perf: default integer arrays to int32 for ~25% memory reduction

doc/release_notes.rst

@@ -19,6 +19,7 @@ Upcoming Version
 
 *Other*
 
+* Default internal integer labels to ``int32`` (configurable via ``linopy.options["label_dtype"]``, set to ``np.int64`` for the old behavior), cutting memory ~25% and speeding up model build 10-35%. Raises ``ValueError`` if labels exceed the int32 maximum.
 * ``add_variables(binary=True, ...)`` now accepts ``lower``/``upper`` bounds, as long as they are 0 or 1. Previously binary bounds could only be set via the ``.lower``/``.upper`` setters after creation. (https://github.com/PyPSA/linopy/issues/776)
 * ``add_piecewise_formulation`` gained an ``active_fill`` parameter that gates a partial ``active`` (defined over a subset of the indexed dimension, or masked) as always-active (``1``) or always-off (``0``); without it, a partial ``active`` — which was previously zeroed silently — now raises. Useful when one formulation mixes gated and ungated entities (e.g. committable and non-committable units sharing a ``status``). ``active_fill`` is transitional and will be removed once v1 semantics make ``active.reindex(coords).fillna(value)`` sufficient. (https://github.com/PyPSA/linopy/issues/796)
 

linopy/common.py

@@ -8,7 +8,6 @@
 from __future__ import annotations
 
 import operator
-import os
 from collections.abc import Callable, Generator, Hashable, Iterable, Sequence
 from functools import cached_property, reduce, wraps
 from pathlib import Path
@@ -159,12 +158,10 @@ def infer_schema_polars(ds: Dataset) -> dict[str, DataTypeClass]:
         dict: A dictionary mapping column names to their corresponding Polars data types.
     """
     schema: dict[str, DataTypeClass] = {}
-    np_major_version = int(np.__version__.split(".")[0])
-    use_int32 = os.name == "nt" and np_major_version < 2
     for name, array in ds.items():
         name = str(name)
         if np.issubdtype(array.dtype, np.integer):
-            schema[name] = pl.Int32 if use_int32 else pl.Int64
+            schema[name] = pl.Int32 if array.dtype.itemsize <= 4 else pl.Int64
         elif np.issubdtype(array.dtype, np.floating):
             schema[name] = pl.Float64
         elif np.issubdtype(array.dtype, np.bool_):
@@ -308,7 +305,7 @@ def save_join(*dataarrays: DataArray, integer_dtype: bool = False) -> Dataset:
         )
         arrs = xr_align(*dataarrays, join="outer")
         if integer_dtype:
-            arrs = tuple([ds.fillna(-1).astype(int) for ds in arrs])
+            arrs = tuple([ds.fillna(-1).astype(options["label_dtype"]) for ds in arrs])
     return Dataset({ds.name: ds for ds in arrs})
 
 

linopy/config.py

@@ -9,6 +9,10 @@
 
 from typing import Any
 
+import numpy as np
+
+_VALID_LABEL_DTYPES = {np.int32, np.int64}
+
 
 class OptionSettings:
     """Runtime configuration knobs (e.g. display widths). Use as a context manager or set values directly via ``options(key=value)``."""
@@ -30,6 +34,10 @@ def set_value(self, **kwargs: Any) -> None:
         for k, v in kwargs.items():
             if k not in self._defaults:
                 raise KeyError(f"{k} is not a valid setting.")
+            if k == "label_dtype" and v not in _VALID_LABEL_DTYPES:
+                raise ValueError(
+                    f"label_dtype must be one of {_VALID_LABEL_DTYPES}, got {v}"
+                )
             self._current_values[k] = v
 
     def get_value(self, name: str) -> Any:
@@ -62,4 +70,5 @@ def __repr__(self) -> str:
 options = OptionSettings(
     display_max_rows=14,
     display_max_terms=6,
+    label_dtype=np.int32,
 )

linopy/constraints.py

@@ -748,7 +748,7 @@ def _to_dataset(self, nterm: int) -> Dataset:
         # Map active row i -> flat position in full shape via con_labels
         active_positions = self.active_positions
         coeffs_2d = np.zeros((full_size, nterm), dtype=csr.dtype)
-        vars_2d = np.full((full_size, nterm), -1, dtype=np.int64)
+        vars_2d = np.full((full_size, nterm), -1, dtype=options["label_dtype"])
         if csr.nnz > 0:
             row_indices = np.repeat(active_positions, counts)
             term_cols = np.arange(csr.nnz) - np.repeat(csr.indptr[:-1], counts)
@@ -772,7 +772,7 @@ def _to_dataset(self, nterm: int) -> Dataset:
         )
         ds = Dataset({"coeffs": coeffs_da, "vars": vars_da})
         if self._cindex is not None:
-            labels_flat = np.full(full_size, -1, dtype=np.int64)
+            labels_flat = np.full(full_size, -1, dtype=options["label_dtype"])
             labels_flat[active_positions] = self._con_labels
             ds = assign_multiindex_safe(
                 ds,
@@ -2181,7 +2181,10 @@ def flat(self) -> pd.DataFrame:
             return pd.DataFrame(columns=["coeffs", "vars", "labels", "key"])
         df = pd.concat(dfs, ignore_index=True)
         unique_labels = df.labels.unique()
-        map_labels = pd.Series(np.arange(len(unique_labels)), index=unique_labels)
+        map_labels = pd.Series(
+            np.arange(len(unique_labels), dtype=options["label_dtype"]),
+            index=unique_labels,
+        )
         df["key"] = df.labels.map(map_labels)
         return df
 

linopy/expressions.py

@@ -451,7 +451,9 @@ def __init__(self, data: Dataset | Any | None, model: Model) -> None:
             )
 
         if np.issubdtype(data.vars, np.floating):
-            data = assign_multiindex_safe(data, vars=data.vars.fillna(-1).astype(int))
+            data = assign_multiindex_safe(
+                data, vars=data.vars.fillna(-1).astype(options["label_dtype"])
+            )
         if not np.issubdtype(data.coeffs, np.floating):
             data["coeffs"].values = data.coeffs.values.astype(float)
 
@@ -1535,7 +1537,7 @@ def sanitize(self) -> Self:
         linopy.LinearExpression
         """
         if not np.issubdtype(self.vars.dtype, np.integer):
-            return self.assign(vars=self.vars.fillna(-1).astype(int))
+            return self.assign(vars=self.vars.fillna(-1).astype(options["label_dtype"]))
 
         return self
 
@@ -1939,12 +1941,12 @@ def _simplify_row(vars_row: np.ndarray, coeffs_row: np.ndarray) -> np.ndarray:
         # Combined has dimensions (.., CV_DIM, TERM_DIM)
 
         # Drop terms where all vars are -1 (i.e., empty terms across all coordinates)
-        vars = combined.isel({CV_DIM: 0}).astype(int)
+        vars = combined.isel({CV_DIM: 0}).astype(options["label_dtype"])
         non_empty_terms = (vars != -1).any(dim=[d for d in vars.dims if d != TERM_DIM])
         combined = combined.isel({TERM_DIM: non_empty_terms})
 
         # Extract vars and coeffs from the combined result
-        vars = combined.isel({CV_DIM: 0}).astype(int)
+        vars = combined.isel({CV_DIM: 0}).astype(options["label_dtype"])
         coeffs = combined.isel({CV_DIM: 1})
 
         # Create new dataset with simplified data

linopy/model.py

@@ -35,6 +35,7 @@
     replace_by_map,
     to_path,
 )
+from linopy.config import options
 from linopy.constants import (
     GREATER_EQUAL,
     HELPER_DIMS,
@@ -824,7 +825,15 @@ def add_variables(
 
         start = self._xCounter
         end = start + data.labels.size
-        data.labels.values = np.arange(start, end).reshape(data.labels.shape)
+        label_dtype = options["label_dtype"]
+        if end > np.iinfo(label_dtype).max:
+            raise ValueError(
+                f"Number of labels ({end}) exceeds the maximum value for "
+                f"{label_dtype.__name__} ({np.iinfo(label_dtype).max})."
+            )
+        data.labels.values = np.arange(
+            start, end, dtype=options["label_dtype"]
+        ).reshape(data.labels.shape)
         self._xCounter += data.labels.size
 
         if mask is not None:
@@ -969,7 +978,15 @@ def _allocate_constraint_labels(
         """Assign label ranges from the constraint counter and apply an optional mask."""
         start = self._cCounter
         end = start + data.labels.size
-        data.labels.values = np.arange(start, end).reshape(data.labels.shape)
+        label_dtype = options["label_dtype"]
+        if end > np.iinfo(label_dtype).max:
+            raise ValueError(
+                f"Number of labels ({end}) exceeds the maximum value for "
+                f"{label_dtype.__name__} ({np.iinfo(label_dtype).max})."
+            )
+        data.labels.values = np.arange(start, end, dtype=label_dtype).reshape(
+            data.labels.shape
+        )
         self._cCounter += data.labels.size
         if mask is not None:
             data.labels.values = np.where(mask.values, data.labels.values, -1)

linopy/variables.py

@@ -1272,7 +1272,9 @@ def ffill(self, dim: str, limit: None = None) -> Variable:
             .map(DataArray.ffill, dim=dim, limit=limit)
             .fillna(self._fill_value)
         )
-        return self.assign_multiindex_safe(labels=data.labels.astype(int))
+        return self.assign_multiindex_safe(
+            labels=data.labels.astype(options["label_dtype"])
+        )
 
     def bfill(self, dim: str, limit: None = None) -> Variable:
         """
@@ -1299,7 +1301,7 @@ def bfill(self, dim: str, limit: None = None) -> Variable:
             .map(DataArray.bfill, dim=dim, limit=limit)
             .fillna(self._fill_value)
         )
-        return self.assign(labels=data.labels.astype(int))
+        return self.assign(labels=data.labels.astype(options["label_dtype"]))
 
     def sanitize(self) -> Variable:
         """
@@ -1310,7 +1312,9 @@ def sanitize(self) -> Variable:
         linopy.Variable
         """
         if issubdtype(self.labels.dtype, floating):
-            return self.assign(labels=self.labels.fillna(-1).astype(int))
+            return self.assign(
+                labels=self.labels.fillna(-1).astype(options["label_dtype"])
+            )
         return self
 
     def equals(self, other: Variable) -> bool:
@@ -2032,7 +2036,10 @@ def flat(self) -> pd.DataFrame:
         """
         df = pd.concat([self[k].flat for k in self], ignore_index=True)
         unique_labels = df.labels.unique()
-        map_labels = pd.Series(np.arange(len(unique_labels)), index=unique_labels)
+        map_labels = pd.Series(
+            np.arange(len(unique_labels), dtype=options["label_dtype"]),
+            index=unique_labels,
+        )
         df["key"] = df.labels.map(map_labels)
         return df
 

test/test_dtypes.py

@@ -0,0 +1,75 @@
+"""Tests for int32 default label dtype."""
+
+import numpy as np
+import pytest
+
+from linopy import Model
+from linopy.config import options
+
+
+def test_default_label_dtype_is_int32() -> None:
+    assert options["label_dtype"] == np.int32
+
+
+def test_variable_labels_are_int32() -> None:
+    m = Model()
+    x = m.add_variables(lower=0, upper=10, coords=[range(5)], name="x")
+    assert x.labels.dtype == np.int32
+
+
+def test_constraint_labels_are_int32() -> None:
+    m = Model()
+    x = m.add_variables(lower=0, upper=10, coords=[range(5)], name="x")
+    m.add_constraints(x >= 1, name="c")
+    assert m.constraints["c"].labels.dtype == np.int32
+
+
+def test_expression_vars_are_int32() -> None:
+    m = Model()
+    x = m.add_variables(lower=0, upper=10, coords=[range(5)], name="x")
+    expr = 2 * x + 1
+    assert expr.vars.dtype == np.int32
+
+
+@pytest.mark.skipif(
+    not pytest.importorskip("highspy", reason="highspy not installed"),
+    reason="highspy not installed",
+)
+def test_solve_with_int32_labels() -> None:
+    m = Model()
+    x = m.add_variables(lower=0, upper=10, name="x")
+    y = m.add_variables(lower=0, upper=10, name="y")
+    m.add_constraints(x + y <= 15, name="c1")
+    m.add_objective(x + 2 * y, sense="max")
+    m.solve("highs")
+    assert m.objective.value == pytest.approx(25.0)
+
+
+def test_overflow_guard_variables() -> None:
+    m = Model()
+    m._xCounter = np.iinfo(np.int32).max - 1
+    with pytest.raises(ValueError, match="exceeds the maximum"):
+        m.add_variables(lower=0, upper=1, coords=[range(5)], name="x")
+
+
+def test_overflow_guard_constraints() -> None:
+    m = Model()
+    x = m.add_variables(lower=0, upper=1, coords=[range(5)], name="x")
+    m._cCounter = np.iinfo(np.int32).max - 1
+    with pytest.raises(ValueError, match="exceeds the maximum"):
+        m.add_constraints(x >= 0, name="c")
+
+
+def test_label_dtype_option_int64() -> None:
+    with options:
+        options["label_dtype"] = np.int64
+        m = Model()
+        x = m.add_variables(lower=0, upper=10, coords=[range(5)], name="x")
+        assert x.labels.dtype == np.int64
+        expr = 2 * x + 1
+        assert expr.vars.dtype == np.int64
+
+
+def test_label_dtype_rejects_invalid() -> None:
+    with pytest.raises(ValueError, match="label_dtype must be one of"):
+        options["label_dtype"] = np.float64