739142755653148,5.0,0,1,0.0,5.0,-3,1,1.0,48_5.0
+48,3,10.036462523941848,5.0,0,1,0.0,5.0,-2,1,1.0,48_5.0
+48,4,10.557346448715858,5.0,0,1,0.0,5.0,-1,1,1.0,48_5.0
+48,5,11.320126492133841,5.0,1,1,2.0,5.0,0,1,1.0,48_5.0
+48,6,12.376923426574146,5.0,1,1,2.0,5.0,1,1,1.0,48_5.0
+48,7,12.52530519550149,5.0,1,1,2.0,5.0,2,1,1.0,48_5.0
+49,2,10.233108990003252,5.0,0,1,0.0,5.0,-3,1,1.0,49_5.0
+49,3,9.93003674799368,5.0,0,1,0.0,5.0,-2,1,1.0,49_5.0
+49,4,9.728215912860774,5.0,0,1,0.0,5.0,-1,1,1.0,49_5.0
+49,5,12.103403154230948,5.0,1,1,2.0,5.0,0,1,1.0,49_5.0
+49,6,11.538525743370705,5.0,1,1,2.0,5.0,1,1,1.0,49_5.0
+49,7,12.200705227099014,5.0,1,1,2.0,5.0,2,1,1.0,49_5.0
diff --git a/benchmarks/data/stacked_did_golden.json b/benchmarks/data/stacked_did_golden.json
new file mode 100644
index 00000000..9431fd27
--- /dev/null
+++ b/benchmarks/data/stacked_did_golden.json
@@ -0,0 +1,141 @@
+{
+  "meta": {
+    "clubSandwich_version": "0.7.0",
+    "R_version": "R version 4.5.2 (2025-10-31)",
+    "panel_descriptor": {
+      "n_units": 50,
+      "n_periods": 8,
+      "cohort_periods": [3, 5, 7],
+      "never_treated_frac": 0.3,
+      "treatment_effect": 2,
+      "dynamic_effects": false,
+      "seed": 20260521
+    },
+    "estimator_descriptor": {
+      "kappa_pre": 2,
+      "kappa_post": 2
+    },
+    "event_times_non_ref": [-2, 0, 1, 2],
+    "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"]
+  },
+  "unit": {
+    "se_cr1_es": [0.200078483415387, 0.1826256881800713, 0.1790924462926418, 0.2225451518363201],
+    "se_cr2_es": [0.200215877612203, 0.1826114978147135, 0.1783881839488361, 0.2223946838242654],
+    "dof_bm_es": [42.220546102162736, 42.220546102162778, 42.220546102162771, 42.220546102162743],
+    "dof_bm_overall": 42.220546102163503,
+    "se_overall_cr1": 0.1569914942914515,
+    "se_overall_cr2": 0.1568044677648716,
+    "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+    "se_cr1_intercept": 0.5002307674660508,
+    "se_cr2_intercept": 0.5024664335898475,
+    "coef_es": [0.1405590947363666, 2.203920170110117, 1.987787707839172, 2.050642070438884]
+  },
+  "unit_subexp": {
+    "se_cr1_es": [0.1870363047946401, 0.1822840556079794, 0.1981449756819671, 0.2273611013086859],
+    "se_cr2_es": [0.1864890303402853, 0.1817266783762599, 0.1974749196027261, 0.2268750375440091],
+    "dof_bm_es": [48.843509657306811, 48.843509657306825, 48.843509657306839, 48.843509657306868],
+    "dof_bm_overall": 48.843509657305873,
+    "se_overall_cr1": 0.1632258518454228,
+    "se_overall_cr2": 0.1628147053040843,
+    "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+    "se_cr1_intercept": 0.3749306928695448,
+    "se_cr2_intercept": 0.3728786437332218,
+    "coef_es": [0.1405590947363666, 2.203920170110117, 1.987787707839172, 2.050642070438884]
+  },
+  "population": {
+    "meta": {
+      "event_times_non_ref": [-2, 0, 1, 2],
+      "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+      "ref_period": -1,
+      "anticipation": 0
+    },
+    "unit": {
+      "se_cr1_es": [0.2055876355330852, 0.1872478941072777, 0.1804446738410741, 0.2304790296984253],
+      "se_cr2_es": [0.2059668479962416, 0.1874175980522263, 0.1798116646995264, 0.2305505425313663],
+      "dof_bm_es": [41.119689028899977, 41.11968902890009, 41.119689028899977, 41.119689028900027],
+      "dof_bm_overall": 41.119689028899927,
+      "se_overall_cr1": 0.1617459655611932,
+      "se_overall_cr2": 0.1617028169620642,
+      "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+      "se_cr1_intercept": 0.5092056342946343,
+      "se_cr2_intercept": 0.5122813872396925,
+      "coef_es": [0.1592684537601917, 2.202558541183745, 2.011016359321843, 2.067053388909866]
+    },
+    "unit_subexp": {
+      "se_cr1_es": [0.1931159955471998, 0.1868877212425096, 0.1989124404085452, 0.2350017927907321],
+      "se_cr2_es": [0.1928712028049674, 0.1865877414187149, 0.1983942713018044, 0.234830388099705],
+      "dof_bm_es": [46.762096192405849, 46.762096192405608, 46.7620961924057, 46.762096192405586],
+      "dof_bm_overall": 46.762096192407405,
+      "se_overall_cr1": 0.1676538114692963,
+      "se_overall_cr2": 0.1674538455644745,
+      "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+      "se_cr1_intercept": 0.3897579339609716,
+      "se_cr2_intercept": 0.388735182137397,
+      "coef_es": [0.1592684537601917, 2.202558541183745, 2.011016359321843, 2.067053388909866]
+    }
+  },
+  "anticipation1": {
+    "meta": {
+      "event_times_non_ref": [-3, -1, 0, 1, 2],
+      "es_coef_names": ["delta_neg3", "delta_neg1", "delta_0", "delta_1", "delta_2"],
+      "ref_period": -2,
+      "anticipation": 1
+    },
+    "unit": {
+      "se_cr1_es": [0.2048338447728567, 0.2001299373145117, 0.1530282199734606, 0.1835610737502558, 0.1642185283764434],
+      "se_cr2_es": [0.2046593583984913, 0.2002158776122035, 0.1527284404654759, 0.1834505565682085, 0.1640239364813267],
+      "dof_bm_es": [42.220546102162807, 42.220546102162729, 42.220546102162757, 42.220546102162771, 42.220546102162835],
+      "dof_bm_overall": 42.220546102163553,
+      "se_overall_cr1": 0.1270856646130946,
+      "se_overall_cr2": 0.1271454151934584,
+      "es_coef_names": ["delta_neg3", "delta_neg1", "delta_0", "delta_1", "delta_2"],
+      "se_cr1_intercept": 0.5197898391070184,
+      "se_cr2_intercept": 0.5219118652053266,
+      "coef_es": [-0.09202034132318355, -0.1405590947363716, 2.06336107537375, 1.8472286131028, 1.910082975702512]
+    },
+    "unit_subexp": {
+      "se_cr1_es": [0.1960381429822047, 0.1870844046552268, 0.1697544218508184, 0.1683966164731217, 0.1545688709713832],
+      "se_cr2_es": [0.1953831407900792, 0.1864890303402968, 0.1692315113031761, 0.1676285479403536, 0.1538971704311764],
+      "dof_bm_es": [48.843509657306775, 48.843509657306655, 48.843509657306768, 48.843509657306818, 48.843509657306903],
+      "dof_bm_overall": 48.843509657305958,
+      "se_overall_cr1": 0.113794880953498,
+      "se_overall_cr2": 0.1133052662417925,
+      "es_coef_names": ["delta_neg3", "delta_neg1", "delta_0", "delta_1", "delta_2"],
+      "se_cr1_intercept": 0.3896150813864238,
+      "se_cr2_intercept": 0.387454300846914,
+      "coef_es": [-0.09202034132318355, -0.1405590947363716, 2.06336107537375, 1.8472286131028, 1.910082975702512]
+    }
+  },
+  "sample_share": {
+    "meta": {
+      "event_times_non_ref": [-2, 0, 1, 2],
+      "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+      "ref_period": -1,
+      "anticipation": 0
+    },
+    "unit": {
+      "se_cr1_es": [0.1926945586681986, 0.1774065826127712, 0.179551540289758, 0.2122392518497908],
+      "se_cr2_es": [0.1924916107265304, 0.1771398302930658, 0.1787869223969842, 0.211801063322419],
+      "dof_bm_es": [43.616923190028679, 43.616923190028658, 43.616923190028672, 43.616923190028579],
+      "dof_bm_overall": 43.616923190028821,
+      "se_overall_cr1": 0.1513357911957259,
+      "se_overall_cr2": 0.1509672871980841,
+      "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+      "se_cr1_intercept": 0.4871327984687643,
+      "se_cr2_intercept": 0.4880622063955147,
+      "coef_es": [0.1108321656167259, 2.20608363527907, 1.950880164652265, 2.024566452331055]
+    },
+    "unit_subexp": {
+      "se_cr1_es": [0.1793915271522487, 0.1770839950451901, 0.1982000403327432, 0.217155905557332],
+      "se_cr2_es": [0.17853000329591, 0.1762727793305481, 0.1973992806659578, 0.2163342225610217],
+      "dof_bm_es": [50.784770788284384, 50.784770788284128, 50.78477078828427, 50.784770788284149],
+      "dof_bm_overall": 50.78477078828643,
+      "se_overall_cr1": 0.1576457481577655,
+      "se_overall_cr2": 0.1570178024670955,
+      "es_coef_names": ["delta_neg2", "delta_0", "delta_1", "delta_2"],
+      "se_cr1_intercept": 0.3600565608113597,
+      "se_cr2_intercept": 0.3569865903344588,
+      "coef_es": [0.1108321656167259, 2.20608363527907, 1.950880164652265, 2.024566452331055]
+    }
+  }
+}
diff --git a/benchmarks/data/stacked_did_population_panel.csv b/benchmarks/data/stacked_did_population_panel.csv
new file mode 100644
index 00000000..387f252e
--- /dev/null
+++ b/benchmarks/data/stacked_did_population_panel.csv
@@ -0,0 +1,326 @@
+unit,period,outcome,first_treat,treated,treat,true_effect,pop,_sub_exp,_event_time,_D_sa,_Q_weight,unit_subexp
+0,1,7.483662635760572,inf,0,0,0.0,51.09703678029105,3.0,-2,0,0.6781447450134034,0_3.0
+0,2,6.841502772126143,inf,0,0,0.0,51.09703678029105,3.0,-1,0,0.6781447450134034,0_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,51.09703678029105,3.0,0,0,0.6781447450134034,0_3.0
+0,4,7.947042738617837,inf,0,0,0.0,51.09703678029105,3.0,1,0,0.6781447450134034,0_3.0
+0,5,6.317944717889617,inf,0,0,0.0,51.09703678029105,3.0,2,0,0.6781447450134034,0_3.0
+1,1,9.509053965877307,inf,0,0,0.0,56.94407154666434,3.0,-2,0,0.6781447450134034,1_3.0
+1,2,10.614366425343084,inf,0,0,0.0,56.94407154666434,3.0,-1,0,0.6781447450134034,1_3.0
+1,3,9.423988866091562,inf,0,0,0.0,56.94407154666434,3.0,0,0,0.6781447450134034,1_3.0
+1,4,11.746966365956794,inf,0,0,0.0,56.94407154666434,3.0,1,0,0.6781447450134034,1_3.0
+1,5,10.014392475247982,inf,0,0,0.0,56.94407154666434,3.0,2,0,0.6781447450134034,1_3.0
+2,1,8.504571437173666,inf,0,0,0.0,51.679680009513135,3.0,-2,0,0.6781447450134034,2_3.0
+2,2,7.968205325235655,inf,0,0,0.0,51.679680009513135,3.0,-1,0,0.6781447450134034,2_3.0
+2,3,8.129454052191617,inf,0,0,0.0,51.679680009513135,3.0,0,0,0.6781447450134034,2_3.0
+2,4,8.633961259377445,inf,0,0,0.0,51.679680009513135,3.0,1,0,0.6781447450134034,2_3.0
+2,5,9.895662186433224,inf,0,0,0.0,51.679680009513135,3.0,2,0,0.6781447450134034,2_3.0
+3,1,13.825385295202855,inf,0,0,0.0,97.24645084148509,3.0,-2,0,0.6781447450134034,3_3.0
+3,2,13.829286373980507,inf,0,0,0.0,97.24645084148509,3.0,-1,0,0.6781447450134034,3_3.0
+3,3,13.948413480279912,inf,0,0,0.0,97.24645084148509,3.0,0,0,0.6781447450134034,3_3.0
+3,4,13.516792177472347,inf,0,0,0.0,97.24645084148509,3.0,1,0,0.6781447450134034,3_3.0
+3,5,14.119361417606083,inf,0,0,0.0,97.24645084148509,3.0,2,0,0.6781447450134034,3_3.0
+4,1,11.72507772055074,inf,0,0,0.0,61.87541105494325,3.0,-2,0,0.6781447450134034,4_3.0
+4,2,11.912646534344725,inf,0,0,0.0,61.87541105494325,3.0,-1,0,0.6781447450134034,4_3.0
+4,3,11.813278538150167,inf,0,0,0.0,61.87541105494325,3.0,0,0,0.6781447450134034,4_3.0
+4,4,13.525689949379311,inf,0,0,0.0,61.87541105494325,3.0,1,0,0.6781447450134034,4_3.0
+4,5,12.732831972997747,inf,0,0,0.0,61.87541105494325,3.0,2,0,0.6781447450134034,4_3.0
+5,1,6.378890705428073,inf,0,0,0.0,57.26006629156302,3.0,-2,0,0.6781447450134034,5_3.0
+5,2,6.440945780933881,inf,0,0,0.0,57.26006629156302,3.0,-1,0,0.6781447450134034,5_3.0
+5,3,6.6127868751986805,inf,0,0,0.0,57.26006629156302,3.0,0,0,0.6781447450134034,5_3.0
+5,4,6.513422311622866,inf,0,0,0.0,57.26006629156302,3.0,1,0,0.6781447450134034,5_3.0
+5,5,6.75979943656341,inf,0,0,0.0,57.26006629156302,3.0,2,0,0.6781447450134034,5_3.0
+6,1,8.633688139950424,inf,0,0,0.0,29.391885564044067,3.0,-2,0,0.6781447450134034,6_3.0
+6,2,8.600953374134306,inf,0,0,0.0,29.391885564044067,3.0,-1,0,0.6781447450134034,6_3.0
+6,3,8.411617732005928,inf,0,0,0.0,29.391885564044067,3.0,0,0,0.6781447450134034,6_3.0
+6,4,8.387370342059572,inf,0,0,0.0,29.391885564044067,3.0,1,0,0.6781447450134034,6_3.0
+6,5,8.945688592547397,inf,0,0,0.0,29.391885564044067,3.0,2,0,0.6781447450134034,6_3.0
+7,1,11.334453872912409,inf,0,0,0.0,55.896633871640766,3.0,-2,0,0.6781447450134034,7_3.0
+7,2,10.898891009750589,inf,0,0,0.0,55.896633871640766,3.0,-1,0,0.6781447450134034,7_3.0
+7,3,10.456288009791654,inf,0,0,0.0,55.896633871640766,3.0,0,0,0.6781447450134034,7_3.0
+7,4,11.086419785594533,inf,0,0,0.0,55.896633871640766,3.0,1,0,0.6781447450134034,7_3.0
+7,5,11.136486986823087,inf,0,0,0.0,55.896633871640766,3.0,2,0,0.6781447450134034,7_3.0
+8,1,12.363906817019398,inf,0,0,0.0,47.284829029539345,3.0,-2,0,0.6781447450134034,8_3.0
+8,2,13.349921038000932,inf,0,0,0.0,47.284829029539345,3.0,-1,0,0.6781447450134034,8_3.0
+8,3,11.289014907437469,inf,0,0,0.0,47.284829029539345,3.0,0,0,0.6781447450134034,8_3.0
+8,4,12.413821661809754,inf,0,0,0.0,47.284829029539345,3.0,1,0,0.6781447450134034,8_3.0
+8,5,12.599831686253742,inf,0,0,0.0,47.284829029539345,3.0,2,0,0.6781447450134034,8_3.0
+9,1,8.226753444350999,inf,0,0,0.0,61.39574279392427,3.0,-2,0,0.6781447450134034,9_3.0
+9,2,8.04357609368333,inf,0,0,0.0,61.39574279392427,3.0,-1,0,0.6781447450134034,9_3.0
+9,3,7.935929232288304,inf,0,0,0.0,61.39574279392427,3.0,0,0,0.6781447450134034,9_3.0
+9,4,8.947274202622067,inf,0,0,0.0,61.39574279392427,3.0,1,0,0.6781447450134034,9_3.0
+9,5,8.525815827970705,inf,0,0,0.0,61.39574279392427,3.0,2,0,0.6781447450134034,9_3.0
+10,1,6.616529053324495,inf,0,0,0.0,93.11380727028694,3.0,-2,0,0.6781447450134034,10_3.0
+10,2,7.904056043755299,inf,0,0,0.0,93.11380727028694,3.0,-1,0,0.6781447450134034,10_3.0
+10,3,7.385638433127546,inf,0,0,0.0,93.11380727028694,3.0,0,0,0.6781447450134034,10_3.0
+10,4,7.6853500749042345,inf,0,0,0.0,93.11380727028694,3.0,1,0,0.6781447450134034,10_3.0
+10,5,7.841052147984791,inf,0,0,0.0,93.11380727028694,3.0,2,0,0.6781447450134034,10_3.0
+11,1,8.904134028447222,inf,0,0,0.0,25.342652980402878,3.0,-2,0,0.6781447450134034,11_3.0
+11,2,9.722452361429282,inf,0,0,0.0,25.342652980402878,3.0,-1,0,0.6781447450134034,11_3.0
+11,3,8.983212026944074,inf,0,0,0.0,25.342652980402878,3.0,0,0,0.6781447450134034,11_3.0
+11,4,9.774571594385389,inf,0,0,0.0,25.342652980402878,3.0,1,0,0.6781447450134034,11_3.0
+11,5,9.962850837041142,inf,0,0,0.0,25.342652980402878,3.0,2,0,0.6781447450134034,11_3.0
+12,1,14.247804715186929,inf,0,0,0.0,31.634395525708975,3.0,-2,0,0.6781447450134034,12_3.0
+12,2,13.59781501202838,inf,0,0,0.0,31.634395525708975,3.0,-1,0,0.6781447450134034,12_3.0
+12,3,14.268684067203575,inf,0,0,0.0,31.634395525708975,3.0,0,0,0.6781447450134034,12_3.0
+12,4,13.089643664151836,inf,0,0,0.0,31.634395525708975,3.0,1,0,0.6781447450134034,12_3.0
+12,5,13.739823106373846,inf,0,0,0.0,31.634395525708975,3.0,2,0,0.6781447450134034,12_3.0
+13,1,12.466946753384763,inf,0,0,0.0,39.716885452850825,3.0,-2,0,0.6781447450134034,13_3.0
+13,2,12.577723255899597,inf,0,0,0.0,39.716885452850825,3.0,-1,0,0.6781447450134034,13_3.0
+13,3,12.989441559722076,inf,0,0,0.0,39.716885452850825,3.0,0,0,0.6781447450134034,13_3.0
+13,4,12.274600589012058,inf,0,0,0.0,39.716885452850825,3.0,1,0,0.6781447450134034,13_3.0
+13,5,12.427220506745886,inf,0,0,0.0,39.716885452850825,3.0,2,0,0.6781447450134034,13_3.0
+14,1,9.321384661956385,inf,0,0,0.0,27.75690150753168,3.0,-2,0,0.6781447450134034,14_3.0
+14,2,8.537850792031616,inf,0,0,0.0,27.75690150753168,3.0,-1,0,0.6781447450134034,14_3.0
+14,3,9.118405772601042,inf,0,0,0.0,27.75690150753168,3.0,0,0,0.6781447450134034,14_3.0
+14,4,9.74251766740271,inf,0,0,0.0,27.75690150753168,3.0,1,0,0.6781447450134034,14_3.0
+14,5,9.254403864326273,inf,0,0,0.0,27.75690150753168,3.0,2,0,0.6781447450134034,14_3.0
+15,1,7.441009679583563,7.0,0,1,0.0,35.65147999343277,3.0,-2,0,0.6781447450134034,15_3.0
+15,2,8.485304881474772,7.0,0,1,0.0,35.65147999343277,3.0,-1,0,0.6781447450134034,15_3.0
+15,3,8.564340753151164,7.0,0,1,0.0,35.65147999343277,3.0,0,0,0.6781447450134034,15_3.0
+15,4,8.1815349724801,7.0,0,1,0.0,35.65147999343277,3.0,1,0,0.6781447450134034,15_3.0
+15,5,9.311637156674976,7.0,0,1,0.0,35.65147999343277,3.0,2,0,0.6781447450134034,15_3.0
+16,1,7.6836207307333035,3.0,0,1,0.0,93.68673941787888,3.0,-2,1,1.0,16_3.0
+16,2,8.357654569597338,3.0,0,1,0.0,93.68673941787888,3.0,-1,1,1.0,16_3.0
+16,3,10.056550820184588,3.0,1,1,2.0,93.68673941787888,3.0,0,1,1.0,16_3.0
+16,4,10.207378999928212,3.0,1,1,2.0,93.68673941787888,3.0,1,1,1.0,16_3.0
+16,5,10.096882823598278,3.0,1,1,2.0,93.68673941787888,3.0,2,1,1.0,16_3.0
+17,1,8.175606640239206,3.0,0,1,0.0,38.41053736783668,3.0,-2,1,1.0,17_3.0
+17,2,8.600350852687033,3.0,0,1,0.0,38.41053736783668,3.0,-1,1,1.0,17_3.0
+17,3,9.60085063995874,3.0,1,1,2.0,38.41053736783668,3.0,0,1,1.0,17_3.0
+17,4,11.265857113284081,3.0,1,1,2.0,38.41053736783668,3.0,1,1,1.0,17_3.0
+17,5,10.805731826455423,3.0,1,1,2.0,38.41053736783668,3.0,2,1,1.0,17_3.0
+19,1,8.287479893276272,3.0,0,1,0.0,5.016156798575755,3.0,-2,1,1.0,19_3.0
+19,2,8.838449778710965,3.0,0,1,0.0,5.016156798575755,3.0,-1,1,1.0,19_3.0
+19,3,10.301088160788524,3.0,1,1,2.0,5.016156798575755,3.0,0,1,1.0,19_3.0
+19,4,10.193467173412117,3.0,1,1,2.0,5.016156798575755,3.0,1,1,1.0,19_3.0
+19,5,10.93570451644154,3.0,1,1,2.0,5.016156798575755,3.0,2,1,1.0,19_3.0
+20,1,6.608825849966525,3.0,0,1,0.0,30.566903992571508,3.0,-2,1,1.0,20_3.0
+20,2,6.192567902449463,3.0,0,1,0.0,30.566903992571508,3.0,-1,1,1.0,20_3.0
+20,3,8.62011884779571,3.0,1,1,2.0,30.566903992571508,3.0,0,1,1.0,20_3.0
+20,4,9.534699361996061,3.0,1,1,2.0,30.566903992571508,3.0,1,1,1.0,20_3.0
+20,5,8.972714344039472,3.0,1,1,2.0,30.566903992571508,3.0,2,1,1.0,20_3.0
+23,1,11.395156619865356,7.0,0,1,0.0,89.09593115974947,3.0,-2,0,0.6781447450134034,23_3.0
+23,2,12.292733093551572,7.0,0,1,0.0,89.09593115974947,3.0,-1,0,0.6781447450134034,23_3.0
+23,3,10.987337838152467,7.0,0,1,0.0,89.09593115974947,3.0,0,0,0.6781447450134034,23_3.0
+23,4,12.858675160893522,7.0,0,1,0.0,89.09593115974947,3.0,1,0,0.6781447450134034,23_3.0
+23,5,11.476064698198886,7.0,0,1,0.0,89.09593115974947,3.0,2,0,0.6781447450134034,23_3.0
+26,1,10.668383782601722,3.0,0,1,0.0,58.952591834886505,3.0,-2,1,1.0,26_3.0
+26,2,10.446234601225571,3.0,0,1,0.0,58.952591834886505,3.0,-1,1,1.0,26_3.0
+26,3,13.556242815369252,3.0,1,1,2.0,58.952591834886505,3.0,0,1,1.0,26_3.0
+26,4,13.576729933292084,3.0,1,1,2.0,58.952591834886505,3.0,1,1,1.0,26_3.0
+26,5,13.922175150825803,3.0,1,1,2.0,58.952591834886505,3.0,2,1,1.0,26_3.0
+27,1,8.580588297539316,7.0,0,1,0.0,60.3140426816271,3.0,-2,0,0.6781447450134034,27_3.0
+27,2,9.053926240501355,7.0,0,1,0.0,60.3140426816271,3.0,-1,0,0.6781447450134034,27_3.0
+27,3,8.688974179281848,7.0,0,1,0.0,60.3140426816271,3.0,0,0,0.6781447450134034,27_3.0
+27,4,7.928575988525038,7.0,0,1,0.0,60.3140426816271,3.0,1,0,0.6781447450134034,27_3.0
+27,5,9.170520905605873,7.0,0,1,0.0,60.3140426816271,3.0,2,0,0.6781447450134034,27_3.0
+28,1,9.371848258038815,3.0,0,1,0.0,65.91287985914114,3.0,-2,1,1.0,28_3.0
+28,2,9.154422149577181,3.0,0,1,0.0,65.91287985914114,3.0,-1,1,1.0,28_3.0
+28,3,11.965196547208972,3.0,1,1,2.0,65.91287985914114,3.0,0,1,1.0,28_3.0
+28,4,11.302065396349953,3.0,1,1,2.0,65.91287985914114,3.0,1,1,1.0,28_3.0
+28,5,11.589704550706,3.0,1,1,2.0,65.91287985914114,3.0,2,1,1.0,28_3.0
+29,1,8.985010266355241,3.0,0,1,0.0,51.616916159081924,3.0,-2,1,1.0,29_3.0
+29,2,8.92226436961385,3.0,0,1,0.0,51.616916159081924,3.0,-1,1,1.0,29_3.0
+29,3,11.587183906367892,3.0,1,1,2.0,51.616916159081924,3.0,0,1,1.0,29_3.0
+29,4,11.061585148973101,3.0,1,1,2.0,51.616916159081924,3.0,1,1,1.0,29_3.0
+29,5,11.32705363934025,3.0,1,1,2.0,51.616916159081924,3.0,2,1,1.0,29_3.0
+30,1,8.376287388505098,3.0,0,1,0.0,20.174295613318602,3.0,-2,1,1.0,30_3.0
+30,2,8.657693030574496,3.0,0,1,0.0,20.174295613318602,3.0,-1,1,1.0,30_3.0
+30,3,10.18781936742593,3.0,1,1,2.0,20.174295613318602,3.0,0,1,1.0,30_3.0
+30,4,11.000739934625502,3.0,1,1,2.0,20.174295613318602,3.0,1,1,1.0,30_3.0
+30,5,10.282065104573295,3.0,1,1,2.0,20.174295613318602,3.0,2,1,1.0,30_3.0
+33,1,11.57454492762582,7.0,0,1,0.0,10.424090898727023,3.0,-2,0,0.6781447450134034,33_3.0
+33,2,11.691765125046665,7.0,0,1,0.0,10.424090898727023,3.0,-1,0,0.6781447450134034,33_3.0
+33,3,11.84979372696007,7.0,0,1,0.0,10.424090898727023,3.0,0,0,0.6781447450134034,33_3.0
+33,4,12.62424994943963,7.0,0,1,0.0,10.424090898727023,3.0,1,0,0.6781447450134034,33_3.0
+33,5,12.3254386483586,7.0,0,1,0.0,10.424090898727023,3.0,2,0,0.6781447450134034,33_3.0
+34,1,12.563769947545987,7.0,0,1,0.0,12.1998072950007,3.0,-2,0,0.6781447450134034,34_3.0
+34,2,11.966597306919729,7.0,0,1,0.0,12.1998072950007,3.0,-1,0,0.6781447450134034,34_3.0
+34,3,12.803090421682993,7.0,0,1,0.0,12.1998072950007,3.0,0,0,0.6781447450134034,34_3.0
+34,4,13.544985407250604,7.0,0,1,0.0,12.1998072950007,3.0,1,0,0.6781447450134034,34_3.0
+34,5,13.17341536766152,7.0,0,1,0.0,12.1998072950007,3.0,2,0,0.6781447450134034,34_3.0
+35,1,13.198034881842501,3.0,0,1,0.0,28.77799746301179,3.0,-2,1,1.0,35_3.0
+35,2,13.13803314498448,3.0,0,1,0.0,28.77799746301179,3.0,-1,1,1.0,35_3.0
+35,3,14.775491475016876,3.0,1,1,2.0,28.77799746301179,3.0,0,1,1.0,35_3.0
+35,4,14.755631298179171,3.0,1,1,2.0,28.77799746301179,3.0,1,1,1.0,35_3.0
+35,5,15.82726562084955,3.0,1,1,2.0,28.77799746301179,3.0,2,1,1.0,35_3.0
+36,1,7.2152253987939226,3.0,0,1,0.0,59.65911885671169,3.0,-2,1,1.0,36_3.0
+36,2,6.610853150968622,3.0,0,1,0.0,59.65911885671169,3.0,-1,1,1.0,36_3.0
+36,3,8.853563997233643,3.0,1,1,2.0,59.65911885671169,3.0,0,1,1.0,36_3.0
+36,4,9.09329932509169,3.0,1,1,2.0,59.65911885671169,3.0,1,1,1.0,36_3.0
+36,5,9.298377296285087,3.0,1,1,2.0,59.65911885671169,3.0,2,1,1.0,36_3.0
+37,1,10.146958297869952,7.0,0,1,0.0,34.979030166470274,3.0,-2,0,0.6781447450134034,37_3.0
+37,2,9.060497069998855,7.0,0,1,0.0,34.979030166470274,3.0,-1,0,0.6781447450134034,37_3.0
+37,3,9.318388049432945,7.0,0,1,0.0,34.979030166470274,3.0,0,0,0.6781447450134034,37_3.0
+37,4,9.26097646005531,7.0,0,1,0.0,34.979030166470274,3.0,1,0,0.6781447450134034,37_3.0
+37,5,8.919210921226544,7.0,0,1,0.0,34.979030166470274,3.0,2,0,0.6781447450134034,37_3.0
+39,1,12.353298527293795,3.0,0,1,0.0,96.57316078735487,3.0,-2,1,1.0,39_3.0
+39,2,12.242756403937744,3.0,0,1,0.0,96.57316078735487,3.0,-1,1,1.0,39_3.0
+39,3,15.149617439612653,3.0,1,1,2.0,96.57316078735487,3.0,0,1,1.0,39_3.0
+39,4,14.088452790656742,3.0,1,1,2.0,96.57316078735487,3.0,1,1,1.0,39_3.0
+39,5,14.060806514376624,3.0,1,1,2.0,96.57316078735487,3.0,2,1,1.0,39_3.0
+40,1,7.342148190809627,7.0,0,1,0.0,33.85216111249103,3.0,-2,0,0.6781447450134034,40_3.0
+40,2,8.356636279545437,7.0,0,1,0.0,33.85216111249103,3.0,-1,0,0.6781447450134034,40_3.0
+40,3,8.853341199215448,7.0,0,1,0.0,33.85216111249103,3.0,0,0,0.6781447450134034,40_3.0
+40,4,7.646273376822638,7.0,0,1,0.0,33.85216111249103,3.0,1,0,0.6781447450134034,40_3.0
+40,5,7.959815635944495,7.0,0,1,0.0,33.85216111249103,3.0,2,0,0.6781447450134034,40_3.0
+41,1,12.529859573199765,3.0,0,1,0.0,24.392128963266284,3.0,-2,1,1.0,41_3.0
+41,2,12.359862327664574,3.0,0,1,0.0,24.392128963266284,3.0,-1,1,1.0,41_3.0
+41,3,14.594595526225248,3.0,1,1,2.0,24.392128963266284,3.0,0,1,1.0,41_3.0
+41,4,14.627150386514797,3.0,1,1,2.0,24.392128963266284,3.0,1,1,1.0,41_3.0
+41,5,15.438306775111634,3.0,1,1,2.0,24.392128963266284,3.0,2,1,1.0,41_3.0
+43,1,11.589875560059742,7.0,0,1,0.0,57.69391670715631,3.0,-2,0,0.6781447450134034,43_3.0
+43,2,11.566062741673996,7.0,0,1,0.0,57.69391670715631,3.0,-1,0,0.6781447450134034,43_3.0
+43,3,11.081605645195344,7.0,0,1,0.0,57.69391670715631,3.0,0,0,0.6781447450134034,43_3.0
+43,4,11.740097124129901,7.0,0,1,0.0,57.69391670715631,3.0,1,0,0.6781447450134034,43_3.0
+43,5,12.466392539854581,7.0,0,1,0.0,57.69391670715631,3.0,2,0,0.6781447450134034,43_3.0
+45,1,8.77119845157113,7.0,0,1,0.0,62.074658112794324,3.0,-2,0,0.6781447450134034,45_3.0
+45,2,8.039535155953265,7.0,0,1,0.0,62.074658112794324,3.0,-1,0,0.6781447450134034,45_3.0
+45,3,8.077961859940832,7.0,0,1,0.0,62.074658112794324,3.0,0,0,0.6781447450134034,45_3.0
+45,4,8.881163504521098,7.0,0,1,0.0,62.074658112794324,3.0,1,0,0.6781447450134034,45_3.0
+45,5,8.557759225694356,7.0,0,1,0.0,62.074658112794324,3.0,2,0,0.6781447450134034,45_3.0
+46,1,12.148563177783899,7.0,0,1,0.0,28.67232929600455,3.0,-2,0,0.6781447450134034,46_3.0
+46,2,12.307754494386236,7.0,0,1,0.0,28.67232929600455,3.0,-1,0,0.6781447450134034,46_3.0
+46,3,12.393620432500153,7.0,0,1,0.0,28.67232929600455,3.0,0,0,0.6781447450134034,46_3.0
+46,4,13.035568610439492,7.0,0,1,0.0,28.67232929600455,3.0,1,0,0.6781447450134034,46_3.0
+46,5,12.995360934519939,7.0,0,1,0.0,28.67232929600455,3.0,2,0,0.6781447450134034,46_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,51.09703678029105,5.0,-2,0,1.5364254249776612,0_5.0
+0,4,7.947042738617837,inf,0,0,0.0,51.09703678029105,5.0,-1,0,1.5364254249776612,0_5.0
+0,5,6.317944717889617,inf,0,0,0.0,51.09703678029105,5.0,0,0,1.5364254249776612,0_5.0
+0,6,8.030151283551504,inf,0,0,0.0,51.09703678029105,5.0,1,0,1.5364254249776612,0_5.0
+0,7,7.857022192736064,inf,0,0,0.0,51.09703678029105,5.0,2,0,1.5364254249776612,0_5.0
+1,3,9.423988866091562,inf,0,0,0.0,56.94407154666434,5.0,-2,0,1.5364254249776612,1_5.0
+1,4,11.746966365956794,inf,0,0,0.0,56.94407154666434,5.0,-1,0,1.5364254249776612,1_5.0
+1,5,10.014392475247982,inf,0,0,0.0,56.94407154666434,5.0,0,0,1.5364254249776612,1_5.0
+1,6,10.364571813434443,inf,0,0,0.0,56.94407154666434,5.0,1,0,1.5364254249776612,1_5.0
+1,7,10.370697302329921,inf,0,0,0.0,56.94407154666434,5.0,2,0,1.5364254249776612,1_5.0
+2,3,8.129454052191617,inf,0,0,0.0,51.679680009513135,5.0,-2,0,1.5364254249776612,2_5.0
+2,4,8.633961259377445,inf,0,0,0.0,51.679680009513135,5.0,-1,0,1.5364254249776612,2_5.0
+2,5,9.895662186433224,inf,0,0,0.0,51.679680009513135,5.0,0,0,1.5364254249776612,2_5.0
+2,6,8.55134266155172,inf,0,0,0.0,51.679680009513135,5.0,1,0,1.5364254249776612,2_5.0
+2,7,8.517976639648715,inf,0,0,0.0,51.679680009513135,5.0,2,0,1.5364254249776612,2_5.0
+3,3,13.948413480279912,inf,0,0,0.0,97.24645084148509,5.0,-2,0,1.5364254249776612,3_5.0
+3,4,13.516792177472347,inf,0,0,0.0,97.24645084148509,5.0,-1,0,1.5364254249776612,3_5.0
+3,5,14.119361417606083,inf,0,0,0.0,97.24645084148509,5.0,0,0,1.5364254249776612,3_5.0
+3,6,13.338205224874367,inf,0,0,0.0,97.24645084148509,5.0,1,0,1.5364254249776612,3_5.0
+3,7,14.4234803289339,inf,0,0,0.0,97.24645084148509,5.0,2,0,1.5364254249776612,3_5.0
+4,3,11.813278538150167,inf,0,0,0.0,61.87541105494325,5.0,-2,0,1.5364254249776612,4_5.0
+4,4,13.525689949379311,inf,0,0,0.0,61.87541105494325,5.0,-1,0,1.5364254249776612,4_5.0
+4,5,12.732831972997747,inf,0,0,0.0,61.87541105494325,5.0,0,0,1.5364254249776612,4_5.0
+4,6,12.618074611389941,inf,0,0,0.0,61.87541105494325,5.0,1,0,1.5364254249776612,4_5.0
+4,7,11.481627079417956,inf,0,0,0.0,61.87541105494325,5.0,2,0,1.5364254249776612,4_5.0
+5,3,6.6127868751986805,inf,0,0,0.0,57.26006629156302,5.0,-2,0,1.5364254249776612,5_5.0
+5,4,6.513422311622866,inf,0,0,0.0,57.26006629156302,5.0,-1,0,1.5364254249776612,5_5.0
+5,5,6.75979943656341,inf,0,0,0.0,57.26006629156302,5.0,0,0,1.5364254249776612,5_5.0
+5,6,6.600117721454566,inf,0,0,0.0,57.26006629156302,5.0,1,0,1.5364254249776612,5_5.0
+5,7,7.410185509058965,inf,0,0,0.0,57.26006629156302,5.0,2,0,1.5364254249776612,5_5.0
+6,3,8.411617732005928,inf,0,0,0.0,29.391885564044067,5.0,-2,0,1.5364254249776612,6_5.0
+6,4,8.387370342059572,inf,0,0,0.0,29.391885564044067,5.0,-1,0,1.5364254249776612,6_5.0
+6,5,8.945688592547397,inf,0,0,0.0,29.391885564044067,5.0,0,0,1.5364254249776612,6_5.0
+6,6,9.424533467045443,inf,0,0,0.0,29.391885564044067,5.0,1,0,1.5364254249776612,6_5.0
+6,7,8.558459954832495,inf,0,0,0.0,29.391885564044067,5.0,2,0,1.5364254249776612,6_5.0
+7,3,10.456288009791654,inf,0,0,0.0,55.896633871640766,5.0,-2,0,1.5364254249776612,7_5.0
+7,4,11.086419785594533,inf,0,0,0.0,55.896633871640766,5.0,-1,0,1.5364254249776612,7_5.0
+7,5,11.136486986823087,inf,0,0,0.0,55.896633871640766,5.0,0,0,1.5364254249776612,7_5.0
+7,6,10.49707772184382,inf,0,0,0.0,55.896633871640766,5.0,1,0,1.5364254249776612,7_5.0
+7,7,11.220866186095618,inf,0,0,0.0,55.896633871640766,5.0,2,0,1.5364254249776612,7_5.0
+8,3,11.289014907437469,inf,0,0,0.0,47.284829029539345,5.0,-2,0,1.5364254249776612,8_5.0
+8,4,12.413821661809754,inf,0,0,0.0,47.284829029539345,5.0,-1,0,1.5364254249776612,8_5.0
+8,5,12.599831686253742,inf,0,0,0.0,47.284829029539345,5.0,0,0,1.5364254249776612,8_5.0
+8,6,12.770672804509399,inf,0,0,0.0,47.284829029539345,5.0,1,0,1.5364254249776612,8_5.0
+8,7,12.12649797254452,inf,0,0,0.0,47.284829029539345,5.0,2,0,1.5364254249776612,8_5.0
+9,3,7.935929232288304,inf,0,0,0.0,61.39574279392427,5.0,-2,0,1.5364254249776612,9_5.0
+9,4,8.947274202622067,inf,0,0,0.0,61.39574279392427,5.0,-1,0,1.5364254249776612,9_5.0
+9,5,8.525815827970705,inf,0,0,0.0,61.39574279392427,5.0,0,0,1.5364254249776612,9_5.0
+9,6,8.29259367272137,inf,0,0,0.0,61.39574279392427,5.0,1,0,1.5364254249776612,9_5.0
+9,7,7.755986984180787,inf,0,0,0.0,61.39574279392427,5.0,2,0,1.5364254249776612,9_5.0
+10,3,7.385638433127546,inf,0,0,0.0,93.11380727028694,5.0,-2,0,1.5364254249776612,10_5.0
+10,4,7.6853500749042345,inf,0,0,0.0,93.11380727028694,5.0,-1,0,1.5364254249776612,10_5.0
+10,5,7.841052147984791,inf,0,0,0.0,93.11380727028694,5.0,0,0,1.5364254249776612,10_5.0
+10,6,7.779603182233025,inf,0,0,0.0,93.11380727028694,5.0,1,0,1.5364254249776612,10_5.0
+10,7,9.262810775890333,inf,0,0,0.0,93.11380727028694,5.0,2,0,1.5364254249776612,10_5.0
+11,3,8.983212026944074,inf,0,0,0.0,25.342652980402878,5.0,-2,0,1.5364254249776612,11_5.0
+11,4,9.774571594385389,inf,0,0,0.0,25.342652980402878,5.0,-1,0,1.5364254249776612,11_5.0
+11,5,9.962850837041142,inf,0,0,0.0,25.342652980402878,5.0,0,0,1.5364254249776612,11_5.0
+11,6,9.456066431566448,inf,0,0,0.0,25.342652980402878,5.0,1,0,1.5364254249776612,11_5.0
+11,7,9.374591364502958,inf,0,0,0.0,25.342652980402878,5.0,2,0,1.5364254249776612,11_5.0
+12,3,14.268684067203575,inf,0,0,0.0,31.634395525708975,5.0,-2,0,1.5364254249776612,12_5.0
+12,4,13.089643664151836,inf,0,0,0.0,31.634395525708975,5.0,-1,0,1.5364254249776612,12_5.0
+12,5,13.739823106373846,inf,0,0,0.0,31.634395525708975,5.0,0,0,1.5364254249776612,12_5.0
+12,6,13.669519287347907,inf,0,0,0.0,31.634395525708975,5.0,1,0,1.5364254249776612,12_5.0
+12,7,13.99578598067457,inf,0,0,0.0,31.634395525708975,5.0,2,0,1.5364254249776612,12_5.0
+13,3,12.989441559722076,inf,0,0,0.0,39.716885452850825,5.0,-2,0,1.5364254249776612,13_5.0
+13,4,12.274600589012058,inf,0,0,0.0,39.716885452850825,5.0,-1,0,1.5364254249776612,13_5.0
+13,5,12.427220506745886,inf,0,0,0.0,39.716885452850825,5.0,0,0,1.5364254249776612,13_5.0
+13,6,12.754093760275953,inf,0,0,0.0,39.716885452850825,5.0,1,0,1.5364254249776612,13_5.0
+13,7,13.718433817762524,inf,0,0,0.0,39.716885452850825,5.0,2,0,1.5364254249776612,13_5.0
+14,3,9.118405772601042,inf,0,0,0.0,27.75690150753168,5.0,-2,0,1.5364254249776612,14_5.0
+14,4,9.74251766740271,inf,0,0,0.0,27.75690150753168,5.0,-1,0,1.5364254249776612,14_5.0
+14,5,9.254403864326273,inf,0,0,0.0,27.75690150753168,5.0,0,0,1.5364254249776612,14_5.0
+14,6,9.330459677777796,inf,0,0,0.0,27.75690150753168,5.0,1,0,1.5364254249776612,14_5.0
+14,7,9.00828730479836,inf,0,0,0.0,27.75690150753168,5.0,2,0,1.5364254249776612,14_5.0
+18,3,8.750153667767103,5.0,0,1,0.0,77.6902716711911,5.0,-2,1,1.0,18_5.0
+18,4,10.369418029568877,5.0,0,1,0.0,77.6902716711911,5.0,-1,1,1.0,18_5.0
+18,5,11.814853522513967,5.0,1,1,2.0,77.6902716711911,5.0,0,1,1.0,18_5.0
+18,6,11.880991600710768,5.0,1,1,2.0,77.6902716711911,5.0,1,1,1.0,18_5.0
+18,7,12.084939640054936,5.0,1,1,2.0,77.6902716711911,5.0,2,1,1.0,18_5.0
+21,3,11.130313135937552,5.0,0,1,0.0,70.55652351914023,5.0,-2,1,1.0,21_5.0
+21,4,11.783557955755562,5.0,0,1,0.0,70.55652351914023,5.0,-1,1,1.0,21_5.0
+21,5,13.932178697330485,5.0,1,1,2.0,70.55652351914023,5.0,0,1,1.0,21_5.0
+21,6,13.311045276660398,5.0,1,1,2.0,70.55652351914023,5.0,1,1,1.0,21_5.0
+21,7,12.672260765381138,5.0,1,1,2.0,70.55652351914023,5.0,2,1,1.0,21_5.0
+22,3,10.370162504733846,5.0,0,1,0.0,45.77300946437004,5.0,-2,1,1.0,22_5.0
+22,4,10.130795877474743,5.0,0,1,0.0,45.77300946437004,5.0,-1,1,1.0,22_5.0
+22,5,13.327665186472212,5.0,1,1,2.0,45.77300946437004,5.0,0,1,1.0,22_5.0
+22,6,14.006133024294815,5.0,1,1,2.0,45.77300946437004,5.0,1,1,1.0,22_5.0
+22,7,11.997025447630525,5.0,1,1,2.0,45.77300946437004,5.0,2,1,1.0,22_5.0
+24,3,12.216528415765985,5.0,0,1,0.0,44.074903125040095,5.0,-2,1,1.0,24_5.0
+24,4,11.844123402747586,5.0,0,1,0.0,44.074903125040095,5.0,-1,1,1.0,24_5.0
+24,5,13.783296021523771,5.0,1,1,2.0,44.074903125040095,5.0,0,1,1.0,24_5.0
+24,6,13.374452489632073,5.0,1,1,2.0,44.074903125040095,5.0,1,1,1.0,24_5.0
+24,7,14.257003341404754,5.0,1,1,2.0,44.074903125040095,5.0,2,1,1.0,24_5.0
+25,3,12.376139740663152,5.0,0,1,0.0,62.73241976658182,5.0,-2,1,1.0,25_5.0
+25,4,12.451932231507831,5.0,0,1,0.0,62.73241976658182,5.0,-1,1,1.0,25_5.0
+25,5,14.40345513916028,5.0,1,1,2.0,62.73241976658182,5.0,0,1,1.0,25_5.0
+25,6,15.399316274400507,5.0,1,1,2.0,62.73241976658182,5.0,1,1,1.0,25_5.0
+25,7,14.121018820768782,5.0,1,1,2.0,62.73241976658182,5.0,2,1,1.0,25_5.0
+31,3,5.699685628062173,5.0,0,1,0.0,7.827770838902669,5.0,-2,1,1.0,31_5.0
+31,4,6.640244849690189,5.0,0,1,0.0,7.827770838902669,5.0,-1,1,1.0,31_5.0
+31,5,8.80287858433971,5.0,1,1,2.0,7.827770838902669,5.0,0,1,1.0,31_5.0
+31,6,7.478011213888342,5.0,1,1,2.0,7.827770838902669,5.0,1,1,1.0,31_5.0
+31,7,7.785449789009849,5.0,1,1,2.0,7.827770838902669,5.0,2,1,1.0,31_5.0
+32,3,13.145351442221973,5.0,0,1,0.0,25.952523503204013,5.0,-2,1,1.0,32_5.0
+32,4,14.213246370804844,5.0,0,1,0.0,25.952523503204013,5.0,-1,1,1.0,32_5.0
+32,5,15.673222444412447,5.0,1,1,2.0,25.952523503204013,5.0,0,1,1.0,32_5.0
+32,6,14.660677350520446,5.0,1,1,2.0,25.952523503204013,5.0,1,1,1.0,32_5.0
+32,7,14.475511437433273,5.0,1,1,2.0,25.952523503204013,5.0,2,1,1.0,32_5.0
+38,3,9.198240092902193,5.0,0,1,0.0,78.00707396817663,5.0,-2,1,1.0,38_5.0
+38,4,8.489100540648527,5.0,0,1,0.0,78.00707396817663,5.0,-1,1,1.0,38_5.0
+38,5,11.45417220010028,5.0,1,1,2.0,78.00707396817663,5.0,0,1,1.0,38_5.0
+38,6,10.634610329560985,5.0,1,1,2.0,78.00707396817663,5.0,1,1,1.0,38_5.0
+38,7,12.158628732093817,5.0,1,1,2.0,78.00707396817663,5.0,2,1,1.0,38_5.0
+42,3,9.165900951128542,5.0,0,1,0.0,69.09529934600081,5.0,-2,1,1.0,42_5.0
+42,4,9.836927314238459,5.0,0,1,0.0,69.09529934600081,5.0,-1,1,1.0,42_5.0
+42,5,11.893273621648287,5.0,1,1,2.0,69.09529934600081,5.0,0,1,1.0,42_5.0
+42,6,10.811622966738568,5.0,1,1,2.0,69.09529934600081,5.0,1,1,1.0,42_5.0
+42,7,12.387788948877349,5.0,1,1,2.0,69.09529934600081,5.0,2,1,1.0,42_5.0
+44,3,13.218744977031886,5.0,0,1,0.0,77.18615844224195,5.0,-2,1,1.0,44_5.0
+44,4,13.584858653200563,5.0,0,1,0.0,77.18615844224195,5.0,-1,1,1.0,44_5.0
+44,5,15.869766671523484,5.0,1,1,2.0,77.18615844224195,5.0,0,1,1.0,44_5.0
+44,6,15.606050081127213,5.0,1,1,2.0,77.18615844224195,5.0,1,1,1.0,44_5.0
+44,7,14.741953786805245,5.0,1,1,2.0,77.18615844224195,5.0,2,1,1.0,44_5.0
+47,3,11.405672205690639,5.0,0,1,0.0,91.1995253876107,5.0,-2,1,1.0,47_5.0
+47,4,10.327840224798175,5.0,0,1,0.0,91.1995253876107,5.0,-1,1,1.0,47_5.0
+47,5,12.972063268138358,5.0,1,1,2.0,91.1995253876107,5.0,0,1,1.0,47_5.0
+47,6,13.3375192672357,5.0,1,1,2.0,91.1995253876107,5.0,1,1,1.0,47_5.0
+47,7,13.949251964590784,5.0,1,1,2.0,91.1995253876107,5.0,2,1,1.0,47_5.0
+48,3,10.036462523941848,5.0,0,1,0.0,57.72070965802602,5.0,-2,1,1.0,48_5.0
+48,4,10.557346448715858,5.0,0,1,0.0,57.72070965802602,5.0,-1,1,1.0,48_5.0
+48,5,11.320126492133841,5.0,1,1,2.0,57.72070965802602,5.0,0,1,1.0,48_5.0
+48,6,12.376923426574146,5.0,1,1,2.0,57.72070965802602,5.0,1,1,1.0,48_5.0
+48,7,12.52530519550149,5.0,1,1,2.0,57.72070965802602,5.0,2,1,1.0,48_5.0
+49,3,9.93003674799368,5.0,0,1,0.0,72.11274206787415,5.0,-2,1,1.0,49_5.0
+49,4,9.728215912860774,5.0,0,1,0.0,72.11274206787415,5.0,-1,1,1.0,49_5.0
+49,5,12.103403154230948,5.0,1,1,2.0,72.11274206787415,5.0,0,1,1.0,49_5.0
+49,6,11.538525743370705,5.0,1,1,2.0,72.11274206787415,5.0,1,1,1.0,49_5.0
+49,7,12.200705227099014,5.0,1,1,2.0,72.11274206787415,5.0,2,1,1.0,49_5.0
diff --git a/benchmarks/data/stacked_did_sample_share_panel.csv b/benchmarks/data/stacked_did_sample_share_panel.csv
new file mode 100644
index 00000000..1166ba81
--- /dev/null
+++ b/benchmarks/data/stacked_did_sample_share_panel.csv
@@ -0,0 +1,326 @@
+unit,period,outcome,first_treat,treated,treat,true_effect,_sub_exp,_event_time,_D_sa,_Q_weight,unit_subexp
+0,1,7.483662635760572,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,0_3.0
+0,2,6.841502772126143,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,0_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,3.0,0,0,0.9107692307692308,0_3.0
+0,4,7.947042738617837,inf,0,0,0.0,3.0,1,0,0.9107692307692308,0_3.0
+0,5,6.317944717889617,inf,0,0,0.0,3.0,2,0,0.9107692307692308,0_3.0
+1,1,9.509053965877307,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,1_3.0
+1,2,10.614366425343084,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,1_3.0
+1,3,9.423988866091562,inf,0,0,0.0,3.0,0,0,0.9107692307692308,1_3.0
+1,4,11.746966365956794,inf,0,0,0.0,3.0,1,0,0.9107692307692308,1_3.0
+1,5,10.014392475247982,inf,0,0,0.0,3.0,2,0,0.9107692307692308,1_3.0
+2,1,8.504571437173666,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,2_3.0
+2,2,7.968205325235655,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,2_3.0
+2,3,8.129454052191617,inf,0,0,0.0,3.0,0,0,0.9107692307692308,2_3.0
+2,4,8.633961259377445,inf,0,0,0.0,3.0,1,0,0.9107692307692308,2_3.0
+2,5,9.895662186433224,inf,0,0,0.0,3.0,2,0,0.9107692307692308,2_3.0
+3,1,13.825385295202855,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,3_3.0
+3,2,13.829286373980507,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,3_3.0
+3,3,13.948413480279912,inf,0,0,0.0,3.0,0,0,0.9107692307692308,3_3.0
+3,4,13.516792177472347,inf,0,0,0.0,3.0,1,0,0.9107692307692308,3_3.0
+3,5,14.119361417606083,inf,0,0,0.0,3.0,2,0,0.9107692307692308,3_3.0
+4,1,11.72507772055074,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,4_3.0
+4,2,11.912646534344725,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,4_3.0
+4,3,11.813278538150167,inf,0,0,0.0,3.0,0,0,0.9107692307692308,4_3.0
+4,4,13.525689949379311,inf,0,0,0.0,3.0,1,0,0.9107692307692308,4_3.0
+4,5,12.732831972997747,inf,0,0,0.0,3.0,2,0,0.9107692307692308,4_3.0
+5,1,6.378890705428073,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,5_3.0
+5,2,6.440945780933881,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,5_3.0
+5,3,6.6127868751986805,inf,0,0,0.0,3.0,0,0,0.9107692307692308,5_3.0
+5,4,6.513422311622866,inf,0,0,0.0,3.0,1,0,0.9107692307692308,5_3.0
+5,5,6.75979943656341,inf,0,0,0.0,3.0,2,0,0.9107692307692308,5_3.0
+6,1,8.633688139950424,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,6_3.0
+6,2,8.600953374134306,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,6_3.0
+6,3,8.411617732005928,inf,0,0,0.0,3.0,0,0,0.9107692307692308,6_3.0
+6,4,8.387370342059572,inf,0,0,0.0,3.0,1,0,0.9107692307692308,6_3.0
+6,5,8.945688592547397,inf,0,0,0.0,3.0,2,0,0.9107692307692308,6_3.0
+7,1,11.334453872912409,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,7_3.0
+7,2,10.898891009750589,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,7_3.0
+7,3,10.456288009791654,inf,0,0,0.0,3.0,0,0,0.9107692307692308,7_3.0
+7,4,11.086419785594533,inf,0,0,0.0,3.0,1,0,0.9107692307692308,7_3.0
+7,5,11.136486986823087,inf,0,0,0.0,3.0,2,0,0.9107692307692308,7_3.0
+8,1,12.363906817019398,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,8_3.0
+8,2,13.349921038000932,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,8_3.0
+8,3,11.289014907437469,inf,0,0,0.0,3.0,0,0,0.9107692307692308,8_3.0
+8,4,12.413821661809754,inf,0,0,0.0,3.0,1,0,0.9107692307692308,8_3.0
+8,5,12.599831686253742,inf,0,0,0.0,3.0,2,0,0.9107692307692308,8_3.0
+9,1,8.226753444350999,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,9_3.0
+9,2,8.04357609368333,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,9_3.0
+9,3,7.935929232288304,inf,0,0,0.0,3.0,0,0,0.9107692307692308,9_3.0
+9,4,8.947274202622067,inf,0,0,0.0,3.0,1,0,0.9107692307692308,9_3.0
+9,5,8.525815827970705,inf,0,0,0.0,3.0,2,0,0.9107692307692308,9_3.0
+10,1,6.616529053324495,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,10_3.0
+10,2,7.904056043755299,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,10_3.0
+10,3,7.385638433127546,inf,0,0,0.0,3.0,0,0,0.9107692307692308,10_3.0
+10,4,7.6853500749042345,inf,0,0,0.0,3.0,1,0,0.9107692307692308,10_3.0
+10,5,7.841052147984791,inf,0,0,0.0,3.0,2,0,0.9107692307692308,10_3.0
+11,1,8.904134028447222,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,11_3.0
+11,2,9.722452361429282,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,11_3.0
+11,3,8.983212026944074,inf,0,0,0.0,3.0,0,0,0.9107692307692308,11_3.0
+11,4,9.774571594385389,inf,0,0,0.0,3.0,1,0,0.9107692307692308,11_3.0
+11,5,9.962850837041142,inf,0,0,0.0,3.0,2,0,0.9107692307692308,11_3.0
+12,1,14.247804715186929,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,12_3.0
+12,2,13.59781501202838,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,12_3.0
+12,3,14.268684067203575,inf,0,0,0.0,3.0,0,0,0.9107692307692308,12_3.0
+12,4,13.089643664151836,inf,0,0,0.0,3.0,1,0,0.9107692307692308,12_3.0
+12,5,13.739823106373846,inf,0,0,0.0,3.0,2,0,0.9107692307692308,12_3.0
+13,1,12.466946753384763,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,13_3.0
+13,2,12.577723255899597,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,13_3.0
+13,3,12.989441559722076,inf,0,0,0.0,3.0,0,0,0.9107692307692308,13_3.0
+13,4,12.274600589012058,inf,0,0,0.0,3.0,1,0,0.9107692307692308,13_3.0
+13,5,12.427220506745886,inf,0,0,0.0,3.0,2,0,0.9107692307692308,13_3.0
+14,1,9.321384661956385,inf,0,0,0.0,3.0,-2,0,0.9107692307692308,14_3.0
+14,2,8.537850792031616,inf,0,0,0.0,3.0,-1,0,0.9107692307692308,14_3.0
+14,3,9.118405772601042,inf,0,0,0.0,3.0,0,0,0.9107692307692308,14_3.0
+14,4,9.74251766740271,inf,0,0,0.0,3.0,1,0,0.9107692307692308,14_3.0
+14,5,9.254403864326273,inf,0,0,0.0,3.0,2,0,0.9107692307692308,14_3.0
+15,1,7.441009679583563,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,15_3.0
+15,2,8.485304881474772,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,15_3.0
+15,3,8.564340753151164,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,15_3.0
+15,4,8.1815349724801,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,15_3.0
+15,5,9.311637156674976,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,15_3.0
+16,1,7.6836207307333035,3.0,0,1,0.0,3.0,-2,1,1.0,16_3.0
+16,2,8.357654569597338,3.0,0,1,0.0,3.0,-1,1,1.0,16_3.0
+16,3,10.056550820184588,3.0,1,1,2.0,3.0,0,1,1.0,16_3.0
+16,4,10.207378999928212,3.0,1,1,2.0,3.0,1,1,1.0,16_3.0
+16,5,10.096882823598278,3.0,1,1,2.0,3.0,2,1,1.0,16_3.0
+17,1,8.175606640239206,3.0,0,1,0.0,3.0,-2,1,1.0,17_3.0
+17,2,8.600350852687033,3.0,0,1,0.0,3.0,-1,1,1.0,17_3.0
+17,3,9.60085063995874,3.0,1,1,2.0,3.0,0,1,1.0,17_3.0
+17,4,11.265857113284081,3.0,1,1,2.0,3.0,1,1,1.0,17_3.0
+17,5,10.805731826455423,3.0,1,1,2.0,3.0,2,1,1.0,17_3.0
+19,1,8.287479893276272,3.0,0,1,0.0,3.0,-2,1,1.0,19_3.0
+19,2,8.838449778710965,3.0,0,1,0.0,3.0,-1,1,1.0,19_3.0
+19,3,10.301088160788524,3.0,1,1,2.0,3.0,0,1,1.0,19_3.0
+19,4,10.193467173412117,3.0,1,1,2.0,3.0,1,1,1.0,19_3.0
+19,5,10.93570451644154,3.0,1,1,2.0,3.0,2,1,1.0,19_3.0
+20,1,6.608825849966525,3.0,0,1,0.0,3.0,-2,1,1.0,20_3.0
+20,2,6.192567902449463,3.0,0,1,0.0,3.0,-1,1,1.0,20_3.0
+20,3,8.62011884779571,3.0,1,1,2.0,3.0,0,1,1.0,20_3.0
+20,4,9.534699361996061,3.0,1,1,2.0,3.0,1,1,1.0,20_3.0
+20,5,8.972714344039472,3.0,1,1,2.0,3.0,2,1,1.0,20_3.0
+23,1,11.395156619865356,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,23_3.0
+23,2,12.292733093551572,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,23_3.0
+23,3,10.987337838152467,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,23_3.0
+23,4,12.858675160893522,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,23_3.0
+23,5,11.476064698198886,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,23_3.0
+26,1,10.668383782601722,3.0,0,1,0.0,3.0,-2,1,1.0,26_3.0
+26,2,10.446234601225571,3.0,0,1,0.0,3.0,-1,1,1.0,26_3.0
+26,3,13.556242815369252,3.0,1,1,2.0,3.0,0,1,1.0,26_3.0
+26,4,13.576729933292084,3.0,1,1,2.0,3.0,1,1,1.0,26_3.0
+26,5,13.922175150825803,3.0,1,1,2.0,3.0,2,1,1.0,26_3.0
+27,1,8.580588297539316,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,27_3.0
+27,2,9.053926240501355,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,27_3.0
+27,3,8.688974179281848,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,27_3.0
+27,4,7.928575988525038,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,27_3.0
+27,5,9.170520905605873,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,27_3.0
+28,1,9.371848258038815,3.0,0,1,0.0,3.0,-2,1,1.0,28_3.0
+28,2,9.154422149577181,3.0,0,1,0.0,3.0,-1,1,1.0,28_3.0
+28,3,11.965196547208972,3.0,1,1,2.0,3.0,0,1,1.0,28_3.0
+28,4,11.302065396349953,3.0,1,1,2.0,3.0,1,1,1.0,28_3.0
+28,5,11.589704550706,3.0,1,1,2.0,3.0,2,1,1.0,28_3.0
+29,1,8.985010266355241,3.0,0,1,0.0,3.0,-2,1,1.0,29_3.0
+29,2,8.92226436961385,3.0,0,1,0.0,3.0,-1,1,1.0,29_3.0
+29,3,11.587183906367892,3.0,1,1,2.0,3.0,0,1,1.0,29_3.0
+29,4,11.061585148973101,3.0,1,1,2.0,3.0,1,1,1.0,29_3.0
+29,5,11.32705363934025,3.0,1,1,2.0,3.0,2,1,1.0,29_3.0
+30,1,8.376287388505098,3.0,0,1,0.0,3.0,-2,1,1.0,30_3.0
+30,2,8.657693030574496,3.0,0,1,0.0,3.0,-1,1,1.0,30_3.0
+30,3,10.18781936742593,3.0,1,1,2.0,3.0,0,1,1.0,30_3.0
+30,4,11.000739934625502,3.0,1,1,2.0,3.0,1,1,1.0,30_3.0
+30,5,10.282065104573295,3.0,1,1,2.0,3.0,2,1,1.0,30_3.0
+33,1,11.57454492762582,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,33_3.0
+33,2,11.691765125046665,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,33_3.0
+33,3,11.84979372696007,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,33_3.0
+33,4,12.62424994943963,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,33_3.0
+33,5,12.3254386483586,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,33_3.0
+34,1,12.563769947545987,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,34_3.0
+34,2,11.966597306919729,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,34_3.0
+34,3,12.803090421682993,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,34_3.0
+34,4,13.544985407250604,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,34_3.0
+34,5,13.17341536766152,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,34_3.0
+35,1,13.198034881842501,3.0,0,1,0.0,3.0,-2,1,1.0,35_3.0
+35,2,13.13803314498448,3.0,0,1,0.0,3.0,-1,1,1.0,35_3.0
+35,3,14.775491475016876,3.0,1,1,2.0,3.0,0,1,1.0,35_3.0
+35,4,14.755631298179171,3.0,1,1,2.0,3.0,1,1,1.0,35_3.0
+35,5,15.82726562084955,3.0,1,1,2.0,3.0,2,1,1.0,35_3.0
+36,1,7.2152253987939226,3.0,0,1,0.0,3.0,-2,1,1.0,36_3.0
+36,2,6.610853150968622,3.0,0,1,0.0,3.0,-1,1,1.0,36_3.0
+36,3,8.853563997233643,3.0,1,1,2.0,3.0,0,1,1.0,36_3.0
+36,4,9.09329932509169,3.0,1,1,2.0,3.0,1,1,1.0,36_3.0
+36,5,9.298377296285087,3.0,1,1,2.0,3.0,2,1,1.0,36_3.0
+37,1,10.146958297869952,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,37_3.0
+37,2,9.060497069998855,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,37_3.0
+37,3,9.318388049432945,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,37_3.0
+37,4,9.26097646005531,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,37_3.0
+37,5,8.919210921226544,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,37_3.0
+39,1,12.353298527293795,3.0,0,1,0.0,3.0,-2,1,1.0,39_3.0
+39,2,12.242756403937744,3.0,0,1,0.0,3.0,-1,1,1.0,39_3.0
+39,3,15.149617439612653,3.0,1,1,2.0,3.0,0,1,1.0,39_3.0
+39,4,14.088452790656742,3.0,1,1,2.0,3.0,1,1,1.0,39_3.0
+39,5,14.060806514376624,3.0,1,1,2.0,3.0,2,1,1.0,39_3.0
+40,1,7.342148190809627,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,40_3.0
+40,2,8.356636279545437,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,40_3.0
+40,3,8.853341199215448,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,40_3.0
+40,4,7.646273376822638,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,40_3.0
+40,5,7.959815635944495,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,40_3.0
+41,1,12.529859573199765,3.0,0,1,0.0,3.0,-2,1,1.0,41_3.0
+41,2,12.359862327664574,3.0,0,1,0.0,3.0,-1,1,1.0,41_3.0
+41,3,14.594595526225248,3.0,1,1,2.0,3.0,0,1,1.0,41_3.0
+41,4,14.627150386514797,3.0,1,1,2.0,3.0,1,1,1.0,41_3.0
+41,5,15.438306775111634,3.0,1,1,2.0,3.0,2,1,1.0,41_3.0
+43,1,11.589875560059742,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,43_3.0
+43,2,11.566062741673996,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,43_3.0
+43,3,11.081605645195344,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,43_3.0
+43,4,11.740097124129901,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,43_3.0
+43,5,12.466392539854581,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,43_3.0
+45,1,8.77119845157113,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,45_3.0
+45,2,8.039535155953265,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,45_3.0
+45,3,8.077961859940832,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,45_3.0
+45,4,8.881163504521098,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,45_3.0
+45,5,8.557759225694356,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,45_3.0
+46,1,12.148563177783899,7.0,0,1,0.0,3.0,-2,0,0.9107692307692308,46_3.0
+46,2,12.307754494386236,7.0,0,1,0.0,3.0,-1,0,0.9107692307692308,46_3.0
+46,3,12.393620432500153,7.0,0,1,0.0,3.0,0,0,0.9107692307692308,46_3.0
+46,4,13.035568610439492,7.0,0,1,0.0,3.0,1,0,0.9107692307692308,46_3.0
+46,5,12.995360934519939,7.0,0,1,0.0,3.0,2,0,0.9107692307692308,46_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,0_5.0
+0,4,7.947042738617837,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,0_5.0
+0,5,6.317944717889617,inf,0,0,0.0,5.0,0,0,1.1487179487179489,0_5.0
+0,6,8.030151283551504,inf,0,0,0.0,5.0,1,0,1.1487179487179489,0_5.0
+0,7,7.857022192736064,inf,0,0,0.0,5.0,2,0,1.1487179487179489,0_5.0
+1,3,9.423988866091562,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,1_5.0
+1,4,11.746966365956794,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,1_5.0
+1,5,10.014392475247982,inf,0,0,0.0,5.0,0,0,1.1487179487179489,1_5.0
+1,6,10.364571813434443,inf,0,0,0.0,5.0,1,0,1.1487179487179489,1_5.0
+1,7,10.370697302329921,inf,0,0,0.0,5.0,2,0,1.1487179487179489,1_5.0
+2,3,8.129454052191617,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,2_5.0
+2,4,8.633961259377445,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,2_5.0
+2,5,9.895662186433224,inf,0,0,0.0,5.0,0,0,1.1487179487179489,2_5.0
+2,6,8.55134266155172,inf,0,0,0.0,5.0,1,0,1.1487179487179489,2_5.0
+2,7,8.517976639648715,inf,0,0,0.0,5.0,2,0,1.1487179487179489,2_5.0
+3,3,13.948413480279912,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,3_5.0
+3,4,13.516792177472347,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,3_5.0
+3,5,14.119361417606083,inf,0,0,0.0,5.0,0,0,1.1487179487179489,3_5.0
+3,6,13.338205224874367,inf,0,0,0.0,5.0,1,0,1.1487179487179489,3_5.0
+3,7,14.4234803289339,inf,0,0,0.0,5.0,2,0,1.1487179487179489,3_5.0
+4,3,11.813278538150167,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,4_5.0
+4,4,13.525689949379311,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,4_5.0
+4,5,12.732831972997747,inf,0,0,0.0,5.0,0,0,1.1487179487179489,4_5.0
+4,6,12.618074611389941,inf,0,0,0.0,5.0,1,0,1.1487179487179489,4_5.0
+4,7,11.481627079417956,inf,0,0,0.0,5.0,2,0,1.1487179487179489,4_5.0
+5,3,6.6127868751986805,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,5_5.0
+5,4,6.513422311622866,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,5_5.0
+5,5,6.75979943656341,inf,0,0,0.0,5.0,0,0,1.1487179487179489,5_5.0
+5,6,6.600117721454566,inf,0,0,0.0,5.0,1,0,1.1487179487179489,5_5.0
+5,7,7.410185509058965,inf,0,0,0.0,5.0,2,0,1.1487179487179489,5_5.0
+6,3,8.411617732005928,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,6_5.0
+6,4,8.387370342059572,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,6_5.0
+6,5,8.945688592547397,inf,0,0,0.0,5.0,0,0,1.1487179487179489,6_5.0
+6,6,9.424533467045443,inf,0,0,0.0,5.0,1,0,1.1487179487179489,6_5.0
+6,7,8.558459954832495,inf,0,0,0.0,5.0,2,0,1.1487179487179489,6_5.0
+7,3,10.456288009791654,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,7_5.0
+7,4,11.086419785594533,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,7_5.0
+7,5,11.136486986823087,inf,0,0,0.0,5.0,0,0,1.1487179487179489,7_5.0
+7,6,10.49707772184382,inf,0,0,0.0,5.0,1,0,1.1487179487179489,7_5.0
+7,7,11.220866186095618,inf,0,0,0.0,5.0,2,0,1.1487179487179489,7_5.0
+8,3,11.289014907437469,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,8_5.0
+8,4,12.413821661809754,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,8_5.0
+8,5,12.599831686253742,inf,0,0,0.0,5.0,0,0,1.1487179487179489,8_5.0
+8,6,12.770672804509399,inf,0,0,0.0,5.0,1,0,1.1487179487179489,8_5.0
+8,7,12.12649797254452,inf,0,0,0.0,5.0,2,0,1.1487179487179489,8_5.0
+9,3,7.935929232288304,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,9_5.0
+9,4,8.947274202622067,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,9_5.0
+9,5,8.525815827970705,inf,0,0,0.0,5.0,0,0,1.1487179487179489,9_5.0
+9,6,8.29259367272137,inf,0,0,0.0,5.0,1,0,1.1487179487179489,9_5.0
+9,7,7.755986984180787,inf,0,0,0.0,5.0,2,0,1.1487179487179489,9_5.0
+10,3,7.385638433127546,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,10_5.0
+10,4,7.6853500749042345,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,10_5.0
+10,5,7.841052147984791,inf,0,0,0.0,5.0,0,0,1.1487179487179489,10_5.0
+10,6,7.779603182233025,inf,0,0,0.0,5.0,1,0,1.1487179487179489,10_5.0
+10,7,9.262810775890333,inf,0,0,0.0,5.0,2,0,1.1487179487179489,10_5.0
+11,3,8.983212026944074,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,11_5.0
+11,4,9.774571594385389,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,11_5.0
+11,5,9.962850837041142,inf,0,0,0.0,5.0,0,0,1.1487179487179489,11_5.0
+11,6,9.456066431566448,inf,0,0,0.0,5.0,1,0,1.1487179487179489,11_5.0
+11,7,9.374591364502958,inf,0,0,0.0,5.0,2,0,1.1487179487179489,11_5.0
+12,3,14.268684067203575,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,12_5.0
+12,4,13.089643664151836,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,12_5.0
+12,5,13.739823106373846,inf,0,0,0.0,5.0,0,0,1.1487179487179489,12_5.0
+12,6,13.669519287347907,inf,0,0,0.0,5.0,1,0,1.1487179487179489,12_5.0
+12,7,13.99578598067457,inf,0,0,0.0,5.0,2,0,1.1487179487179489,12_5.0
+13,3,12.989441559722076,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,13_5.0
+13,4,12.274600589012058,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,13_5.0
+13,5,12.427220506745886,inf,0,0,0.0,5.0,0,0,1.1487179487179489,13_5.0
+13,6,12.754093760275953,inf,0,0,0.0,5.0,1,0,1.1487179487179489,13_5.0
+13,7,13.718433817762524,inf,0,0,0.0,5.0,2,0,1.1487179487179489,13_5.0
+14,3,9.118405772601042,inf,0,0,0.0,5.0,-2,0,1.1487179487179489,14_5.0
+14,4,9.74251766740271,inf,0,0,0.0,5.0,-1,0,1.1487179487179489,14_5.0
+14,5,9.254403864326273,inf,0,0,0.0,5.0,0,0,1.1487179487179489,14_5.0
+14,6,9.330459677777796,inf,0,0,0.0,5.0,1,0,1.1487179487179489,14_5.0
+14,7,9.00828730479836,inf,0,0,0.0,5.0,2,0,1.1487179487179489,14_5.0
+18,3,8.750153667767103,5.0,0,1,0.0,5.0,-2,1,1.0,18_5.0
+18,4,10.369418029568877,5.0,0,1,0.0,5.0,-1,1,1.0,18_5.0
+18,5,11.814853522513967,5.0,1,1,2.0,5.0,0,1,1.0,18_5.0
+18,6,11.880991600710768,5.0,1,1,2.0,5.0,1,1,1.0,18_5.0
+18,7,12.084939640054936,5.0,1,1,2.0,5.0,2,1,1.0,18_5.0
+21,3,11.130313135937552,5.0,0,1,0.0,5.0,-2,1,1.0,21_5.0
+21,4,11.783557955755562,5.0,0,1,0.0,5.0,-1,1,1.0,21_5.0
+21,5,13.932178697330485,5.0,1,1,2.0,5.0,0,1,1.0,21_5.0
+21,6,13.311045276660398,5.0,1,1,2.0,5.0,1,1,1.0,21_5.0
+21,7,12.672260765381138,5.0,1,1,2.0,5.0,2,1,1.0,21_5.0
+22,3,10.370162504733846,5.0,0,1,0.0,5.0,-2,1,1.0,22_5.0
+22,4,10.130795877474743,5.0,0,1,0.0,5.0,-1,1,1.0,22_5.0
+22,5,13.327665186472212,5.0,1,1,2.0,5.0,0,1,1.0,22_5.0
+22,6,14.006133024294815,5.0,1,1,2.0,5.0,1,1,1.0,22_5.0
+22,7,11.997025447630525,5.0,1,1,2.0,5.0,2,1,1.0,22_5.0
+24,3,12.216528415765985,5.0,0,1,0.0,5.0,-2,1,1.0,24_5.0
+24,4,11.844123402747586,5.0,0,1,0.0,5.0,-1,1,1.0,24_5.0
+24,5,13.783296021523771,5.0,1,1,2.0,5.0,0,1,1.0,24_5.0
+24,6,13.374452489632073,5.0,1,1,2.0,5.0,1,1,1.0,24_5.0
+24,7,14.257003341404754,5.0,1,1,2.0,5.0,2,1,1.0,24_5.0
+25,3,12.376139740663152,5.0,0,1,0.0,5.0,-2,1,1.0,25_5.0
+25,4,12.451932231507831,5.0,0,1,0.0,5.0,-1,1,1.0,25_5.0
+25,5,14.40345513916028,5.0,1,1,2.0,5.0,0,1,1.0,25_5.0
+25,6,15.399316274400507,5.0,1,1,2.0,5.0,1,1,1.0,25_5.0
+25,7,14.121018820768782,5.0,1,1,2.0,5.0,2,1,1.0,25_5.0
+31,3,5.699685628062173,5.0,0,1,0.0,5.0,-2,1,1.0,31_5.0
+31,4,6.640244849690189,5.0,0,1,0.0,5.0,-1,1,1.0,31_5.0
+31,5,8.80287858433971,5.0,1,1,2.0,5.0,0,1,1.0,31_5.0
+31,6,7.478011213888342,5.0,1,1,2.0,5.0,1,1,1.0,31_5.0
+31,7,7.785449789009849,5.0,1,1,2.0,5.0,2,1,1.0,31_5.0
+32,3,13.145351442221973,5.0,0,1,0.0,5.0,-2,1,1.0,32_5.0
+32,4,14.213246370804844,5.0,0,1,0.0,5.0,-1,1,1.0,32_5.0
+32,5,15.673222444412447,5.0,1,1,2.0,5.0,0,1,1.0,32_5.0
+32,6,14.660677350520446,5.0,1,1,2.0,5.0,1,1,1.0,32_5.0
+32,7,14.475511437433273,5.0,1,1,2.0,5.0,2,1,1.0,32_5.0
+38,3,9.198240092902193,5.0,0,1,0.0,5.0,-2,1,1.0,38_5.0
+38,4,8.489100540648527,5.0,0,1,0.0,5.0,-1,1,1.0,38_5.0
+38,5,11.45417220010028,5.0,1,1,2.0,5.0,0,1,1.0,38_5.0
+38,6,10.634610329560985,5.0,1,1,2.0,5.0,1,1,1.0,38_5.0
+38,7,12.158628732093817,5.0,1,1,2.0,5.0,2,1,1.0,38_5.0
+42,3,9.165900951128542,5.0,0,1,0.0,5.0,-2,1,1.0,42_5.0
+42,4,9.836927314238459,5.0,0,1,0.0,5.0,-1,1,1.0,42_5.0
+42,5,11.893273621648287,5.0,1,1,2.0,5.0,0,1,1.0,42_5.0
+42,6,10.811622966738568,5.0,1,1,2.0,5.0,1,1,1.0,42_5.0
+42,7,12.387788948877349,5.0,1,1,2.0,5.0,2,1,1.0,42_5.0
+44,3,13.218744977031886,5.0,0,1,0.0,5.0,-2,1,1.0,44_5.0
+44,4,13.584858653200563,5.0,0,1,0.0,5.0,-1,1,1.0,44_5.0
+44,5,15.869766671523484,5.0,1,1,2.0,5.0,0,1,1.0,44_5.0
+44,6,15.606050081127213,5.0,1,1,2.0,5.0,1,1,1.0,44_5.0
+44,7,14.741953786805245,5.0,1,1,2.0,5.0,2,1,1.0,44_5.0
+47,3,11.405672205690639,5.0,0,1,0.0,5.0,-2,1,1.0,47_5.0
+47,4,10.327840224798175,5.0,0,1,0.0,5.0,-1,1,1.0,47_5.0
+47,5,12.972063268138358,5.0,1,1,2.0,5.0,0,1,1.0,47_5.0
+47,6,13.3375192672357,5.0,1,1,2.0,5.0,1,1,1.0,47_5.0
+47,7,13.949251964590784,5.0,1,1,2.0,5.0,2,1,1.0,47_5.0
+48,3,10.036462523941848,5.0,0,1,0.0,5.0,-2,1,1.0,48_5.0
+48,4,10.557346448715858,5.0,0,1,0.0,5.0,-1,1,1.0,48_5.0
+48,5,11.320126492133841,5.0,1,1,2.0,5.0,0,1,1.0,48_5.0
+48,6,12.376923426574146,5.0,1,1,2.0,5.0,1,1,1.0,48_5.0
+48,7,12.52530519550149,5.0,1,1,2.0,5.0,2,1,1.0,48_5.0
+49,3,9.93003674799368,5.0,0,1,0.0,5.0,-2,1,1.0,49_5.0
+49,4,9.728215912860774,5.0,0,1,0.0,5.0,-1,1,1.0,49_5.0
+49,5,12.103403154230948,5.0,1,1,2.0,5.0,0,1,1.0,49_5.0
+49,6,11.538525743370705,5.0,1,1,2.0,5.0,1,1,1.0,49_5.0
+49,7,12.200705227099014,5.0,1,1,2.0,5.0,2,1,1.0,49_5.0
diff --git a/benchmarks/data/stacked_did_test_panel.csv b/benchmarks/data/stacked_did_test_panel.csv
new file mode 100644
index 00000000..58dfdced
--- /dev/null
+++ b/benchmarks/data/stacked_did_test_panel.csv
@@ -0,0 +1,326 @@
+unit,period,outcome,first_treat,treated,treat,true_effect,_sub_exp,_event_time,_D_sa,_Q_weight,unit_subexp
+0,1,7.483662635760572,inf,0,0,0.0,3.0,-2,0,0.768,0_3.0
+0,2,6.841502772126143,inf,0,0,0.0,3.0,-1,0,0.768,0_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,3.0,0,0,0.768,0_3.0
+0,4,7.947042738617837,inf,0,0,0.0,3.0,1,0,0.768,0_3.0
+0,5,6.317944717889617,inf,0,0,0.0,3.0,2,0,0.768,0_3.0
+1,1,9.509053965877307,inf,0,0,0.0,3.0,-2,0,0.768,1_3.0
+1,2,10.614366425343084,inf,0,0,0.0,3.0,-1,0,0.768,1_3.0
+1,3,9.423988866091562,inf,0,0,0.0,3.0,0,0,0.768,1_3.0
+1,4,11.746966365956794,inf,0,0,0.0,3.0,1,0,0.768,1_3.0
+1,5,10.014392475247982,inf,0,0,0.0,3.0,2,0,0.768,1_3.0
+2,1,8.504571437173666,inf,0,0,0.0,3.0,-2,0,0.768,2_3.0
+2,2,7.968205325235655,inf,0,0,0.0,3.0,-1,0,0.768,2_3.0
+2,3,8.129454052191617,inf,0,0,0.0,3.0,0,0,0.768,2_3.0
+2,4,8.633961259377445,inf,0,0,0.0,3.0,1,0,0.768,2_3.0
+2,5,9.895662186433224,inf,0,0,0.0,3.0,2,0,0.768,2_3.0
+3,1,13.825385295202855,inf,0,0,0.0,3.0,-2,0,0.768,3_3.0
+3,2,13.829286373980507,inf,0,0,0.0,3.0,-1,0,0.768,3_3.0
+3,3,13.948413480279912,inf,0,0,0.0,3.0,0,0,0.768,3_3.0
+3,4,13.516792177472347,inf,0,0,0.0,3.0,1,0,0.768,3_3.0
+3,5,14.119361417606083,inf,0,0,0.0,3.0,2,0,0.768,3_3.0
+4,1,11.72507772055074,inf,0,0,0.0,3.0,-2,0,0.768,4_3.0
+4,2,11.912646534344725,inf,0,0,0.0,3.0,-1,0,0.768,4_3.0
+4,3,11.813278538150167,inf,0,0,0.0,3.0,0,0,0.768,4_3.0
+4,4,13.525689949379311,inf,0,0,0.0,3.0,1,0,0.768,4_3.0
+4,5,12.732831972997747,inf,0,0,0.0,3.0,2,0,0.768,4_3.0
+5,1,6.378890705428073,inf,0,0,0.0,3.0,-2,0,0.768,5_3.0
+5,2,6.440945780933881,inf,0,0,0.0,3.0,-1,0,0.768,5_3.0
+5,3,6.6127868751986805,inf,0,0,0.0,3.0,0,0,0.768,5_3.0
+5,4,6.513422311622866,inf,0,0,0.0,3.0,1,0,0.768,5_3.0
+5,5,6.75979943656341,inf,0,0,0.0,3.0,2,0,0.768,5_3.0
+6,1,8.633688139950424,inf,0,0,0.0,3.0,-2,0,0.768,6_3.0
+6,2,8.600953374134306,inf,0,0,0.0,3.0,-1,0,0.768,6_3.0
+6,3,8.411617732005928,inf,0,0,0.0,3.0,0,0,0.768,6_3.0
+6,4,8.387370342059572,inf,0,0,0.0,3.0,1,0,0.768,6_3.0
+6,5,8.945688592547397,inf,0,0,0.0,3.0,2,0,0.768,6_3.0
+7,1,11.334453872912409,inf,0,0,0.0,3.0,-2,0,0.768,7_3.0
+7,2,10.898891009750589,inf,0,0,0.0,3.0,-1,0,0.768,7_3.0
+7,3,10.456288009791654,inf,0,0,0.0,3.0,0,0,0.768,7_3.0
+7,4,11.086419785594533,inf,0,0,0.0,3.0,1,0,0.768,7_3.0
+7,5,11.136486986823087,inf,0,0,0.0,3.0,2,0,0.768,7_3.0
+8,1,12.363906817019398,inf,0,0,0.0,3.0,-2,0,0.768,8_3.0
+8,2,13.349921038000932,inf,0,0,0.0,3.0,-1,0,0.768,8_3.0
+8,3,11.289014907437469,inf,0,0,0.0,3.0,0,0,0.768,8_3.0
+8,4,12.413821661809754,inf,0,0,0.0,3.0,1,0,0.768,8_3.0
+8,5,12.599831686253742,inf,0,0,0.0,3.0,2,0,0.768,8_3.0
+9,1,8.226753444350999,inf,0,0,0.0,3.0,-2,0,0.768,9_3.0
+9,2,8.04357609368333,inf,0,0,0.0,3.0,-1,0,0.768,9_3.0
+9,3,7.935929232288304,inf,0,0,0.0,3.0,0,0,0.768,9_3.0
+9,4,8.947274202622067,inf,0,0,0.0,3.0,1,0,0.768,9_3.0
+9,5,8.525815827970705,inf,0,0,0.0,3.0,2,0,0.768,9_3.0
+10,1,6.616529053324495,inf,0,0,0.0,3.0,-2,0,0.768,10_3.0
+10,2,7.904056043755299,inf,0,0,0.0,3.0,-1,0,0.768,10_3.0
+10,3,7.385638433127546,inf,0,0,0.0,3.0,0,0,0.768,10_3.0
+10,4,7.6853500749042345,inf,0,0,0.0,3.0,1,0,0.768,10_3.0
+10,5,7.841052147984791,inf,0,0,0.0,3.0,2,0,0.768,10_3.0
+11,1,8.904134028447222,inf,0,0,0.0,3.0,-2,0,0.768,11_3.0
+11,2,9.722452361429282,inf,0,0,0.0,3.0,-1,0,0.768,11_3.0
+11,3,8.983212026944074,inf,0,0,0.0,3.0,0,0,0.768,11_3.0
+11,4,9.774571594385389,inf,0,0,0.0,3.0,1,0,0.768,11_3.0
+11,5,9.962850837041142,inf,0,0,0.0,3.0,2,0,0.768,11_3.0
+12,1,14.247804715186929,inf,0,0,0.0,3.0,-2,0,0.768,12_3.0
+12,2,13.59781501202838,inf,0,0,0.0,3.0,-1,0,0.768,12_3.0
+12,3,14.268684067203575,inf,0,0,0.0,3.0,0,0,0.768,12_3.0
+12,4,13.089643664151836,inf,0,0,0.0,3.0,1,0,0.768,12_3.0
+12,5,13.739823106373846,inf,0,0,0.0,3.0,2,0,0.768,12_3.0
+13,1,12.466946753384763,inf,0,0,0.0,3.0,-2,0,0.768,13_3.0
+13,2,12.577723255899597,inf,0,0,0.0,3.0,-1,0,0.768,13_3.0
+13,3,12.989441559722076,inf,0,0,0.0,3.0,0,0,0.768,13_3.0
+13,4,12.274600589012058,inf,0,0,0.0,3.0,1,0,0.768,13_3.0
+13,5,12.427220506745886,inf,0,0,0.0,3.0,2,0,0.768,13_3.0
+14,1,9.321384661956385,inf,0,0,0.0,3.0,-2,0,0.768,14_3.0
+14,2,8.537850792031616,inf,0,0,0.0,3.0,-1,0,0.768,14_3.0
+14,3,9.118405772601042,inf,0,0,0.0,3.0,0,0,0.768,14_3.0
+14,4,9.74251766740271,inf,0,0,0.0,3.0,1,0,0.768,14_3.0
+14,5,9.254403864326273,inf,0,0,0.0,3.0,2,0,0.768,14_3.0
+15,1,7.441009679583563,7.0,0,1,0.0,3.0,-2,0,0.768,15_3.0
+15,2,8.485304881474772,7.0,0,1,0.0,3.0,-1,0,0.768,15_3.0
+15,3,8.564340753151164,7.0,0,1,0.0,3.0,0,0,0.768,15_3.0
+15,4,8.1815349724801,7.0,0,1,0.0,3.0,1,0,0.768,15_3.0
+15,5,9.311637156674976,7.0,0,1,0.0,3.0,2,0,0.768,15_3.0
+16,1,7.6836207307333035,3.0,0,1,0.0,3.0,-2,1,1.0,16_3.0
+16,2,8.357654569597338,3.0,0,1,0.0,3.0,-1,1,1.0,16_3.0
+16,3,10.056550820184588,3.0,1,1,2.0,3.0,0,1,1.0,16_3.0
+16,4,10.207378999928212,3.0,1,1,2.0,3.0,1,1,1.0,16_3.0
+16,5,10.096882823598278,3.0,1,1,2.0,3.0,2,1,1.0,16_3.0
+17,1,8.175606640239206,3.0,0,1,0.0,3.0,-2,1,1.0,17_3.0
+17,2,8.600350852687033,3.0,0,1,0.0,3.0,-1,1,1.0,17_3.0
+17,3,9.60085063995874,3.0,1,1,2.0,3.0,0,1,1.0,17_3.0
+17,4,11.265857113284081,3.0,1,1,2.0,3.0,1,1,1.0,17_3.0
+17,5,10.805731826455423,3.0,1,1,2.0,3.0,2,1,1.0,17_3.0
+19,1,8.287479893276272,3.0,0,1,0.0,3.0,-2,1,1.0,19_3.0
+19,2,8.838449778710965,3.0,0,1,0.0,3.0,-1,1,1.0,19_3.0
+19,3,10.301088160788524,3.0,1,1,2.0,3.0,0,1,1.0,19_3.0
+19,4,10.193467173412117,3.0,1,1,2.0,3.0,1,1,1.0,19_3.0
+19,5,10.93570451644154,3.0,1,1,2.0,3.0,2,1,1.0,19_3.0
+20,1,6.608825849966525,3.0,0,1,0.0,3.0,-2,1,1.0,20_3.0
+20,2,6.192567902449463,3.0,0,1,0.0,3.0,-1,1,1.0,20_3.0
+20,3,8.62011884779571,3.0,1,1,2.0,3.0,0,1,1.0,20_3.0
+20,4,9.534699361996061,3.0,1,1,2.0,3.0,1,1,1.0,20_3.0
+20,5,8.972714344039472,3.0,1,1,2.0,3.0,2,1,1.0,20_3.0
+23,1,11.395156619865356,7.0,0,1,0.0,3.0,-2,0,0.768,23_3.0
+23,2,12.292733093551572,7.0,0,1,0.0,3.0,-1,0,0.768,23_3.0
+23,3,10.987337838152467,7.0,0,1,0.0,3.0,0,0,0.768,23_3.0
+23,4,12.858675160893522,7.0,0,1,0.0,3.0,1,0,0.768,23_3.0
+23,5,11.476064698198886,7.0,0,1,0.0,3.0,2,0,0.768,23_3.0
+26,1,10.668383782601722,3.0,0,1,0.0,3.0,-2,1,1.0,26_3.0
+26,2,10.446234601225571,3.0,0,1,0.0,3.0,-1,1,1.0,26_3.0
+26,3,13.556242815369252,3.0,1,1,2.0,3.0,0,1,1.0,26_3.0
+26,4,13.576729933292084,3.0,1,1,2.0,3.0,1,1,1.0,26_3.0
+26,5,13.922175150825803,3.0,1,1,2.0,3.0,2,1,1.0,26_3.0
+27,1,8.580588297539316,7.0,0,1,0.0,3.0,-2,0,0.768,27_3.0
+27,2,9.053926240501355,7.0,0,1,0.0,3.0,-1,0,0.768,27_3.0
+27,3,8.688974179281848,7.0,0,1,0.0,3.0,0,0,0.768,27_3.0
+27,4,7.928575988525038,7.0,0,1,0.0,3.0,1,0,0.768,27_3.0
+27,5,9.170520905605873,7.0,0,1,0.0,3.0,2,0,0.768,27_3.0
+28,1,9.371848258038815,3.0,0,1,0.0,3.0,-2,1,1.0,28_3.0
+28,2,9.154422149577181,3.0,0,1,0.0,3.0,-1,1,1.0,28_3.0
+28,3,11.965196547208972,3.0,1,1,2.0,3.0,0,1,1.0,28_3.0
+28,4,11.302065396349953,3.0,1,1,2.0,3.0,1,1,1.0,28_3.0
+28,5,11.589704550706,3.0,1,1,2.0,3.0,2,1,1.0,28_3.0
+29,1,8.985010266355241,3.0,0,1,0.0,3.0,-2,1,1.0,29_3.0
+29,2,8.92226436961385,3.0,0,1,0.0,3.0,-1,1,1.0,29_3.0
+29,3,11.587183906367892,3.0,1,1,2.0,3.0,0,1,1.0,29_3.0
+29,4,11.061585148973101,3.0,1,1,2.0,3.0,1,1,1.0,29_3.0
+29,5,11.32705363934025,3.0,1,1,2.0,3.0,2,1,1.0,29_3.0
+30,1,8.376287388505098,3.0,0,1,0.0,3.0,-2,1,1.0,30_3.0
+30,2,8.657693030574496,3.0,0,1,0.0,3.0,-1,1,1.0,30_3.0
+30,3,10.18781936742593,3.0,1,1,2.0,3.0,0,1,1.0,30_3.0
+30,4,11.000739934625502,3.0,1,1,2.0,3.0,1,1,1.0,30_3.0
+30,5,10.282065104573295,3.0,1,1,2.0,3.0,2,1,1.0,30_3.0
+33,1,11.57454492762582,7.0,0,1,0.0,3.0,-2,0,0.768,33_3.0
+33,2,11.691765125046665,7.0,0,1,0.0,3.0,-1,0,0.768,33_3.0
+33,3,11.84979372696007,7.0,0,1,0.0,3.0,0,0,0.768,33_3.0
+33,4,12.62424994943963,7.0,0,1,0.0,3.0,1,0,0.768,33_3.0
+33,5,12.3254386483586,7.0,0,1,0.0,3.0,2,0,0.768,33_3.0
+34,1,12.563769947545987,7.0,0,1,0.0,3.0,-2,0,0.768,34_3.0
+34,2,11.966597306919729,7.0,0,1,0.0,3.0,-1,0,0.768,34_3.0
+34,3,12.803090421682993,7.0,0,1,0.0,3.0,0,0,0.768,34_3.0
+34,4,13.544985407250604,7.0,0,1,0.0,3.0,1,0,0.768,34_3.0
+34,5,13.17341536766152,7.0,0,1,0.0,3.0,2,0,0.768,34_3.0
+35,1,13.198034881842501,3.0,0,1,0.0,3.0,-2,1,1.0,35_3.0
+35,2,13.13803314498448,3.0,0,1,0.0,3.0,-1,1,1.0,35_3.0
+35,3,14.775491475016876,3.0,1,1,2.0,3.0,0,1,1.0,35_3.0
+35,4,14.755631298179171,3.0,1,1,2.0,3.0,1,1,1.0,35_3.0
+35,5,15.82726562084955,3.0,1,1,2.0,3.0,2,1,1.0,35_3.0
+36,1,7.2152253987939226,3.0,0,1,0.0,3.0,-2,1,1.0,36_3.0
+36,2,6.610853150968622,3.0,0,1,0.0,3.0,-1,1,1.0,36_3.0
+36,3,8.853563997233643,3.0,1,1,2.0,3.0,0,1,1.0,36_3.0
+36,4,9.09329932509169,3.0,1,1,2.0,3.0,1,1,1.0,36_3.0
+36,5,9.298377296285087,3.0,1,1,2.0,3.0,2,1,1.0,36_3.0
+37,1,10.146958297869952,7.0,0,1,0.0,3.0,-2,0,0.768,37_3.0
+37,2,9.060497069998855,7.0,0,1,0.0,3.0,-1,0,0.768,37_3.0
+37,3,9.318388049432945,7.0,0,1,0.0,3.0,0,0,0.768,37_3.0
+37,4,9.26097646005531,7.0,0,1,0.0,3.0,1,0,0.768,37_3.0
+37,5,8.919210921226544,7.0,0,1,0.0,3.0,2,0,0.768,37_3.0
+39,1,12.353298527293795,3.0,0,1,0.0,3.0,-2,1,1.0,39_3.0
+39,2,12.242756403937744,3.0,0,1,0.0,3.0,-1,1,1.0,39_3.0
+39,3,15.149617439612653,3.0,1,1,2.0,3.0,0,1,1.0,39_3.0
+39,4,14.088452790656742,3.0,1,1,2.0,3.0,1,1,1.0,39_3.0
+39,5,14.060806514376624,3.0,1,1,2.0,3.0,2,1,1.0,39_3.0
+40,1,7.342148190809627,7.0,0,1,0.0,3.0,-2,0,0.768,40_3.0
+40,2,8.356636279545437,7.0,0,1,0.0,3.0,-1,0,0.768,40_3.0
+40,3,8.853341199215448,7.0,0,1,0.0,3.0,0,0,0.768,40_3.0
+40,4,7.646273376822638,7.0,0,1,0.0,3.0,1,0,0.768,40_3.0
+40,5,7.959815635944495,7.0,0,1,0.0,3.0,2,0,0.768,40_3.0
+41,1,12.529859573199765,3.0,0,1,0.0,3.0,-2,1,1.0,41_3.0
+41,2,12.359862327664574,3.0,0,1,0.0,3.0,-1,1,1.0,41_3.0
+41,3,14.594595526225248,3.0,1,1,2.0,3.0,0,1,1.0,41_3.0
+41,4,14.627150386514797,3.0,1,1,2.0,3.0,1,1,1.0,41_3.0
+41,5,15.438306775111634,3.0,1,1,2.0,3.0,2,1,1.0,41_3.0
+43,1,11.589875560059742,7.0,0,1,0.0,3.0,-2,0,0.768,43_3.0
+43,2,11.566062741673996,7.0,0,1,0.0,3.0,-1,0,0.768,43_3.0
+43,3,11.081605645195344,7.0,0,1,0.0,3.0,0,0,0.768,43_3.0
+43,4,11.740097124129901,7.0,0,1,0.0,3.0,1,0,0.768,43_3.0
+43,5,12.466392539854581,7.0,0,1,0.0,3.0,2,0,0.768,43_3.0
+45,1,8.77119845157113,7.0,0,1,0.0,3.0,-2,0,0.768,45_3.0
+45,2,8.039535155953265,7.0,0,1,0.0,3.0,-1,0,0.768,45_3.0
+45,3,8.077961859940832,7.0,0,1,0.0,3.0,0,0,0.768,45_3.0
+45,4,8.881163504521098,7.0,0,1,0.0,3.0,1,0,0.768,45_3.0
+45,5,8.557759225694356,7.0,0,1,0.0,3.0,2,0,0.768,45_3.0
+46,1,12.148563177783899,7.0,0,1,0.0,3.0,-2,0,0.768,46_3.0
+46,2,12.307754494386236,7.0,0,1,0.0,3.0,-1,0,0.768,46_3.0
+46,3,12.393620432500153,7.0,0,1,0.0,3.0,0,0,0.768,46_3.0
+46,4,13.035568610439492,7.0,0,1,0.0,3.0,1,0,0.768,46_3.0
+46,5,12.995360934519939,7.0,0,1,0.0,3.0,2,0,0.768,46_3.0
+0,3,7.9831226286999195,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,0_5.0
+0,4,7.947042738617837,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,0_5.0
+0,5,6.317944717889617,inf,0,0,0.0,5.0,0,0,1.3866666666666667,0_5.0
+0,6,8.030151283551504,inf,0,0,0.0,5.0,1,0,1.3866666666666667,0_5.0
+0,7,7.857022192736064,inf,0,0,0.0,5.0,2,0,1.3866666666666667,0_5.0
+1,3,9.423988866091562,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,1_5.0
+1,4,11.746966365956794,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,1_5.0
+1,5,10.014392475247982,inf,0,0,0.0,5.0,0,0,1.3866666666666667,1_5.0
+1,6,10.364571813434443,inf,0,0,0.0,5.0,1,0,1.3866666666666667,1_5.0
+1,7,10.370697302329921,inf,0,0,0.0,5.0,2,0,1.3866666666666667,1_5.0
+2,3,8.129454052191617,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,2_5.0
+2,4,8.633961259377445,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,2_5.0
+2,5,9.895662186433224,inf,0,0,0.0,5.0,0,0,1.3866666666666667,2_5.0
+2,6,8.55134266155172,inf,0,0,0.0,5.0,1,0,1.3866666666666667,2_5.0
+2,7,8.517976639648715,inf,0,0,0.0,5.0,2,0,1.3866666666666667,2_5.0
+3,3,13.948413480279912,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,3_5.0
+3,4,13.516792177472347,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,3_5.0
+3,5,14.119361417606083,inf,0,0,0.0,5.0,0,0,1.3866666666666667,3_5.0
+3,6,13.338205224874367,inf,0,0,0.0,5.0,1,0,1.3866666666666667,3_5.0
+3,7,14.4234803289339,inf,0,0,0.0,5.0,2,0,1.3866666666666667,3_5.0
+4,3,11.813278538150167,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,4_5.0
+4,4,13.525689949379311,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,4_5.0
+4,5,12.732831972997747,inf,0,0,0.0,5.0,0,0,1.3866666666666667,4_5.0
+4,6,12.618074611389941,inf,0,0,0.0,5.0,1,0,1.3866666666666667,4_5.0
+4,7,11.481627079417956,inf,0,0,0.0,5.0,2,0,1.3866666666666667,4_5.0
+5,3,6.6127868751986805,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,5_5.0
+5,4,6.513422311622866,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,5_5.0
+5,5,6.75979943656341,inf,0,0,0.0,5.0,0,0,1.3866666666666667,5_5.0
+5,6,6.600117721454566,inf,0,0,0.0,5.0,1,0,1.3866666666666667,5_5.0
+5,7,7.410185509058965,inf,0,0,0.0,5.0,2,0,1.3866666666666667,5_5.0
+6,3,8.411617732005928,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,6_5.0
+6,4,8.387370342059572,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,6_5.0
+6,5,8.945688592547397,inf,0,0,0.0,5.0,0,0,1.3866666666666667,6_5.0
+6,6,9.424533467045443,inf,0,0,0.0,5.0,1,0,1.3866666666666667,6_5.0
+6,7,8.558459954832495,inf,0,0,0.0,5.0,2,0,1.3866666666666667,6_5.0
+7,3,10.456288009791654,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,7_5.0
+7,4,11.086419785594533,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,7_5.0
+7,5,11.136486986823087,inf,0,0,0.0,5.0,0,0,1.3866666666666667,7_5.0
+7,6,10.49707772184382,inf,0,0,0.0,5.0,1,0,1.3866666666666667,7_5.0
+7,7,11.220866186095618,inf,0,0,0.0,5.0,2,0,1.3866666666666667,7_5.0
+8,3,11.289014907437469,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,8_5.0
+8,4,12.413821661809754,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,8_5.0
+8,5,12.599831686253742,inf,0,0,0.0,5.0,0,0,1.3866666666666667,8_5.0
+8,6,12.770672804509399,inf,0,0,0.0,5.0,1,0,1.3866666666666667,8_5.0
+8,7,12.12649797254452,inf,0,0,0.0,5.0,2,0,1.3866666666666667,8_5.0
+9,3,7.935929232288304,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,9_5.0
+9,4,8.947274202622067,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,9_5.0
+9,5,8.525815827970705,inf,0,0,0.0,5.0,0,0,1.3866666666666667,9_5.0
+9,6,8.29259367272137,inf,0,0,0.0,5.0,1,0,1.3866666666666667,9_5.0
+9,7,7.755986984180787,inf,0,0,0.0,5.0,2,0,1.3866666666666667,9_5.0
+10,3,7.385638433127546,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,10_5.0
+10,4,7.6853500749042345,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,10_5.0
+10,5,7.841052147984791,inf,0,0,0.0,5.0,0,0,1.3866666666666667,10_5.0
+10,6,7.779603182233025,inf,0,0,0.0,5.0,1,0,1.3866666666666667,10_5.0
+10,7,9.262810775890333,inf,0,0,0.0,5.0,2,0,1.3866666666666667,10_5.0
+11,3,8.983212026944074,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,11_5.0
+11,4,9.774571594385389,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,11_5.0
+11,5,9.962850837041142,inf,0,0,0.0,5.0,0,0,1.3866666666666667,11_5.0
+11,6,9.456066431566448,inf,0,0,0.0,5.0,1,0,1.3866666666666667,11_5.0
+11,7,9.374591364502958,inf,0,0,0.0,5.0,2,0,1.3866666666666667,11_5.0
+12,3,14.268684067203575,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,12_5.0
+12,4,13.089643664151836,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,12_5.0
+12,5,13.739823106373846,inf,0,0,0.0,5.0,0,0,1.3866666666666667,12_5.0
+12,6,13.669519287347907,inf,0,0,0.0,5.0,1,0,1.3866666666666667,12_5.0
+12,7,13.99578598067457,inf,0,0,0.0,5.0,2,0,1.3866666666666667,12_5.0
+13,3,12.989441559722076,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,13_5.0
+13,4,12.274600589012058,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,13_5.0
+13,5,12.427220506745886,inf,0,0,0.0,5.0,0,0,1.3866666666666667,13_5.0
+13,6,12.754093760275953,inf,0,0,0.0,5.0,1,0,1.3866666666666667,13_5.0
+13,7,13.718433817762524,inf,0,0,0.0,5.0,2,0,1.3866666666666667,13_5.0
+14,3,9.118405772601042,inf,0,0,0.0,5.0,-2,0,1.3866666666666667,14_5.0
+14,4,9.74251766740271,inf,0,0,0.0,5.0,-1,0,1.3866666666666667,14_5.0
+14,5,9.254403864326273,inf,0,0,0.0,5.0,0,0,1.3866666666666667,14_5.0
+14,6,9.330459677777796,inf,0,0,0.0,5.0,1,0,1.3866666666666667,14_5.0
+14,7,9.00828730479836,inf,0,0,0.0,5.0,2,0,1.3866666666666667,14_5.0
+18,3,8.750153667767103,5.0,0,1,0.0,5.0,-2,1,1.0,18_5.0
+18,4,10.369418029568877,5.0,0,1,0.0,5.0,-1,1,1.0,18_5.0
+18,5,11.814853522513967,5.0,1,1,2.0,5.0,0,1,1.0,18_5.0
+18,6,11.880991600710768,5.0,1,1,2.0,5.0,1,1,1.0,18_5.0
+18,7,12.084939640054936,5.0,1,1,2.0,5.0,2,1,1.0,18_5.0
+21,3,11.130313135937552,5.0,0,1,0.0,5.0,-2,1,1.0,21_5.0
+21,4,11.783557955755562,5.0,0,1,0.0,5.0,-1,1,1.0,21_5.0
+21,5,13.932178697330485,5.0,1,1,2.0,5.0,0,1,1.0,21_5.0
+21,6,13.311045276660398,5.0,1,1,2.0,5.0,1,1,1.0,21_5.0
+21,7,12.672260765381138,5.0,1,1,2.0,5.0,2,1,1.0,21_5.0
+22,3,10.370162504733846,5.0,0,1,0.0,5.0,-2,1,1.0,22_5.0
+22,4,10.130795877474743,5.0,0,1,0.0,5.0,-1,1,1.0,22_5.0
+22,5,13.327665186472212,5.0,1,1,2.0,5.0,0,1,1.0,22_5.0
+22,6,14.006133024294815,5.0,1,1,2.0,5.0,1,1,1.0,22_5.0
+22,7,11.997025447630525,5.0,1,1,2.0,5.0,2,1,1.0,22_5.0
+24,3,12.216528415765985,5.0,0,1,0.0,5.0,-2,1,1.0,24_5.0
+24,4,11.844123402747586,5.0,0,1,0.0,5.0,-1,1,1.0,24_5.0
+24,5,13.783296021523771,5.0,1,1,2.0,5.0,0,1,1.0,24_5.0
+24,6,13.374452489632073,5.0,1,1,2.0,5.0,1,1,1.0,24_5.0
+24,7,14.257003341404754,5.0,1,1,2.0,5.0,2,1,1.0,24_5.0
+25,3,12.376139740663152,5.0,0,1,0.0,5.0,-2,1,1.0,25_5.0
+25,4,12.451932231507831,5.0,0,1,0.0,5.0,-1,1,1.0,25_5.0
+25,5,14.40345513916028,5.0,1,1,2.0,5.0,0,1,1.0,25_5.0
+25,6,15.399316274400507,5.0,1,1,2.0,5.0,1,1,1.0,25_5.0
+25,7,14.121018820768782,5.0,1,1,2.0,5.0,2,1,1.0,25_5.0
+31,3,5.699685628062173,5.0,0,1,0.0,5.0,-2,1,1.0,31_5.0
+31,4,6.640244849690189,5.0,0,1,0.0,5.0,-1,1,1.0,31_5.0
+31,5,8.80287858433971,5.0,1,1,2.0,5.0,0,1,1.0,31_5.0
+31,6,7.478011213888342,5.0,1,1,2.0,5.0,1,1,1.0,31_5.0
+31,7,7.785449789009849,5.0,1,1,2.0,5.0,2,1,1.0,31_5.0
+32,3,13.145351442221973,5.0,0,1,0.0,5.0,-2,1,1.0,32_5.0
+32,4,14.213246370804844,5.0,0,1,0.0,5.0,-1,1,1.0,32_5.0
+32,5,15.673222444412447,5.0,1,1,2.0,5.0,0,1,1.0,32_5.0
+32,6,14.660677350520446,5.0,1,1,2.0,5.0,1,1,1.0,32_5.0
+32,7,14.475511437433273,5.0,1,1,2.0,5.0,2,1,1.0,32_5.0
+38,3,9.198240092902193,5.0,0,1,0.0,5.0,-2,1,1.0,38_5.0
+38,4,8.489100540648527,5.0,0,1,0.0,5.0,-1,1,1.0,38_5.0
+38,5,11.45417220010028,5.0,1,1,2.0,5.0,0,1,1.0,38_5.0
+38,6,10.634610329560985,5.0,1,1,2.0,5.0,1,1,1.0,38_5.0
+38,7,12.158628732093817,5.0,1,1,2.0,5.0,2,1,1.0,38_5.0
+42,3,9.165900951128542,5.0,0,1,0.0,5.0,-2,1,1.0,42_5.0
+42,4,9.836927314238459,5.0,0,1,0.0,5.0,-1,1,1.0,42_5.0
+42,5,11.893273621648287,5.0,1,1,2.0,5.0,0,1,1.0,42_5.0
+42,6,10.811622966738568,5.0,1,1,2.0,5.0,1,1,1.0,42_5.0
+42,7,12.387788948877349,5.0,1,1,2.0,5.0,2,1,1.0,42_5.0
+44,3,13.218744977031886,5.0,0,1,0.0,5.0,-2,1,1.0,44_5.0
+44,4,13.584858653200563,5.0,0,1,0.0,5.0,-1,1,1.0,44_5.0
+44,5,15.869766671523484,5.0,1,1,2.0,5.0,0,1,1.0,44_5.0
+44,6,15.606050081127213,5.0,1,1,2.0,5.0,1,1,1.0,44_5.0
+44,7,14.741953786805245,5.0,1,1,2.0,5.0,2,1,1.0,44_5.0
+47,3,11.405672205690639,5.0,0,1,0.0,5.0,-2,1,1.0,47_5.0
+47,4,10.327840224798175,5.0,0,1,0.0,5.0,-1,1,1.0,47_5.0
+47,5,12.972063268138358,5.0,1,1,2.0,5.0,0,1,1.0,47_5.0
+47,6,13.3375192672357,5.0,1,1,2.0,5.0,1,1,1.0,47_5.0
+47,7,13.949251964590784,5.0,1,1,2.0,5.0,2,1,1.0,47_5.0
+48,3,10.036462523941848,5.0,0,1,0.0,5.0,-2,1,1.0,48_5.0
+48,4,10.557346448715858,5.0,0,1,0.0,5.0,-1,1,1.0,48_5.0
+48,5,11.320126492133841,5.0,1,1,2.0,5.0,0,1,1.0,48_5.0
+48,6,12.376923426574146,5.0,1,1,2.0,5.0,1,1,1.0,48_5.0
+48,7,12.52530519550149,5.0,1,1,2.0,5.0,2,1,1.0,48_5.0
+49,3,9.93003674799368,5.0,0,1,0.0,5.0,-2,1,1.0,49_5.0
+49,4,9.728215912860774,5.0,0,1,0.0,5.0,-1,1,1.0,49_5.0
+49,5,12.103403154230948,5.0,1,1,2.0,5.0,0,1,1.0,49_5.0
+49,6,11.538525743370705,5.0,1,1,2.0,5.0,1,1,1.0,49_5.0
+49,7,12.200705227099014,5.0,1,1,2.0,5.0,2,1,1.0,49_5.0
diff --git a/diff_diff/guides/llms-full.txt b/diff_diff/guides/llms-full.txt
index 8bfbc55a..4d685444 100644
--- a/diff_diff/guides/llms-full.txt
+++ b/diff_diff/guides/llms-full.txt
@@ -743,6 +743,7 @@ StackedDiD(
     alpha: float = 0.05,
     anticipation: int = 0,
     rank_deficient_action: str = "warn",
+    vcov_type: str = "hc1",                   # {"hc1","hc2_bm"}; classical/hc2 rejected (intrinsically clustered), conley deferred. survey_design=... requires hc1
 )
 ```
 
diff --git a/diff_diff/stacked_did.py b/diff_diff/stacked_did.py
index 69e4ffd2..741c07d5 100644
--- a/diff_diff/stacked_did.py
+++ b/diff_diff/stacked_did.py
@@ -78,6 +78,41 @@ class StackedDiD:
         - "warn": Issue warning and drop linearly dependent columns
         - "error": Raise ValueError
         - "silent": Drop columns silently
+    vcov_type : {"classical","hc1","hc2","hc2_bm"}, default="hc1"
+        Analytical variance family for the stacked WLS regression. StackedDiD
+        is intrinsically clustered (``cluster`` is required, no ``cluster=None``
+        opt-out), so one-way families that don't compose with cluster_ids are
+        rejected at ``__init__``:
+
+        - ``"hc1"`` (default): CR1 Liang-Zeger cluster-robust on the Q-weighted
+          design via ``solve_ols(weights=composed_weights, vcov_type="hc1")``.
+          Bit-equal to the prior bake-Q-into-X output up to float64 multiplication
+          ordering at machine precision (HC1 WLS sandwich is algebraically
+          invariant between the two forms). Matches
+          ``clubSandwich::vcovCR(lm(weights=Q,...), cluster=~unit, type="CR1S")``
+          at atol=1e-10 (target is ``CR1S`` — Stata-style ``G/(G-1) * (n-1)/(n-p)``
+          finite-sample correction — NOT plain ``CR1`` which omits the
+          ``(n-1)/(n-p)`` factor and would diverge by ~1.4%).
+        - ``"hc2_bm"``: CR2 Bell-McCaffrey via
+          ``solve_ols(weights=composed_weights, vcov_type="hc2_bm")``, routed
+          through the clubSandwich WLS-CR2 port (matches
+          ``clubSandwich::vcovCR(lm(weights=Q,...), cluster=~unit, type="CR2")
+          + coef_test()$df_Satt`` at atol=1e-10). See ``REGISTRY.md`` Phase 1a
+          ``hc2_bm + weights`` row for the algebra (W not √W in hat matrix,
+          W² in bias term, unweighted residuals in score).
+        - ``"classical"`` and ``"hc2"`` are REJECTED at ``__init__`` with a
+          cluster-incompatibility ``ValueError``: StackedDiD requires a cluster
+          structure, so one-way families don't compose with the linalg validator.
+          Use ``"hc1"`` or ``"hc2_bm"``.
+        - ``"conley"`` is REJECTED at ``__init__`` (deferred; would require
+          threading the six ``conley_*`` params through ``solve_ols`` — tracked
+          in TODO.md).
+
+        Survey-design precedence: when ``survey_design=`` is supplied to
+        ``fit()`` with ``vcov_type != "hc1"``, a ``NotImplementedError`` is
+        raised — the survey Taylor-series linearization (or replicate-weight
+        refit) variance overrides the analytical sandwich. Use the default
+        ``vcov_type="hc1"`` for survey designs.
 
     Attributes
     ----------
@@ -128,6 +163,7 @@ def __init__(
         alpha: float = 0.05,
         anticipation: int = 0,
         rank_deficient_action: str = "warn",
+        vcov_type: str = "hc1",
     ):
         if weighting not in ("aggregate", "population", "sample_share"):
             raise ValueError(
@@ -146,6 +182,9 @@ def __init__(
                 f"rank_deficient_action must be 'warn', 'error', or 'silent', "
                 f"got '{rank_deficient_action}'"
             )
+        # vcov_type validation (Phase 1b 2/8: thread through StackedDiD).
+        # Factored into _validate_vcov_type so set_params() can re-validate.
+        self._validate_vcov_type(vcov_type)
 
         self.kappa_pre = kappa_pre
         self.kappa_post = kappa_post
@@ -155,10 +194,40 @@ def __init__(
         self.alpha = alpha
         self.anticipation = anticipation
         self.rank_deficient_action = rank_deficient_action
+        self.vcov_type = vcov_type
 
         self.is_fitted_ = False
         self.results_: Optional[StackedDiDResults] = None
 
+    @staticmethod
+    def _validate_vcov_type(vcov_type: str) -> None:
+        """Validate vcov_type. Called from __init__ AND set_params so that
+        sklearn-style mutation (`est.set_params(vcov_type="bad")`) hits the
+        estimator-level guard rather than failing later in the linalg layer
+        with a different message."""
+        if vcov_type == "conley":
+            raise ValueError(
+                "vcov_type='conley' is not yet supported on StackedDiD. "
+                "Threading conley_coords / conley_cutoff_km / conley_metric / "
+                "conley_kernel / conley_time / conley_unit / conley_lag_cutoff "
+                "through solve_ols requires a follow-up PR (tracked in TODO.md "
+                "alongside the SunAbraham conley follow-up). Use vcov_type='hc1' "
+                "(default, CR1) or 'hc2_bm' (CR2 Bell-McCaffrey)."
+            )
+        if vcov_type not in ("classical", "hc1", "hc2", "hc2_bm"):
+            raise ValueError(
+                f"vcov_type must be one of {{'classical', 'hc1', 'hc2', 'hc2_bm'}}, "
+                f"got '{vcov_type}'"
+            )
+        if vcov_type in ("classical", "hc2"):
+            raise ValueError(
+                "StackedDiD clusters intrinsically at 'unit' or 'unit_subexp' "
+                "(no cluster=None opt-out). One-way vcov_type='classical'/'hc2' "
+                "is rejected by the linalg validator when combined with "
+                "cluster_ids. Use vcov_type='hc1' (CR1 Liang-Zeger) or "
+                "'hc2_bm' (CR2 Bell-McCaffrey)."
+            )
+
     def fit(
         self,
         data: pd.DataFrame,
@@ -242,9 +311,7 @@ def fit(
         resolved_survey, survey_weights, survey_weight_type, survey_metadata = (
             _resolve_survey_for_fit(survey_design, data, "analytical")
         )
-        _uses_replicate_sd = (
-            resolved_survey is not None and resolved_survey.uses_replicate_variance
-        )
+        _uses_replicate_sd = resolved_survey is not None and resolved_survey.uses_replicate_variance
 
         # Reject fweight and aweight — Q-weight composition is ratio-valued
         # and breaks both frequency-weight (integer) and analytic-weight
@@ -260,6 +327,24 @@ def fit(
                 "Use weight_type='pweight' (default) instead."
             )
 
+        # Survey-design precedence: when survey_design is supplied, the survey
+        # Taylor-series linearization (or replicate-weight refit) variance
+        # overrides the analytical sandwich. The non-hc1 analytical families
+        # are blocked. Reject ordering matters here: the fweight/aweight check
+        # above fires FIRST so users hit the Q-weight semantics error before
+        # the vcov error (two-step educational path, matches SA precedent).
+        # Future refactors must not swap the order without re-validating tests
+        # `test_aweight_plus_hc2_bm_rejected_by_stacked_did_level_guard`.
+        if resolved_survey is not None and self.vcov_type != "hc1":
+            raise NotImplementedError(
+                f"StackedDiD(vcov_type='{self.vcov_type}') is not supported with "
+                "survey_design=. The survey TSL (or replicate-weight refit) "
+                "variance overrides the analytical sandwich family, so the "
+                "small-sample CR2 Bell-McCaffrey correction cannot compose "
+                "with the survey variance machinery. Use vcov_type='hc1' "
+                "(default) for survey designs."
+            )
+
         # Collect survey design column names for propagation through sub-experiments
         survey_cols: List[str] = []
         if survey_design is not None and isinstance(survey_design, SurveyDesign):
@@ -402,10 +487,7 @@ def fit(
         else:
             composed_weights = Q_weights
 
-        sqrt_w = np.sqrt(composed_weights)
         Y = stacked_df[outcome].values
-        Y_t = Y * sqrt_w
-        X_t = X * sqrt_w[:, np.newaxis]
 
         # Cluster IDs
         if self.cluster == "unit":
@@ -415,16 +497,127 @@ def fit(
                 stacked_df[unit].astype(str) + "_" + stacked_df["_sub_exp"].astype(str)
             ).values
 
-        # Run OLS on transformed data (= WLS)
-        coef, residuals, vcov = solve_ols(
-            X_t,
-            Y_t,
+        # WLS with weights=composed_weights. solve_ols internally bakes
+        # sqrt(w) for the coefficient solve and back-transforms to compute
+        # vcov on original-scale data via clubSandwich's WLS-CR2 algebra for
+        # hc2_bm (PR #475). The hc1 path remains bit-equal to the prior
+        # bake-Q-into-X form (WLS-CR1 score is invariant). Note: this path
+        # routes through the Python backend regardless of vcov_type per
+        # `linalg.py:747-751` (Rust skips weighted vcov); the prior bake-Q
+        # path also went through Python in practice on stacked designs.
+        coef, _residuals_unused, vcov = solve_ols(
+            X,
+            Y,
             cluster_ids=cluster_ids,
+            weights=composed_weights,
+            weight_type="pweight",
+            vcov_type=self.vcov_type,
             return_vcov=True,
             rank_deficient_action=self.rank_deficient_action,
         )
         assert vcov is not None
 
+        # Bell-McCaffrey Satterthwaite contrast DOF for hc2_bm. Per the
+        # registry contract for `vcov_type="hc2_bm"`, the user-facing
+        # aggregated inference (event_study_effects[h]['p_value']/['conf_int']
+        # and overall_p_value/overall_conf_int) must use CR2 Bell-McCaffrey
+        # Satterthwaite DOF for each contrast — not the normal distribution
+        # that safe_inference(df=None) would otherwise default to. Mirrors
+        # the SunAbraham aggregated-inference pattern from PR #472
+        # (sun_abraham.py:997-1097) and the MPD avg_att pattern from PR #465.
+        # Computed BEFORE constructing event_study_effects / overall_*
+        # inference so the DOFs can be threaded into the safe_inference calls.
+        _bm_contrast_dof_per_event: Dict[int, float] = {}
+        _bm_contrast_dof_overall: Optional[float] = None
+        if self.vcov_type == "hc2_bm" and not _uses_replicate_sd and not np.all(np.isnan(coef)):
+            from diff_diff.linalg import _compute_cr2_bm_contrast_dof
+
+            # Mirror the MultiPeriodDiD rank-deficient pattern (PR #465,
+            # estimators.py:1860-1913): solve_ols emits NaN for dropped
+            # coefficients under R-style rank handling. Subset X, bread,
+            # and contrast vectors to the identified-column block BEFORE
+            # calling _compute_cr2_bm_contrast_dof; otherwise the singular
+            # full-design bread would raise LinAlgError and downgrade
+            # identified contrasts to normal-theory inference (R2 codex
+            # P1: catch-and-fallback was too aggressive for identified
+            # target contrasts).
+            _identified = ~np.isnan(coef)
+            _kept = np.where(_identified)[0]
+            X_kept = X[:, _kept]
+            bread_kept = X_kept.T @ (X_kept * composed_weights[:, np.newaxis])
+            k_design = X.shape[1]
+            # Per-event-time contrast: unit vector at the delta_h column.
+            # Only build contrasts whose target column is identified; if a
+            # delta_h column itself was dropped, that event-time will get
+            # NaN inference (left to safe_inference's df=None path).
+            # Per CI codex R1 P3: skip per-event contrast DOFs when the
+            # event-study surface is not user-visible (aggregate != "event_study").
+            # The overall ATT contrast still gets computed below.
+            es_keys: List[int] = []
+            es_cols_full: List[np.ndarray] = []
+            if aggregate == "event_study":
+                for h in event_times:
+                    if h in interaction_indices and _identified[interaction_indices[h]]:
+                        c = np.zeros(k_design)
+                        c[interaction_indices[h]] = 1.0
+                        es_keys.append(h)
+                        es_cols_full.append(c)
+            # Overall ATT contrast: average of post-period delta_h columns
+            # (the same 1/K * ones contrast used for overall_se below). Only
+            # construct if ALL post-period delta_h are identified — otherwise
+            # the contrast is undefined.
+            _post_event_times_preview = [
+                h for h in event_times if h >= -self.anticipation and h in interaction_indices
+            ]
+            _post_all_identified = all(
+                _identified[interaction_indices[h]] for h in _post_event_times_preview
+            )
+            overall_col_full: Optional[np.ndarray] = None
+            if len(_post_event_times_preview) > 0 and _post_all_identified:
+                K_prev = len(_post_event_times_preview)
+                overall_col_full = np.zeros(k_design)
+                for h in _post_event_times_preview:
+                    overall_col_full[interaction_indices[h]] = 1.0 / K_prev
+            if es_cols_full or overall_col_full is not None:
+                # Subset all contrasts to the kept columns. Since each contrast
+                # is non-zero only at identified columns (by construction
+                # above), no information is lost in the subset.
+                cols_full = list(es_cols_full)
+                if overall_col_full is not None:
+                    cols_full.append(overall_col_full)
+                contrasts_full = np.column_stack(cols_full)
+                contrasts_kept = contrasts_full[_kept, :]
+                try:
+                    dof_vec = _compute_cr2_bm_contrast_dof(
+                        X_kept,
+                        cluster_ids,
+                        bread_kept,
+                        contrasts_kept,
+                        weights=composed_weights,
+                    )
+                    for idx, h in enumerate(es_keys):
+                        _bm_contrast_dof_per_event[h] = float(dof_vec[idx])
+                    if overall_col_full is not None:
+                        _bm_contrast_dof_overall = float(dof_vec[-1])
+                except (ValueError, np.linalg.LinAlgError) as exc:
+                    # Genuine singularity on the IDENTIFIED design (very rare
+                    # — the rank-deficient handling above already subsets to
+                    # identified columns). Emit a UserWarning; the downstream
+                    # inference path NaN-closes (per the fail-closed contract
+                    # added in this PR) so the user receives undefined
+                    # inference rather than silent normal-theory fallback.
+                    warnings.warn(
+                        f"StackedDiD(vcov_type='hc2_bm') aggregated inference "
+                        f"could not compute Bell-McCaffrey contrast DOF on the "
+                        f"identified-column design ({type(exc).__name__}: "
+                        f"{exc}). Aggregated p-values, t-statistics, and "
+                        "confidence intervals will be returned as NaN to "
+                        "preserve the hc2_bm contract (small-sample inference "
+                        "must use BM Satterthwaite DOF, not normal-theory).",
+                        UserWarning,
+                        stacklevel=2,
+                    )
+
         # ---- Survey VCV override ----
         _n_valid_rep_sd = None
         resolved_stacked = None
@@ -434,17 +627,27 @@ def fit(
 
             resolved_stacked = survey_design.resolve(stacked_df)
 
-            # Refit closure: compose Q-weights with replicate survey weights
+            # Refit closure: compose Q-weights with replicate survey weights.
+            # Threads vcov_type=self.vcov_type for grep-consistency though the
+            # closure uses return_vcov=False (only the coef is consumed by
+            # compute_replicate_refit_variance). The vcov_type passed here is
+            # always "hc1" at runtime because the survey + non-hc1 reject in
+            # fit() fires before this branch can be reached for any other
+            # vcov_type.
             def _refit_stacked(w_r):
                 composed_r = Q_weights * w_r
                 w_sum = np.sum(composed_r)
                 if w_sum > 0:
                     composed_r = composed_r * (n_stacked / w_sum)
-                sqrt_w_r = np.sqrt(composed_r)
                 coef_r, _, _ = solve_ols(
-                    X * sqrt_w_r[:, np.newaxis], Y * sqrt_w_r,
+                    X,
+                    Y,
                     cluster_ids=cluster_ids,
-                    rank_deficient_action="silent", return_vcov=False,
+                    weights=composed_r,
+                    weight_type="pweight",
+                    vcov_type=self.vcov_type,
+                    rank_deficient_action="silent",
+                    return_vcov=False,
                 )
                 return coef_r
 
@@ -524,9 +727,26 @@ def _refit_stacked(w_r):
                         _survey_df = _n_valid_rep_sd - 1 if _n_valid_rep_sd > 1 else 0
                         if survey_metadata is not None:
                             survey_metadata.df_survey = _survey_df if _survey_df > 0 else None
-                t_stat, p_value, conf_int = safe_inference(
-                    effect, se, alpha=self.alpha, df=_survey_df
-                )
+                # Use BM contrast DOF for hc2_bm. Fail-closed: when the
+                # hc2_bm contract is in effect but BM DOF is unavailable (helper
+                # failed OR noise-floor NaN guard fired), emit all-NaN inference
+                # rather than fall back to normal-theory CIs/p-values. Mirrors
+                # the fix in LinearRegression.get_inference() from PR #475 R7
+                # (linalg.py:3689-3706). Without this, safe_inference(df=NaN)
+                # would pass df comparison >= 0 (NaN < 0 is False) and emit
+                # finite t_stat with NaN p/CI — silent wrong inference.
+                _is_hc2bm_path = self.vcov_type == "hc2_bm" and not _uses_replicate_sd
+                _bm_df = _bm_contrast_dof_per_event.get(h)
+                if _is_hc2bm_path and (_bm_df is None or not np.isfinite(_bm_df)):
+                    # BM DOF unavailable on hc2_bm path: NaN-out inference.
+                    t_stat = float("nan")
+                    p_value = float("nan")
+                    conf_int = (float("nan"), float("nan"))
+                else:
+                    _df_eff = _bm_df if _bm_df is not None else _survey_df
+                    t_stat, p_value, conf_int = safe_inference(
+                        effect, se, alpha=self.alpha, df=_df_eff
+                    )
                 n_obs_h = int(np.sum((et_vals == h) & (d_vals == 1)))
                 event_study_effects[h] = {
                     "effect": effect,
@@ -565,10 +785,30 @@ def _refit_stacked(w_r):
             if _n_valid_rep_sd < resolved_stacked.n_replicates:
                 _survey_df_overall = _n_valid_rep_sd - 1 if _n_valid_rep_sd > 1 else 0
                 if survey_metadata is not None:
-                    survey_metadata.df_survey = _survey_df_overall if _survey_df_overall > 0 else None
-        overall_t, overall_p, overall_ci = safe_inference(
-            overall_att, overall_se, alpha=self.alpha, df=_survey_df_overall
-        )
+                    survey_metadata.df_survey = (
+                        _survey_df_overall if _survey_df_overall > 0 else None
+                    )
+        # Use BM contrast DOF for overall ATT (hc2_bm). Fail-closed: when
+        # the hc2_bm contract is in effect but BM DOF is unavailable, emit
+        # all-NaN inference (per the LinearRegression.get_inference pattern
+        # from PR #475 R7). Without this, normal-theory fallback would
+        # silently produce wrong p-values/CIs on the overall_* surface.
+        _is_hc2bm_path_overall = self.vcov_type == "hc2_bm" and not _uses_replicate_sd
+        if _is_hc2bm_path_overall and (
+            _bm_contrast_dof_overall is None or not np.isfinite(_bm_contrast_dof_overall)
+        ):
+            overall_t = float("nan")
+            overall_p = float("nan")
+            overall_ci = (float("nan"), float("nan"))
+        else:
+            _df_overall_eff = (
+                _bm_contrast_dof_overall
+                if _bm_contrast_dof_overall is not None
+                else _survey_df_overall
+            )
+            overall_t, overall_p, overall_ci = safe_inference(
+                overall_att, overall_se, alpha=self.alpha, df=_df_overall_eff
+            )
 
         # ---- Construct results ----
         self.results_ = StackedDiDResults(
@@ -594,6 +834,9 @@ def _refit_stacked(w_r):
             clean_control=self.clean_control,
             alpha=self.alpha,
             anticipation=self.anticipation,
+            vcov_type=self.vcov_type,
+            cluster_name=self.cluster,
+            n_clusters=int(np.unique(cluster_ids).size),
             survey_metadata=survey_metadata,
         )
 
@@ -952,10 +1195,21 @@ def get_params(self) -> Dict[str, Any]:
             "alpha": self.alpha,
             "anticipation": self.anticipation,
             "rank_deficient_action": self.rank_deficient_action,
+            "vcov_type": self.vcov_type,
         }
 
     def set_params(self, **params: Any) -> "StackedDiD":
-        """Set estimator parameters (sklearn-compatible)."""
+        """Set estimator parameters (sklearn-compatible).
+
+        Re-validates `vcov_type` via the shared `_validate_vcov_type`
+        helper so sklearn-style mutation hits the estimator-level guard
+        before fit() (avoids a later, less-informative failure in the
+        linalg layer).
+        """
+        # Validate vcov_type up-front if it's being set, so the same
+        # error surface as __init__ applies.
+        if "vcov_type" in params:
+            self._validate_vcov_type(params["vcov_type"])
         for key, value in params.items():
             if hasattr(self, key):
                 setattr(self, key, value)
diff --git a/diff_diff/stacked_did_results.py b/diff_diff/stacked_did_results.py
index e6a38e6e..5a5cbeed 100644
--- a/diff_diff/stacked_did_results.py
+++ b/diff_diff/stacked_did_results.py
@@ -94,6 +94,17 @@ class StackedDiDResults:
     clean_control: str = "not_yet_treated"
     alpha: float = 0.05
     anticipation: int = 0
+    # Analytical variance family configured at fit time (Phase 1b 2/8). When
+    # survey_design= is supplied the survey TSL/replicate variance overrides
+    # the analytical family; this field still records the configured value.
+    vcov_type: str = "hc1"
+    # Cluster identity ("unit" or "unit_subexp") and realized cluster count
+    # at fit time. Used by summary() to render the correct CR1/CR2-BM label
+    # via `_format_vcov_label(cluster_name=, n_clusters=)`. Per CI codex R2
+    # P2: passing cluster_name=None mislabelled clustered StackedDiD fits
+    # as one-way HC1/HC2-BM. StackedDiD is intrinsically clustered.
+    cluster_name: Optional[str] = None
+    n_clusters: Optional[int] = None
     # Survey design metadata (SurveyMetadata instance from diff_diff.survey)
     survey_metadata: Optional[Any] = field(default=None)
 
@@ -172,6 +183,24 @@ def summary(self, alpha: Optional[float] = None) -> str:
             "",
         ]
 
+        # Variance family label (per CI codex R1 P2): surface the analytical
+        # vcov_type when the survey path didn't override. Per R2 P2: pass
+        # cluster_name + n_clusters so the label renders as "CR1 cluster-
+        # robust at unit, G=N" rather than the one-way "HC1 heteroskedasticity-
+        # robust" — StackedDiD is intrinsically clustered.
+        if self.survey_metadata is None and self.vcov_type:
+            from diff_diff.results import _format_vcov_label
+
+            label = _format_vcov_label(
+                self.vcov_type,
+                cluster_name=self.cluster_name,
+                n_clusters=self.n_clusters,
+                n_obs=self.n_stacked_obs,
+            )
+            if label is not None:
+                lines.append(f"{'Variance:':<30} {label:>50}")
+                lines.append("")
+
         # Add survey design info
         if self.survey_metadata is not None:
             sm = self.survey_metadata
diff --git a/docs/methodology/REGISTRY.md b/docs/methodology/REGISTRY.md
index 39bb1f45..279ce571 100644
--- a/docs/methodology/REGISTRY.md
+++ b/docs/methodology/REGISTRY.md
@@ -1364,6 +1364,15 @@ Estimated via WLS with Q-weights. The delta_h coefficients identify theta_kappa^
 - No special bootstrap procedure specified; standard cluster-robust SEs recommended
 - For post-period average: delta method or `lincom`/`marginaleffects`
 
+*Variance families (`vcov_type`):*
+- `hc1` (default) — CR1 Liang-Zeger cluster-robust via `solve_ols(weights=composed_weights, vcov_type="hc1")` (Stata-style `G/(G-1) * (n-1)/(n-p)` finite-sample correction). Matches `clubSandwich::vcovCR(lm(weights=Q,...), cluster=~unit, type="CR1S")` at atol=1e-10 on the new `benchmarks/data/stacked_did_golden.json` fixture (R-side target is `CR1S` not plain `CR1` — the latter omits the `(n-1)/(n-p)` term and diverges by ~1.4% on a 325-row, 10-column stacked design). Bit-equal to the prior bake-Q-into-X output up to float64 multiplication ordering (atol=1e-13; WLS-CR1 score is algebraically invariant between the two forms).
+- `hc2_bm` — CR2 Bell-McCaffrey via `solve_ols(weights=composed_weights, vcov_type="hc2_bm")` routed through the clubSandwich WLS-CR2 port (PR #475). The diff-diff implementation matches clubSandwich's specific algebra (W not √W in the hat matrix, W² in the bias-correction term, unweighted residuals in the score construction) — see Phase 1a `hc2_bm + weights` registry row for the full derivation. Matches `clubSandwich::vcovCR(lm(weights=Q,...), cluster=~unit, type="CR2") + coef_test()$df_Satt` at atol=1e-10 (pinned in `tests/test_methodology_stacked_did.py`). **Bell-McCaffrey Satterthwaite DOF is threaded into the user-facing aggregated inference** (`event_study_effects[h]['p_value']`/`['conf_int']` use the per-event-time contrast DOF from `_compute_cr2_bm_contrast_dof`; `overall_p_value`/`overall_conf_int` use the post-period-average contrast DOF, matching R `Wald_test(constraints=row, vcov=CR2, test="HTZ")$df_denom` at atol=1e-10). Mirrors the SunAbraham aggregated-inference pattern from PR #472.
+- `classical`, `hc2` — REJECTED at `__init__` with `ValueError`. StackedDiD clusters intrinsically at `'unit'` or `'unit_subexp'` (no `cluster=None` opt-out); the linalg validator rejects one-way families paired with `cluster_ids`. Use `vcov_type='hc1'` or `'hc2_bm'`.
+- `conley` — REJECTED at `__init__` with `ValueError` (deferred; threading the six `conley_*` params through `solve_ols` is tracked in `TODO.md` alongside the SunAbraham conley follow-up).
+- `survey_design=` + `vcov_type ∈ {classical, hc2, hc2_bm}` — REJECTED at `fit()` with `NotImplementedError`; the survey TSL or replicate-weight refit variance overrides the analytical sandwich. Use `vcov_type='hc1'` (default) for survey designs. Reject order: the existing fweight/aweight check at `stacked_did.py:309` fires first (Q-weight ratio semantics), then the survey + non-hc1 check at `stacked_did.py:~325` — locked by `test_aweight_plus_hc2_bm_rejected_by_stacked_did_level_guard`.
+
+**Note:** This routing inherits the WLS-CR2 methodology choice from the Phase 1a clubSandwich port (PR #475 / REGISTRY Phase 1a `hc2_bm + weights` row); see that row for the full PT2018-§3.3-vs-clubSandwich algebra deviation derivation. No new methodology choice is introduced in this PR. The change is purely surface: switching from the prior bake-Q-into-X pattern (`X_t = X * sqrt(Q)`, `solve_ols(X_t, Y_t, cluster_ids=)`) to `solve_ols(weights=composed_weights, vcov_type=...)` opens the hc2_bm path without modifying the small-sample WLS-CR2 algebra. The HC1 path is preserved bit-equal (up to float64 multiplication ordering at machine precision).
+
 *Edge cases:*
 - All events trimmed: `len(Omega_kappa) == 0` -> ValueError suggesting reduced kappa
 - No clean controls for event a: IC2 check fails -> Trim event, warn user
diff --git a/tests/test_methodology_stacked_did.py b/tests/test_methodology_stacked_did.py
new file mode 100644
index 00000000..641d14a3
--- /dev/null
+++ b/tests/test_methodology_stacked_did.py
@@ -0,0 +1,669 @@
+"""R-parity tests for StackedDiD's vcov_type threading (Phase 1b 2/8).
+
+Pins Python StackedDiD output against R `clubSandwich::vcovCR(lm(weights=Q,...),
+type="CR1S"|"CR2") + coef_test()$df_Satt` and `Wald_test(test="HTZ")$df_denom`
+at atol=1e-10. The clubSandwich port in `diff_diff/linalg.py` (PR #475)
+supplies the underlying WLS-CR2 algebra; this file verifies that StackedDiD's
+call into `solve_ols(weights=composed_weights, vcov_type=...)` produces output
+consistent with the R reference, including the Bell-McCaffrey Satterthwaite
+DOF threading at both the per-event-time level and the post-period-average
+contrast level.
+
+Goldens are generated by `benchmarks/R/generate_stacked_did_golden.R` (loads
+the pre-stacked panel CSV at `benchmarks/data/stacked_did_test_panel.csv`,
+fits `lm(weights=Q,...)`, dumps CR1S + CR2-BM + Satterthwaite DOF for the
+event-study coefficients AND the post-period-average ATT contrast at both
+`cluster=unit` and `cluster=unit_subexp`).
+
+Note on `CR1` vs `CR1S`: diff-diff's HC1 + cluster uses Stata-style
+`G/(G-1) * (n-1)/(n-p)` finite-sample correction, which matches
+`clubSandwich::vcovCR(type="CR1S")`. The plain `clubSandwich::vcovCR(type="CR1")`
+omits the `(n-1)/(n-p)` term and diverges by ~1.4% on this fixture.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+
+import numpy as np
+import pytest
+from scipy.optimize import brentq
+from scipy.stats import t as t_dist
+
+from diff_diff import StackedDiD
+from diff_diff.prep_dgp import generate_staggered_data
+
+_GOLDEN_PATH = os.path.join(
+    os.path.dirname(__file__),
+    "..",
+    "benchmarks",
+    "data",
+    "stacked_did_golden.json",
+)
+_R_FIXTURE_AVAILABLE = os.path.exists(_GOLDEN_PATH)
+
+
+@pytest.fixture(scope="module")
+def goldens():
+    if not _R_FIXTURE_AVAILABLE:
+        pytest.skip(
+            "stacked_did_golden.json not generated; run "
+            "Rscript benchmarks/R/generate_stacked_did_golden.R"
+        )
+    with open(_GOLDEN_PATH) as f:
+        return json.load(f)
+
+
+_PANEL_CSV_PATH = os.path.join(
+    os.path.dirname(__file__),
+    "..",
+    "benchmarks",
+    "data",
+    "stacked_did_test_panel.csv",
+)
+
+
+@pytest.fixture(scope="module")
+def panel():
+    """Same panel descriptor as the R fixture; must match exactly.
+
+    Per R6 P3: cross-check the regenerated panel against the committed
+    `stacked_did_test_panel.csv` to detect any drift in
+    `generate_staggered_data` that would otherwise silently invalidate the
+    R-parity assertions. The CSV is the post-stacking panel; we verify
+    that the pre-stacking columns (unit, period, outcome, first_treat)
+    match column-by-column. If `generate_staggered_data` ever changes,
+    this test will FAIL with a clear DGP-drift message rather than
+    producing wrong parity output.
+    """
+    data = generate_staggered_data(
+        n_units=50,
+        n_periods=8,
+        cohort_periods=[3, 5, 7],
+        never_treated_frac=0.3,
+        treatment_effect=2.0,
+        dynamic_effects=False,
+        seed=20260521,
+    )
+    # DGP drift check: load committed CSV (post-stacking, contains a subset
+    # of original rows duplicated across sub-experiments) and verify that
+    # the (unit, period) tuples present in the CSV have matching `outcome`
+    # values in the regenerated panel. The CSV is the canonical reference
+    # used by the R fixture script; any drift in `generate_staggered_data`
+    # that changes the underlying outcomes would otherwise silently
+    # invalidate the R-parity assertions below.
+    if os.path.exists(_PANEL_CSV_PATH):
+        import pandas as pd
+
+        csv = pd.read_csv(_PANEL_CSV_PATH)
+        csv_unique = csv.drop_duplicates(subset=["unit", "period"]).copy()
+        merged = csv_unique.merge(
+            data[["unit", "period", "outcome", "first_treat"]],
+            on=["unit", "period"],
+            how="inner",
+            suffixes=("_csv", "_regen"),
+        )
+        assert len(merged) == len(csv_unique), (
+            f"DGP drift: regenerated panel missing (unit, period) rows present "
+            f"in committed CSV. CSV unique rows={len(csv_unique)}, "
+            f"merged rows={len(merged)}. Regenerate the CSV if the DGP "
+            "helper change is intentional."
+        )
+        # Outcome parity (drives the R-side fit).
+        np.testing.assert_allclose(
+            merged["outcome_regen"].astype(float).values,
+            merged["outcome_csv"].astype(float).values,
+            atol=1e-12,
+            err_msg=(
+                "DGP drift detected: generate_staggered_data(...) produces "
+                "different 'outcome' values than the committed "
+                "benchmarks/data/stacked_did_test_panel.csv on the "
+                "matching (unit, period) rows. Regenerate the CSV "
+                "if the change is intentional."
+            ),
+        )
+        # Treatment-timing parity: normalize first_treat (CSV stores inf
+        # for never-treated; regenerated panel uses 0 in some pandas
+        # versions). Per R7 P3: pin the treatment-timing inputs too so
+        # a future drift in cohort assignment is caught independently of
+        # outcome drift.
+        regen_ft = merged["first_treat_regen"].astype(float).replace(0.0, np.inf).values
+        csv_ft = merged["first_treat_csv"].astype(float).values
+        # CSV stores 0 for never-treated in some serializations too; normalize.
+        csv_ft = np.where(csv_ft == 0.0, np.inf, csv_ft)
+        np.testing.assert_array_equal(
+            regen_ft,
+            csv_ft,
+            err_msg=(
+                "DGP drift detected: generate_staggered_data(...) produces "
+                "different treatment timing ('first_treat') than the committed "
+                "CSV. Cohort assignment changed; regenerate CSV if intentional."
+            ),
+        )
+    return data
+
+
+def _fit(panel, vcov_type, cluster):
+    est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type=vcov_type, cluster=cluster)
+    return est.fit(
+        panel,
+        outcome="outcome",
+        unit="unit",
+        time="period",
+        first_treat="first_treat",
+        aggregate="event_study",
+    )
+
+
+def _es_se_vector(res, event_times_non_ref):
+    """Extract event-study SE vector in the order given by event_times_non_ref."""
+    return np.array([res.event_study_effects[h]["se"] for h in event_times_non_ref])
+
+
+def _recover_dof_from_ci(att, se, ci_hi, alpha):
+    """Recover DOF used by safe_inference from the CI half-width.
+
+    Inverse of `ci_hi = att + se * t.ppf(1 - alpha/2, df)`. Used to assert
+    the BM Satterthwaite DOF threaded correctly into the result-class
+    inference fields without relying on internal estimator state.
+
+    Raises ValueError if the inversion fails (e.g., when safe_inference
+    used the normal distribution instead of t — meaning the BM DOF did
+    NOT propagate and the test should FAIL, not silently skip).
+    """
+    half_width = ci_hi - att
+    t_crit = half_width / se
+    target_alpha = alpha / 2  # 0.025 for alpha=0.05
+
+    def f(df):
+        return t_dist.sf(t_crit, df) - target_alpha
+
+    # Use a wide bracket; brentq raises ValueError if no root.
+    return brentq(f, 1.0, 1e9, xtol=1e-9)
+
+
+class TestStackedDiDParityR:
+    """Pin StackedDiD output against clubSandwich at atol=1e-10.
+
+    Coverage:
+      - hc1 + cluster=unit SE matches CR1S (Stata-style correction)
+      - hc2_bm + cluster=unit SE matches CR2 Bell-McCaffrey
+      - hc2_bm + cluster=unit per-event-time DOF matches coef_test()$df_Satt
+      - hc2_bm + cluster=unit overall ATT DOF matches Wald_test(HTZ)$df_denom
+      - hc2_bm + cluster=unit_subexp SE matches CR2 (symmetric)
+
+    The DOF parity tests use CI-width inversion to recover the DOF that
+    `safe_inference` actually used. If the BM DOF doesn't propagate (the
+    R1 P0 bug), `safe_inference` uses the normal distribution and the
+    half-width is `se * 1.96`, which inverts to `df → infinity` — that
+    inversion FAILS within the bracket [1, 1e9] and raises ValueError,
+    which surfaces as a test failure (not a silent skip).
+    """
+
+    def test_hc1_se_matches_clubsandwich_cr1s_unit_cluster(self, goldens, panel):
+        """Default cluster=unit, vcov_type=hc1 matches R `vcovCR(type='CR1S')`.
+
+        Pins BOTH per-event-time SE AND the post-period-average overall_se
+        (per R6 P3: prior fixture only pinned per-event SE; the overall
+        delta-method SE was untested, leaving a gap where post-period
+        covariance regressions could slip through).
+
+        See module docstring: 'CR1S' (Stata-style) is the correct comparison
+        target; plain 'CR1' (clubSandwich's no-(n-1)/(n-p) form) would diverge
+        by ~1.4% on this fixture.
+        """
+        event_times = goldens["meta"]["event_times_non_ref"]
+        r_se = np.array(goldens["unit"]["se_cr1_es"])
+        r_overall_se = float(goldens["unit"]["se_overall_cr1"])
+        res = _fit(panel, vcov_type="hc1", cluster="unit")
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg=(
+                "StackedDiD hc1 event-study SE must match clubSandwich CR1S at atol=1e-10. "
+                "Note: target is CR1S (Stata-style), not plain CR1."
+            ),
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg=(
+                "StackedDiD hc1 overall_se (post-period delta-method) must "
+                "match clubSandwich CR1S contrast variance at atol=1e-10"
+            ),
+        )
+
+    def test_hc2_bm_se_matches_clubsandwich_cr2_unit_cluster(self, goldens, panel):
+        """cluster=unit, vcov_type=hc2_bm matches R `vcovCR(type='CR2')`.
+
+        Pins BOTH per-event-time SE AND overall_se contract variance
+        (per R6 P3).
+        """
+        event_times = goldens["meta"]["event_times_non_ref"]
+        r_se = np.array(goldens["unit"]["se_cr2_es"])
+        r_overall_se = float(goldens["unit"]["se_overall_cr2"])
+        res = _fit(panel, vcov_type="hc2_bm", cluster="unit")
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="StackedDiD hc2_bm event-study SE must match clubSandwich CR2 at atol=1e-10",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg=(
+                "StackedDiD hc2_bm overall_se must match clubSandwich CR2 "
+                "contrast variance at atol=1e-10"
+            ),
+        )
+
+    def test_hc2_bm_per_event_dof_matches_coef_test_df_satt_unit_cluster(self, goldens, panel):
+        """Per-event-time BM DOF propagated via safe_inference matches
+        `coef_test(..., vcov=CR2)$df_Satt`.
+
+        Recovers the DOF from CI half-width inversion. If BM DOF didn't
+        propagate, `safe_inference` would use the normal distribution
+        (df=None) and the recovery would fail with ValueError — which
+        surfaces as a test failure, NOT a silent skip. This was the R1 P0
+        failure mode caught by CI codex.
+        """
+        event_times = goldens["meta"]["event_times_non_ref"]
+        r_dof = np.array(goldens["unit"]["dof_bm_es"])
+        res = _fit(panel, vcov_type="hc2_bm", cluster="unit")
+        validated_count = 0
+        for h, r_dof_h in zip(event_times, r_dof):
+            es = res.event_study_effects[h]
+            att, se, ci_hi = es["effect"], es["se"], es["conf_int"][1]
+            # If propagation failed, this raises ValueError → test fails.
+            df_python = _recover_dof_from_ci(att, se, ci_hi, res.alpha)
+            np.testing.assert_allclose(
+                df_python,
+                r_dof_h,
+                atol=1e-6,
+                rtol=1e-6,
+                err_msg=(
+                    f"BM DOF for event_time={h} must match clubSandwich coef_test "
+                    f"df_Satt; recovered df_python={df_python}, R df_Satt={r_dof_h}"
+                ),
+            )
+            validated_count += 1
+        # Hard guard: ensure we actually validated every coefficient.
+        assert validated_count == len(event_times), (
+            f"Expected to validate DOF for all {len(event_times)} event-time "
+            f"coefficients, only validated {validated_count}"
+        )
+
+    def test_hc2_bm_overall_att_dof_matches_wald_test_htz_unit_cluster(self, goldens, panel):
+        """Overall ATT BM DOF matches `Wald_test(HTZ)$df_denom` (per review P2:
+        the post-period-average contrast was missing from the prior fixture)."""
+        r_dof_overall = float(goldens["unit"]["dof_bm_overall"])
+        res = _fit(panel, vcov_type="hc2_bm", cluster="unit")
+        att, se, ci_hi = res.overall_att, res.overall_se, res.overall_conf_int[1]
+        df_python = _recover_dof_from_ci(att, se, ci_hi, res.alpha)
+        np.testing.assert_allclose(
+            df_python,
+            r_dof_overall,
+            atol=1e-6,
+            rtol=1e-6,
+            err_msg=(
+                f"Overall ATT BM DOF must match clubSandwich Wald_test(HTZ) "
+                f"df_denom; recovered df_python={df_python}, R df={r_dof_overall}"
+            ),
+        )
+
+    def test_hc2_bm_se_matches_clubsandwich_cr2_unit_subexp_cluster(self, goldens, panel):
+        """Same parity for the alternate cluster level (per-event SE +
+        overall_se contract variance)."""
+        event_times = goldens["meta"]["event_times_non_ref"]
+        r_se = np.array(goldens["unit_subexp"]["se_cr2_es"])
+        r_overall_se = float(goldens["unit_subexp"]["se_overall_cr2"])
+        res = _fit(panel, vcov_type="hc2_bm", cluster="unit_subexp")
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg=(
+                "StackedDiD hc2_bm + cluster=unit_subexp must match "
+                "clubSandwich CR2 at atol=1e-10"
+            ),
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg=(
+                "StackedDiD hc2_bm + cluster=unit_subexp overall_se must "
+                "match clubSandwich CR2 at atol=1e-10"
+            ),
+        )
+
+    def test_hc2_bm_overall_att_dof_matches_wald_test_htz_unit_subexp_cluster(self, goldens, panel):
+        """Symmetric overall ATT DOF parity at the alternate cluster level."""
+        r_dof_overall = float(goldens["unit_subexp"]["dof_bm_overall"])
+        res = _fit(panel, vcov_type="hc2_bm", cluster="unit_subexp")
+        att, se, ci_hi = res.overall_att, res.overall_se, res.overall_conf_int[1]
+        df_python = _recover_dof_from_ci(att, se, ci_hi, res.alpha)
+        np.testing.assert_allclose(
+            df_python,
+            r_dof_overall,
+            atol=1e-6,
+            rtol=1e-6,
+            err_msg=(
+                f"Overall ATT BM DOF (unit_subexp) must match clubSandwich; "
+                f"recovered df_python={df_python}, R df={r_dof_overall}"
+            ),
+        )
+
+
+# =============================================================================
+# Variant parity (per R8 P1): pin non-default vcov_type surfaces to R goldens.
+# =============================================================================
+
+
+@pytest.fixture(scope="module")
+def population_panel():
+    """Heterogeneous-population panel matching the R variant fixture.
+
+    Same seed/population assignment as the Python helper that dumped
+    `stacked_did_population_panel.csv`. Per R8 P1: the prior `population`
+    smoke test used a constant pop=1.0 which is numerically degenerate
+    with aggregate weighting. This fixture forces heterogeneous Q-weights.
+    """
+    rng = np.random.default_rng(99)
+    data = generate_staggered_data(
+        n_units=50,
+        n_periods=8,
+        cohort_periods=[3, 5, 7],
+        never_treated_frac=0.3,
+        treatment_effect=2.0,
+        dynamic_effects=False,
+        seed=20260521,
+    )
+    unit_pops = rng.uniform(1.0, 100.0, size=data["unit"].nunique())
+    data = data.copy()
+    data["pop"] = data["unit"].map(lambda u: unit_pops[int(u)])
+    return data
+
+
+class TestStackedDiDParityRVariants:
+    """R-parity for non-default vcov_type surfaces (R8 P1 coverage)."""
+
+    def test_population_weighting_hc1_matches_clubsandwich_cr1s(self, goldens, population_panel):
+        """weighting='population' with heterogeneous Q-weights produces
+        SE that differs from the aggregate-weight case AND matches R."""
+        if "population" not in goldens:
+            pytest.skip("population variant not in goldens; regenerate fixture")
+        r_se_es = np.array(goldens["population"]["unit"]["se_cr1_es"])
+        r_overall_se = float(goldens["population"]["unit"]["se_overall_cr1"])
+        event_times = goldens["population"]["meta"]["event_times_non_ref"]
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1", weighting="population")
+        res = est.fit(
+            population_panel,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            population="pop",
+            aggregate="event_study",
+        )
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se_es,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="StackedDiD weighting='population' hc1 event-study SE must match R CR1S",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            err_msg="StackedDiD weighting='population' hc1 overall_se must match R",
+        )
+
+    def test_population_weighting_hc2_bm_matches_clubsandwich_cr2(self, goldens, population_panel):
+        if "population" not in goldens:
+            pytest.skip("population variant not in goldens")
+        r_se_es = np.array(goldens["population"]["unit"]["se_cr2_es"])
+        r_overall_se = float(goldens["population"]["unit"]["se_overall_cr2"])
+        event_times = goldens["population"]["meta"]["event_times_non_ref"]
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", weighting="population")
+        res = est.fit(
+            population_panel,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            population="pop",
+            aggregate="event_study",
+        )
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se_es,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="StackedDiD weighting='population' hc2_bm event-study SE must match R CR2",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            err_msg="StackedDiD weighting='population' hc2_bm overall_se must match R",
+        )
+        # DOF parity (per R9 P1)
+        r_dof_es = np.array(goldens["population"]["unit"]["dof_bm_es"])
+        r_dof_overall = float(goldens["population"]["unit"]["dof_bm_overall"])
+        for h, r_dof_h in zip(event_times, r_dof_es):
+            es = res.event_study_effects[h]
+            if es["se"] == 0:
+                continue
+            df_py = _recover_dof_from_ci(es["effect"], es["se"], es["conf_int"][1], res.alpha)
+            np.testing.assert_allclose(
+                df_py,
+                r_dof_h,
+                atol=1e-6,
+                rtol=1e-6,
+                err_msg=f"population hc2_bm BM DOF at h={h} must match R",
+            )
+        df_overall = _recover_dof_from_ci(
+            res.overall_att, res.overall_se, res.overall_conf_int[1], res.alpha
+        )
+        np.testing.assert_allclose(
+            df_overall,
+            r_dof_overall,
+            atol=1e-6,
+            rtol=1e-6,
+            err_msg="population hc2_bm overall BM DOF must match R Wald_test(HTZ)",
+        )
+
+    def test_anticipation1_hc1_matches_clubsandwich_cr1s(self, goldens, panel):
+        """anticipation=1: reference period shifts to e=-2, post-period
+        starts at h=-1, event window expands by 1 pre-period. Verify R
+        parity on both event-study and overall SE."""
+        if "anticipation1" not in goldens:
+            pytest.skip("anticipation1 variant not in goldens")
+        r_se_es = np.array(goldens["anticipation1"]["unit"]["se_cr1_es"])
+        r_overall_se = float(goldens["anticipation1"]["unit"]["se_overall_cr1"])
+        event_times = goldens["anticipation1"]["meta"]["event_times_non_ref"]
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1", anticipation=1)
+        res = est.fit(
+            panel,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se_es,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="StackedDiD anticipation=1 hc1 event-study SE must match R CR1S",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            err_msg="StackedDiD anticipation=1 hc1 overall_se must match R",
+        )
+
+    def test_anticipation1_hc2_bm_matches_clubsandwich_cr2(self, goldens, panel):
+        if "anticipation1" not in goldens:
+            pytest.skip("anticipation1 variant not in goldens")
+        r_se_es = np.array(goldens["anticipation1"]["unit"]["se_cr2_es"])
+        r_overall_se = float(goldens["anticipation1"]["unit"]["se_overall_cr2"])
+        event_times = goldens["anticipation1"]["meta"]["event_times_non_ref"]
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", anticipation=1)
+        res = est.fit(
+            panel,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se_es,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="StackedDiD anticipation=1 hc2_bm event-study SE must match R CR2",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            err_msg="StackedDiD anticipation=1 hc2_bm overall_se must match R",
+        )
+        # DOF parity (per R9 P1): under anticipation=1, the post-period
+        # contrast includes h=-1, so the BM DOF should reflect that extension.
+        r_dof_es = np.array(goldens["anticipation1"]["unit"]["dof_bm_es"])
+        r_dof_overall = float(goldens["anticipation1"]["unit"]["dof_bm_overall"])
+        for h, r_dof_h in zip(event_times, r_dof_es):
+            es = res.event_study_effects[h]
+            if es["se"] == 0:
+                continue
+            df_py = _recover_dof_from_ci(es["effect"], es["se"], es["conf_int"][1], res.alpha)
+            np.testing.assert_allclose(
+                df_py,
+                r_dof_h,
+                atol=1e-6,
+                rtol=1e-6,
+                err_msg=f"anticipation1 hc2_bm BM DOF at h={h} must match R",
+            )
+        df_overall = _recover_dof_from_ci(
+            res.overall_att, res.overall_se, res.overall_conf_int[1], res.alpha
+        )
+        np.testing.assert_allclose(
+            df_overall,
+            r_dof_overall,
+            atol=1e-6,
+            rtol=1e-6,
+            err_msg="anticipation1 hc2_bm overall BM DOF must match R Wald_test(HTZ)",
+        )
+
+    def test_sample_share_weighting_hc1_matches_clubsandwich_cr1s(self, goldens, panel):
+        """weighting='sample_share' uses a distinct Q-weight formula
+        `Q_sa = ((N^D_a + N^C_a)/(N^D+N^C)) / (N^C_a/N^C)` for controls.
+        Pin R parity at atol=1e-10 (per R9 P1)."""
+        if "sample_share" not in goldens:
+            pytest.skip("sample_share variant not in goldens")
+        r_se_es = np.array(goldens["sample_share"]["unit"]["se_cr1_es"])
+        r_overall_se = float(goldens["sample_share"]["unit"]["se_overall_cr1"])
+        event_times = goldens["sample_share"]["meta"]["event_times_non_ref"]
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1", weighting="sample_share")
+        res = est.fit(
+            panel,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se_es,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="sample_share hc1 event-study SE must match R CR1S",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            err_msg="sample_share hc1 overall_se must match R",
+        )
+
+    def test_sample_share_weighting_hc2_bm_matches_clubsandwich_cr2(self, goldens, panel):
+        """sample_share + hc2_bm: pin SE + BM DOF parity."""
+        if "sample_share" not in goldens:
+            pytest.skip("sample_share variant not in goldens")
+        r_se_es = np.array(goldens["sample_share"]["unit"]["se_cr2_es"])
+        r_overall_se = float(goldens["sample_share"]["unit"]["se_overall_cr2"])
+        r_dof_es = np.array(goldens["sample_share"]["unit"]["dof_bm_es"])
+        r_dof_overall = float(goldens["sample_share"]["unit"]["dof_bm_overall"])
+        event_times = goldens["sample_share"]["meta"]["event_times_non_ref"]
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", weighting="sample_share")
+        res = est.fit(
+            panel,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        py_se = _es_se_vector(res, event_times)
+        np.testing.assert_allclose(
+            py_se,
+            r_se_es,
+            atol=1e-10,
+            rtol=1e-10,
+            err_msg="sample_share hc2_bm event-study SE must match R CR2",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            r_overall_se,
+            atol=1e-10,
+            err_msg="sample_share hc2_bm overall_se must match R",
+        )
+        for h, r_dof_h in zip(event_times, r_dof_es):
+            es = res.event_study_effects[h]
+            if es["se"] == 0:
+                continue
+            df_py = _recover_dof_from_ci(es["effect"], es["se"], es["conf_int"][1], res.alpha)
+            np.testing.assert_allclose(
+                df_py,
+                r_dof_h,
+                atol=1e-6,
+                rtol=1e-6,
+                err_msg=f"sample_share hc2_bm BM DOF at h={h} must match R",
+            )
+        df_overall = _recover_dof_from_ci(
+            res.overall_att, res.overall_se, res.overall_conf_int[1], res.alpha
+        )
+        np.testing.assert_allclose(
+            df_overall,
+            r_dof_overall,
+            atol=1e-6,
+            rtol=1e-6,
+            err_msg="sample_share hc2_bm overall BM DOF must match R Wald_test(HTZ)",
+        )
diff --git a/tests/test_stacked_did.py b/tests/test_stacked_did.py
index 34a3a150..7fdf9925 100644
--- a/tests/test_stacked_did.py
+++ b/tests/test_stacked_did.py
@@ -925,3 +925,793 @@ def test_no_treated_units(self):
                 time="period",
                 first_treat="first_treat",
             )
+
+
+# =============================================================================
+# TestStackedDiDVcovType — Phase 1b 2/8: vcov_type threading
+# =============================================================================
+
+
+@pytest.fixture
+def baseline_panel():
+    """Fixed-seed panel matching the captured pre-PR HC1 SE baseline.
+
+    Captured on commit 955aa4be0887c71defb8cdab402e955f7c36e48d (PR #475
+    clubSandwich port merge) BEFORE Phase 1b 2/8 source edits. Used by
+    `test_hc1_se_bit_equal_to_pre_pr_baseline` to lock the bake-Q-into-X
+    -> explicit-weights= switch as bit-equal on the hc1 path.
+    """
+    return generate_staggered_data(
+        n_units=50,
+        n_periods=8,
+        cohort_periods=[3, 5, 7],
+        never_treated_frac=0.3,
+        treatment_effect=2.0,
+        dynamic_effects=False,
+        seed=20260521,
+    )
+
+
+class TestStackedDiDVcovType:
+    """Phase 1b 2/8: vcov_type input contract, reject paths, and bit-equality.
+
+    Mirrors the SunAbraham Phase 1b 1/8 test pattern. The plan locks 19
+    sub-tests covering: default behavior, hc1 bit-equality vs prior bake-Q
+    pattern, hc2_bm functional check, six reject paths, surface contract
+    (get_params/set_params/results.vcov_type), clone idempotency, and the
+    replicate-refit closure smoke.
+    """
+
+    def test_default_vcov_type_is_hc1(self):
+        assert StackedDiD().vcov_type == "hc1"
+
+    def test_default_bit_equal_to_explicit_hc1(self, staggered_data):
+        """Default fit and explicit vcov_type='hc1' fit produce identical SE."""
+        est_default = StackedDiD(kappa_pre=2, kappa_post=2)
+        est_explicit = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1")
+        kwargs = dict(outcome="outcome", unit="unit", time="period", first_treat="first_treat")
+        res_default = est_default.fit(staggered_data, **kwargs)
+        res_explicit = est_explicit.fit(staggered_data, **kwargs)
+        np.testing.assert_allclose(res_default.overall_se, res_explicit.overall_se, atol=1e-15)
+        np.testing.assert_allclose(res_default.overall_att, res_explicit.overall_att, atol=1e-15)
+
+    def test_hc1_se_bit_equal_to_pre_pr_baseline(self, baseline_panel):
+        """HC1 SE matches the captured pre-PR baseline at machine precision.
+
+        Baseline captured on commit 955aa4be (PR #475 clubSandwich merge),
+        BEFORE switching from bake-Q-into-X to explicit weights= pattern in
+        the StackedDiD solve_ols call. Locks the WLS-CR1 invariance claim
+        (HC1 score is identical between the two algebraic forms; only
+        multiplication ordering differs at ULP scale).
+
+        Tolerance: atol=1e-13. The plan originally targeted atol=1e-14 but
+        empirically the bake-w vs explicit-weights paths drift by ~2 ULPs at
+        SE scale due to NumPy internal multiplication ordering — well within
+        the "no methodologically significant drift" band.
+
+        Panel descriptor (regenerate if test fails):
+            generate_staggered_data(n_units=50, n_periods=8, cohort_periods=[3, 5, 7],
+                                    never_treated_frac=0.3, treatment_effect=2.0,
+                                    dynamic_effects=False, seed=20260521)
+            StackedDiD(kappa_pre=2, kappa_post=2)  # defaults otherwise
+        """
+        BASELINE_OVERALL_ATT = 2.08078331612939
+        BASELINE_OVERALL_SE = 0.15699149429146309
+        est = StackedDiD(kappa_pre=2, kappa_post=2)  # default hc1
+        res = est.fit(
+            baseline_panel, outcome="outcome", unit="unit", time="period", first_treat="first_treat"
+        )
+        np.testing.assert_allclose(
+            res.overall_att,
+            BASELINE_OVERALL_ATT,
+            atol=1e-13,
+            err_msg="HC1 overall_att drifted from pre-PR baseline",
+        )
+        np.testing.assert_allclose(
+            res.overall_se,
+            BASELINE_OVERALL_SE,
+            atol=1e-13,
+            err_msg="HC1 overall_se drifted from pre-PR baseline",
+        )
+
+    def test_hc2_bm_finite_and_att_identical_to_hc1(self, staggered_data):
+        """hc2_bm produces finite SE; ATT identical to hc1 (only the variance
+        family changes, not the point estimate). The relationship hc2_bm SE
+        vs hc1 SE depends on cluster leverage and is NOT monotonic — both
+        smaller and larger values are valid depending on the design."""
+        kwargs = dict(
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        res_hc1 = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1").fit(
+            staggered_data, **kwargs
+        )
+        res_hc2bm = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm").fit(
+            staggered_data, **kwargs
+        )
+        np.testing.assert_allclose(
+            res_hc1.overall_att,
+            res_hc2bm.overall_att,
+            atol=1e-13,
+            err_msg="ATT must be identical across vcov_type",
+        )
+        assert np.isfinite(res_hc2bm.overall_se) and res_hc2bm.overall_se > 0
+        # Per-event-time SE also all finite under hc2_bm
+        for h, eff in res_hc2bm.event_study_effects.items():
+            if h == -1:
+                continue  # reference period (SE=0 by construction)
+            assert np.isfinite(eff["se"]) and eff["se"] > 0, f"event_time {h} hc2_bm SE not finite"
+        # And hc2_bm event-study SE differs from hc1 event-study SE on at least one
+        # event-time (proves the leverage/DOF adjustment actually fired).
+        hc1_es = {h: eff["se"] for h, eff in res_hc1.event_study_effects.items() if h != -1}
+        hc2_es = {h: eff["se"] for h, eff in res_hc2bm.event_study_effects.items() if h != -1}
+        diffs = [abs(hc1_es[h] - hc2_es[h]) for h in hc1_es]
+        assert (
+            max(diffs) > 1e-6
+        ), "hc2_bm event-study SEs identical to hc1 — leverage adjustment didn't fire"
+
+    def test_classical_rejected_at_init(self):
+        with pytest.raises(ValueError, match="clusters intrinsically"):
+            StackedDiD(vcov_type="classical")
+
+    def test_hc2_rejected_at_init(self):
+        with pytest.raises(ValueError, match="clusters intrinsically"):
+            StackedDiD(vcov_type="hc2")
+
+    def test_conley_rejected_at_init_with_deferral(self):
+        with pytest.raises(ValueError, match="conley"):
+            StackedDiD(vcov_type="conley")
+
+    def test_invalid_vcov_type_rejected(self):
+        with pytest.raises(ValueError, match="hc1.*hc2_bm|hc2_bm.*hc1"):
+            StackedDiD(vcov_type="hc4")
+
+    def test_survey_design_plus_hc2_bm_rejected(self, staggered_data):
+        """survey_design + non-hc1 vcov_type raises NotImplementedError.
+
+        Reject order locked: fweight/aweight check fires first (per
+        stacked_did.py:309), then the survey + non-hc1 vcov check.
+        """
+        from diff_diff.survey import SurveyDesign
+
+        # Add a uniform pweight column to satisfy SurveyDesign
+        data = staggered_data.copy()
+        data["w"] = 1.0
+        design = SurveyDesign(weights="w", weight_type="pweight")
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        with pytest.raises(NotImplementedError, match="survey TSL"):
+            est.fit(
+                data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                survey_design=design,
+            )
+
+    def test_survey_design_plus_classical_rejected(self, staggered_data):
+        """The classical reject fires at __init__ (before fit), so this test
+        verifies the symmetric path — that a survey-design fit with the
+        already-rejected classical vcov_type fails on the __init__ guard."""
+        with pytest.raises(ValueError, match="clusters intrinsically"):
+            StackedDiD(vcov_type="classical")
+
+    def test_get_params_includes_vcov_type(self):
+        params = StackedDiD(vcov_type="hc2_bm").get_params()
+        assert "vcov_type" in params
+        assert params["vcov_type"] == "hc2_bm"
+
+    def test_set_params_updates_vcov_type(self):
+        est = StackedDiD(vcov_type="hc1")
+        est.set_params(vcov_type="hc2_bm")
+        assert est.vcov_type == "hc2_bm"
+
+    def test_set_params_revalidates_vcov_type(self):
+        """Per local codex R4 P3: `set_params(vcov_type=...)` must re-validate
+        via the same estimator-level guard as `__init__`, not silently accept
+        any value and fail later in the linalg layer with a different message.
+        """
+        est = StackedDiD(vcov_type="hc1")
+        # All three reject paths must fire from set_params too:
+        with pytest.raises(ValueError, match="clusters intrinsically"):
+            est.set_params(vcov_type="classical")
+        assert est.vcov_type == "hc1"  # estimator state unchanged after reject
+        with pytest.raises(ValueError, match="clusters intrinsically"):
+            est.set_params(vcov_type="hc2")
+        with pytest.raises(ValueError, match="conley"):
+            est.set_params(vcov_type="conley")
+        with pytest.raises(ValueError, match="hc1.*hc2_bm|hc2_bm.*hc1"):
+            est.set_params(vcov_type="hc4")
+        # Valid mutation still works
+        est.set_params(vcov_type="hc2_bm")
+        assert est.vcov_type == "hc2_bm"
+
+    def test_results_carries_vcov_type(self, staggered_data):
+        for vcov in ["hc1", "hc2_bm"]:
+            est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type=vcov)
+            res = est.fit(
+                staggered_data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+            )
+            assert res.vcov_type == vcov
+
+    def test_unit_subexp_cluster_plus_hc2_bm_finite(self, staggered_data):
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", cluster="unit_subexp")
+        res = est.fit(
+            staggered_data, outcome="outcome", unit="unit", time="period", first_treat="first_treat"
+        )
+        assert np.isfinite(res.overall_se) and res.overall_se > 0
+        assert res.vcov_type == "hc2_bm"
+
+    def test_replicate_refit_smoke_with_default_hc1(self, staggered_data):
+        """Replicate-weight survey + default hc1 fits cleanly through the
+        _refit_stacked closure (which now uses explicit weights= per
+        stacked_did.py post-PR). Smoke test only — variance correctness is
+        covered by separate replicate-refit tests."""
+        from diff_diff.survey import SurveyDesign
+
+        rng = np.random.default_rng(2026)
+        data = staggered_data.copy()
+        # Generate 20 replicate weight columns (simulating jackknife/BRR-style)
+        n_units = data["unit"].nunique()
+        rep_w = rng.uniform(0.5, 1.5, size=(n_units, 20))
+        rep_w_cols = [f"rep_w{i}" for i in range(20)]
+        unit_to_rep = {u: rep_w[i] for i, u in enumerate(sorted(data["unit"].unique()))}
+        for j, col in enumerate(rep_w_cols):
+            data[col] = data["unit"].map(lambda u, j=j: unit_to_rep[u][j])
+        data["w"] = 1.0
+        design = SurveyDesign(
+            weights="w",
+            weight_type="pweight",
+            replicate_weights=rep_w_cols,
+            replicate_method="JK1",
+        )
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1")
+        res = est.fit(
+            data,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            survey_design=design,
+        )
+        assert np.isfinite(res.overall_se) and res.overall_se > 0
+
+    def test_fit_clone_idempotent_on_vcov_type(self, staggered_data):
+        """Per `feedback_fit_does_not_mutate_config`: fit, clone the
+        estimator config via get_params/set_params, refit, assert SE
+        bit-equal. Locks that fit() doesn't mutate self.vcov_type."""
+        kwargs = dict(outcome="outcome", unit="unit", time="period", first_treat="first_treat")
+        est_a = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        res_a = est_a.fit(staggered_data, **kwargs)
+        # Clone via get_params
+        est_b = StackedDiD(**est_a.get_params())
+        assert est_a.vcov_type == est_b.vcov_type == "hc2_bm"
+        res_b = est_b.fit(staggered_data, **kwargs)
+        np.testing.assert_allclose(res_a.overall_se, res_b.overall_se, atol=1e-15)
+        # And refitting est_a doesn't mutate its config
+        _ = est_a.fit(staggered_data, **kwargs)
+        assert est_a.vcov_type == "hc2_bm"
+
+    def test_aweight_plus_hc2_bm_rejected_by_stacked_did_level_guard(self, staggered_data):
+        """Per review MEDIUM #1: the existing fweight/aweight reject at
+        stacked_did.py:309 fires BEFORE the new vcov_type=non-hc1 reject.
+        This locks the order so a future refactor swapping the checks would
+        silently change error messaging from 'Q-weight ratio semantics' to
+        'survey TSL' on the same input.
+        """
+        from diff_diff.survey import SurveyDesign
+
+        data = staggered_data.copy()
+        data["w"] = 1.0
+        design = SurveyDesign(weights="w", weight_type="aweight")
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        # Expect the Q-weight semantics error (stacked_did.py:309), NOT the
+        # survey TSL vcov error. The match pattern checks for the Q-weight
+        # phrasing specifically.
+        with pytest.raises(ValueError, match="weight_type='aweight'.*Q-weight"):
+            est.fit(
+                data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                survey_design=design,
+            )
+
+    def test_survey_design_plus_hc2_bm_rejected_unit_subexp_cluster(self, staggered_data):
+        """Per review MEDIUM #2: the survey+non-hc1 reject must fire
+        regardless of cluster level."""
+        from diff_diff.survey import SurveyDesign
+
+        data = staggered_data.copy()
+        data["w"] = 1.0
+        design = SurveyDesign(weights="w", weight_type="pweight")
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", cluster="unit_subexp")
+        with pytest.raises(NotImplementedError, match="survey TSL"):
+            est.fit(
+                data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                survey_design=design,
+            )
+
+    def test_replicate_refit_coef_bit_equal_vs_bake_w_baseline(self, baseline_panel):
+        """Per review Q2: pin the replicate-refit overall_att/SE on default
+        hc1 + replicate-weight survey at atol=1e-13. Catches any float64
+        multiplication-ordering drift introduced by the bake-Q-into-X ->
+        explicit weights= switch in the _refit_stacked closure.
+
+        Baseline panel + estimator config matches
+        test_hc1_se_bit_equal_to_pre_pr_baseline. The replicate weights
+        below are seeded deterministically — same seed must produce same
+        result pre- and post-switch.
+        """
+        from diff_diff.survey import SurveyDesign
+
+        rng = np.random.default_rng(20260521)
+        data = baseline_panel.copy()
+        n_units = data["unit"].nunique()
+        rep_w = rng.uniform(0.5, 1.5, size=(n_units, 16))
+        rep_w_cols = [f"rep_w{i}" for i in range(16)]
+        unit_to_rep = {u: rep_w[i] for i, u in enumerate(sorted(data["unit"].unique()))}
+        for j, col in enumerate(rep_w_cols):
+            data[col] = data["unit"].map(lambda u, j=j: unit_to_rep[u][j])
+        data["w"] = 1.0
+        design = SurveyDesign(
+            weights="w",
+            weight_type="pweight",
+            replicate_weights=rep_w_cols,
+            replicate_method="JK1",
+        )
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1")
+        res = est.fit(
+            data,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            survey_design=design,
+        )
+        # Drift lock: hardcoded baseline values captured on the post-PR
+        # branch with the explicit weights= pattern. If multiplication
+        # ordering ever shifts (e.g., a future linalg refactor changes the
+        # bake-w internals in solve_ols), this test catches it. Coef is
+        # pinned at atol=1e-13 (per-replicate drift typically 1-2 ULPs);
+        # the variance compounds the per-replicate drift through
+        # compute_replicate_refit_variance's squared-difference
+        # aggregation, so SE gets atol=1e-10. Both bands match
+        # similar "no methodologically significant drift" tolerances
+        # used elsewhere in the project.
+        BASELINE_ATT = 2.0807833161293945
+        BASELINE_SE = 0.14697256517699428
+        np.testing.assert_allclose(res.overall_att, BASELINE_ATT, atol=1e-13)
+        np.testing.assert_allclose(res.overall_se, BASELINE_SE, atol=1e-10)
+
+    def test_hc2_bm_uses_t_distribution_not_normal_on_well_conditioned_design(self, staggered_data):
+        """Smoke-test on a well-conditioned design: hc2_bm CIs use t(BM DOF),
+        not normal-theory z=1.96. End-to-end check that BM DOF threading
+        from PR R1 fix is wired through to safe_inference."""
+        from scipy.stats import norm
+
+        z_975 = norm.ppf(0.975)  # ~1.96
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        with warnings.catch_warnings(record=True) as caught:
+            warnings.simplefilter("always")
+            res = est.fit(
+                staggered_data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                aggregate="event_study",
+            )
+        # No fallback warning on a well-conditioned design.
+        fallback_warns = [w for w in caught if "Bell-McCaffrey contrast DOF" in str(w.message)]
+        assert len(fallback_warns) == 0, (
+            f"hc2_bm should not warn on a well-conditioned design; "
+            f"got {len(fallback_warns)} warning(s)"
+        )
+        # CI half-width / SE > z_0.975 for at least one post-treatment effect
+        # AND for the overall ATT (BM DOF inflates the critical value).
+        any_t_dist_used = False
+        for h in [0, 1, 2]:
+            if h in res.event_study_effects:
+                es = res.event_study_effects[h]
+                if es["n_obs"] == 0 or es["se"] == 0:
+                    continue
+                half_width = es["conf_int"][1] - es["effect"]
+                t_crit = half_width / es["se"]
+                if t_crit > z_975 * 1.0001:
+                    any_t_dist_used = True
+                    break
+        assert any_t_dist_used, (
+            "hc2_bm CIs should use t(BM DOF), not normal — at least one "
+            "post-treatment event_study CI half-width must exceed z_0.975 * SE"
+        )
+        overall_half_width = res.overall_conf_int[1] - res.overall_att
+        overall_t_crit = overall_half_width / res.overall_se
+        assert overall_t_crit > z_975 * 1.0001
+
+    def test_hc2_bm_rank_deficient_keeps_bm_dof_on_identified_contrasts(
+        self, staggered_data, monkeypatch
+    ):
+        """Per local codex R2 P1 (and R5 P3 test-coverage clarification):
+        when solve_ols's rank-deficient handler drops a column (returns
+        NaN in coef[j] for that column), the SURVIVING delta_h coefficients
+        should still get BM DOF (t-distribution). The dropped coefficient
+        gets NaN inference; identified ones keep finite t-distribution CIs.
+
+        Forces rank deficiency by monkeypatching solve_ols to return a coef
+        vector with one NaN entry at a non-target column (specifically: the
+        intercept column, which is NOT one of the event-study delta_h
+        targets). This exercises the reduced-design code path at
+        stacked_did.py:529-577 added in the R2 P1 fix.
+
+        Verifies:
+          1. The fit doesn't emit the fallback warning (BM DOF computed
+             successfully on the reduced design).
+          2. event_study_effects[h] for identified delta_h have finite
+             inference using t-distribution (CI half-width > z_0.975 * SE).
+          3. The dropped coefficient (intercept) gets NaN inference IF the
+             estimator surfaces it — for StackedDiD it doesn't, since only
+             delta_h coefficients are exposed via event_study_effects.
+        """
+        import importlib
+
+        from scipy.stats import norm
+
+        # Use importlib to get the MODULE (not the same-name function exported
+        # at the diff_diff package level).
+        sd_module = importlib.import_module("diff_diff.stacked_did")
+
+        z_975 = norm.ppf(0.975)
+
+        # Wrap solve_ols to inject NaN at the intercept column (index 0).
+        # The intercept is NOT a target delta_h, so dropping it shouldn't
+        # affect event-study identification — just exercises the
+        # rank-deficient code path.
+        _orig_solve_ols = sd_module.solve_ols
+
+        def _fake_solve_ols(*args, **kwargs):
+            coef, residuals, vcov = _orig_solve_ols(*args, **kwargs)
+            # Inject NaN at intercept (column 0) to simulate rank deficiency
+            coef_modified = coef.copy()
+            coef_modified[0] = np.nan
+            # Also NaN-out vcov row/col 0 to match what _expand_vcov_with_nan
+            # would produce for a dropped column.
+            if vcov is not None:
+                vcov_modified = vcov.copy()
+                vcov_modified[0, :] = np.nan
+                vcov_modified[:, 0] = np.nan
+            else:
+                vcov_modified = vcov
+            return coef_modified, residuals, vcov_modified
+
+        monkeypatch.setattr(sd_module, "solve_ols", _fake_solve_ols)
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        with warnings.catch_warnings(record=True) as caught:
+            warnings.simplefilter("always")
+            res = est.fit(
+                staggered_data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                aggregate="event_study",
+            )
+        # The reduced-design code path should fire without emitting the
+        # fallback warning (the helper succeeds on the kept-column subset).
+        fallback_warns = [w for w in caught if "Bell-McCaffrey contrast DOF" in str(w.message)]
+        assert len(fallback_warns) == 0, (
+            f"Rank-deficient handling should subset to kept columns and "
+            f"succeed; got {len(fallback_warns)} fallback warning(s): "
+            f"{[str(w.message) for w in fallback_warns]}"
+        )
+        # All event-study delta_h coefficients (none dropped here) should
+        # have finite t-distribution inference.
+        for h, eff in res.event_study_effects.items():
+            if h == -1:
+                continue
+            if eff["n_obs"] == 0 or eff["se"] == 0:
+                continue
+            assert np.isfinite(eff["effect"]), f"delta_{h} effect should survive"
+            assert np.isfinite(eff["se"])
+            assert np.isfinite(eff["t_stat"]), f"delta_{h} t_stat should be finite (t-dist)"
+            half_width = eff["conf_int"][1] - eff["effect"]
+            t_crit = half_width / eff["se"]
+            assert t_crit > z_975 * 1.0001, (
+                f"delta_{h} CI must use t(BM DOF) on rank-deficient design; " f"got t_crit={t_crit}"
+            )
+        # Overall ATT also keeps t-distribution
+        if np.isfinite(res.overall_se) and res.overall_se > 0:
+            overall_t_crit = (res.overall_conf_int[1] - res.overall_att) / res.overall_se
+            assert overall_t_crit > z_975 * 1.0001
+
+    def test_hc2_bm_nan_dof_fails_closed_with_all_nan_inference(self, staggered_data, monkeypatch):
+        """Per local codex R3 P1: when BM contrast DOF returns NaN (noise-
+        floor guard fires from PR #475), StackedDiD must emit all-NaN
+        inference fields on the affected contrast — NOT mixed finite-t /
+        NaN-p (which would happen if NaN df flowed through safe_inference)
+        and NOT normal-theory fallback (which would silently produce wrong
+        small-sample CIs).
+
+        Forces NaN DOF by monkeypatching `_compute_cr2_bm_contrast_dof` to
+        return a NaN-only vector. Verifies that:
+          - All event_study_effects entries on the hc2_bm path have NaN
+            t_stat, p_value, conf_int.
+          - overall_* fields are all NaN.
+          - effect and se themselves remain finite (only inference is
+            suppressed, mirroring the LinearRegression.get_inference
+            pattern from PR #475 R7).
+        """
+
+        def _fake_contrast_dof(*args, **kwargs):
+            # Match the m-dimensional output shape expected by callers.
+            n_contrasts = args[3].shape[1] if len(args) >= 4 else kwargs["contrasts"].shape[1]
+            return np.full(n_contrasts, np.nan)
+
+        # Patch the import inside stacked_did's module namespace + the
+        # source module (in case of dynamic re-import).
+        from diff_diff import linalg as _linalg_mod
+
+        monkeypatch.setattr(_linalg_mod, "_compute_cr2_bm_contrast_dof", _fake_contrast_dof)
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        res = est.fit(
+            staggered_data,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        # All event_study_effects (except the ref period) should have NaN
+        # inference but finite effect+se.
+        for h, eff in res.event_study_effects.items():
+            if h == -1:
+                continue  # ref period: SE=0 by construction
+            assert np.isfinite(eff["effect"]), f"effect at h={h} should be finite"
+            assert np.isfinite(eff["se"]) and eff["se"] > 0, f"se at h={h} should be finite > 0"
+            assert np.isnan(eff["t_stat"]), (
+                f"t_stat at h={h} must be NaN when BM DOF NaN-guarded; "
+                f"got {eff['t_stat']} (silent wrong inference)"
+            )
+            assert np.isnan(eff["p_value"]), f"p_value at h={h} must be NaN; got {eff['p_value']}"
+            assert all(
+                np.isnan(b) for b in eff["conf_int"]
+            ), f"conf_int at h={h} must be all-NaN; got {eff['conf_int']}"
+        # Overall ATT: same fail-closed expectation
+        assert np.isfinite(res.overall_att) and np.isfinite(res.overall_se)
+        assert np.isnan(
+            res.overall_t_stat
+        ), f"overall_t_stat must be NaN under NaN DOF; got {res.overall_t_stat}"
+        assert np.isnan(res.overall_p_value)
+        assert all(np.isnan(b) for b in res.overall_conf_int)
+
+    def test_hc2_bm_helper_raises_fails_closed_with_all_nan_inference(
+        self, staggered_data, monkeypatch
+    ):
+        """Per local codex R3 P1: when `_compute_cr2_bm_contrast_dof` raises
+        (e.g., genuine singularity on the identified design), the estimator
+        must emit all-NaN inference rather than fall back to normal-theory
+        CIs/p-values."""
+        from diff_diff import linalg as _linalg_mod
+
+        def _fake_raises(*args, **kwargs):
+            raise np.linalg.LinAlgError("forced linalg failure for test")
+
+        monkeypatch.setattr(_linalg_mod, "_compute_cr2_bm_contrast_dof", _fake_raises)
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm")
+        with warnings.catch_warnings(record=True) as caught:
+            warnings.simplefilter("always")
+            res = est.fit(
+                staggered_data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                aggregate="event_study",
+            )
+        # Warning fired (informational that DOF was unavailable)
+        warning_msgs = [str(w.message) for w in caught]
+        assert any(
+            "Bell-McCaffrey contrast DOF" in m for m in warning_msgs
+        ), f"Should warn on helper failure; got: {warning_msgs}"
+        # Inference NaN-closed for all event-study + overall
+        for h, eff in res.event_study_effects.items():
+            if h == -1:
+                continue
+            assert np.isnan(eff["t_stat"]) and np.isnan(
+                eff["p_value"]
+            ), f"event_time h={h} inference must NaN-close, not fall back to normal"
+        assert np.isnan(res.overall_t_stat), "overall_t_stat must NaN-close on helper failure"
+        assert np.isnan(res.overall_p_value)
+
+    def test_anticipation_plus_hc2_bm_threads_bm_dof(self, staggered_data):
+        """Per local codex R7 P1: the BM-DOF code path uses anticipation in
+        post-period selection (`h >= -self.anticipation` at stacked_did.py:
+        546-548 and 754-758). Verify hc2_bm + anticipation=1 still threads
+        BM DOF into both event-study and overall inference (CI half-width >
+        z_0.975 * SE), the shifted reference period is honored, and the
+        overall ATT contrast includes h=-1 (per anticipation extension)."""
+        from scipy.stats import norm
+
+        z_975 = norm.ppf(0.975)
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", anticipation=1)
+        with warnings.catch_warnings(record=True) as caught:
+            warnings.simplefilter("always")
+            res = est.fit(
+                staggered_data,
+                outcome="outcome",
+                unit="unit",
+                time="period",
+                first_treat="first_treat",
+                aggregate="event_study",
+            )
+        fallback_warns = [w for w in caught if "Bell-McCaffrey contrast DOF" in str(w.message)]
+        assert (
+            len(fallback_warns) == 0
+        ), f"hc2_bm + anticipation should not warn; got {len(fallback_warns)}"
+        # Reference period under anticipation=1 shifts to e=-2 (=-1-1)
+        # so event_study_effects[-2] should have SE=0 (reference).
+        assert -2 in res.event_study_effects
+        assert res.event_study_effects[-2]["se"] == 0.0, (
+            f"Reference period should shift to -2 under anticipation=1; "
+            f"got SE={res.event_study_effects[-2]['se']}"
+        )
+        # Overall ATT: includes h=-1 per anticipation extension; verify
+        # the overall CI uses t(BM DOF), not normal.
+        if np.isfinite(res.overall_se) and res.overall_se > 0:
+            overall_t_crit = (res.overall_conf_int[1] - res.overall_att) / res.overall_se
+            assert overall_t_crit > z_975 * 1.0001, (
+                f"Overall ATT CI under anticipation=1 must use t(BM DOF); "
+                f"got t_crit={overall_t_crit}"
+            )
+        # At least one event-study CI uses t-distribution
+        any_t = False
+        for h, eff in res.event_study_effects.items():
+            if h == -2 or eff["n_obs"] == 0 or eff["se"] == 0:
+                continue
+            tc = (eff["conf_int"][1] - eff["effect"]) / eff["se"]
+            if tc > z_975 * 1.0001:
+                any_t = True
+                break
+        assert any_t, "hc2_bm + anticipation=1 event-study CIs should use t(BM DOF)"
+
+    def test_population_weighting_plus_hc2_bm_finite_threads_bm_dof(self, staggered_data):
+        """Per local codex R7 P1: weighting='population' uses a different
+        Q-weight formula (`Q_sa = (Pop^D_a / Pop^D) / (N^C_a / N^C)` for
+        controls per StackedDiD registry); verify that hc2_bm still threads
+        BM DOF correctly when composed_weights changes formula."""
+        from scipy.stats import norm
+
+        z_975 = norm.ppf(0.975)
+
+        # Need a population column. Add one (constant for simplicity).
+        data = staggered_data.copy()
+        data["pop"] = 1.0  # Equal populations ⇒ same as aggregate weighting
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", weighting="population")
+        res = est.fit(
+            data,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            population="pop",
+            aggregate="event_study",
+        )
+        assert np.isfinite(res.overall_att) and np.isfinite(res.overall_se)
+        assert res.overall_se > 0
+        # Verify BM DOF threaded (t-distribution used)
+        if np.isfinite(res.overall_se) and res.overall_se > 0:
+            overall_t_crit = (res.overall_conf_int[1] - res.overall_att) / res.overall_se
+            assert (
+                overall_t_crit > z_975 * 1.0001
+            ), "hc2_bm + weighting='population' overall CI must use t(BM DOF)"
+
+    def test_sample_share_weighting_plus_hc2_bm_finite_threads_bm_dof(self, staggered_data):
+        """Per local codex R7 P1: weighting='sample_share' uses a third
+        Q-weight formula (`Q_sa = ((N^D_a + N^C_a) / (N^D + N^C)) / (N^C_a / N^C)`
+        for controls per StackedDiD registry); verify hc2_bm threads BM DOF."""
+        from scipy.stats import norm
+
+        z_975 = norm.ppf(0.975)
+
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", weighting="sample_share")
+        res = est.fit(
+            staggered_data,
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        assert np.isfinite(res.overall_att) and np.isfinite(res.overall_se)
+        assert res.overall_se > 0
+        if np.isfinite(res.overall_se) and res.overall_se > 0:
+            overall_t_crit = (res.overall_conf_int[1] - res.overall_att) / res.overall_se
+            assert (
+                overall_t_crit > z_975 * 1.0001
+            ), "hc2_bm + weighting='sample_share' overall CI must use t(BM DOF)"
+
+    def test_hc1_vs_hc2_bm_differ_under_anticipation(self, staggered_data):
+        """Sanity check: hc1 and hc2_bm produce different SEs under
+        anticipation=1 (proves the BM-DOF logic actually fires under
+        non-default params, not just defaults)."""
+        kwargs = dict(
+            outcome="outcome",
+            unit="unit",
+            time="period",
+            first_treat="first_treat",
+            aggregate="event_study",
+        )
+        res_hc1 = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc1", anticipation=1).fit(
+            staggered_data, **kwargs
+        )
+        res_hc2bm = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", anticipation=1).fit(
+            staggered_data, **kwargs
+        )
+        # ATT identical (vcov-only change); SE differs on event-study
+        np.testing.assert_allclose(res_hc1.overall_att, res_hc2bm.overall_att, atol=1e-13)
+        hc1_es = {
+            h: e["se"] for h, e in res_hc1.event_study_effects.items() if h != -2 and e["se"] > 0
+        }
+        hc2_es = {
+            h: e["se"] for h, e in res_hc2bm.event_study_effects.items() if h != -2 and e["se"] > 0
+        }
+        diffs = [abs(hc1_es[h] - hc2_es[h]) for h in hc1_es if h in hc2_es]
+        assert max(diffs) > 1e-6, (
+            "Under anticipation=1, hc1 vs hc2_bm SEs must differ — "
+            "leverage/DOF adjustment didn't fire"
+        )
+
+    def test_summary_renders_clustered_variance_label(self, staggered_data):
+        """Per CI codex R2 P2: summary() must render the clustered CR1/CR2-BM
+        label ('CR1 cluster-robust at unit, G=N' / 'CR2 Bell-McCaffrey
+        cluster-robust at unit, G=N'), NOT the one-way label ('HC1
+        heteroskedasticity-robust' / 'HC2 + Bell-McCaffrey DOF (one-way)').
+        StackedDiD is intrinsically clustered."""
+        kwargs = dict(outcome="outcome", unit="unit", time="period", first_treat="first_treat")
+        for vcov, expected_substr in [
+            ("hc1", "CR1 cluster-robust at unit"),
+            ("hc2_bm", "CR2 Bell-McCaffrey cluster-robust at unit"),
+        ]:
+            est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type=vcov, cluster="unit")
+            res = est.fit(staggered_data, **kwargs)
+            s = res.summary()
+            assert "Variance:" in s, f"summary() missing Variance: line for {vcov}"
+            variance_lines = [line for line in s.split("\n") if "Variance:" in line]
+            assert any(expected_substr in line for line in variance_lines), (
+                f"summary() variance label for vcov_type={vcov} should contain "
+                f"'{expected_substr}', got: {variance_lines}"
+            )
+            # MUST NOT render as one-way (would mislead users)
+            assert not any(
+                "(one-way" in line for line in variance_lines
+            ), f"vcov_type={vcov} mislabeled as one-way: {variance_lines}"
+            # G count present
+            assert any(
+                "G=" in line for line in variance_lines
+            ), f"summary() should include cluster count G= for {vcov}: {variance_lines}"
+        # Also test unit_subexp cluster level
+        est = StackedDiD(kappa_pre=2, kappa_post=2, vcov_type="hc2_bm", cluster="unit_subexp")
+        res = est.fit(staggered_data, **kwargs)
+        s = res.summary()
+        variance_lines = [line for line in s.split("\n") if "Variance:" in line]
+        assert any(
+            "at unit_subexp" in line for line in variance_lines
+        ), f"cluster='unit_subexp' should render in label: {variance_lines}"