Note (verbatim, from paper/manuscript_marine.qmd around line 350, under ## Frequency band, coda window, reference and stacking {#sec:params}):
each of these parameter (frequency band, coda window, reference, and stacking) deserve
their own pararaph and test. always assume that you will quantify the bias and variance
of the recovered \dvv\ given reasonable choices in these parameters.
Context: the section that follows (lines 355-380) currently covers all four parameters (frequency band, reference, stacking; coda window is implied via Fig. demo_5_window_band.png but not discussed in its own paragraph) in a handful of shared paragraphs referencing Fig. fig:params, without a dedicated bias/variance quantification per parameter.
Acceptance criterion: four distinct paragraphs, one per parameter (frequency band, coda window, reference, stacking length), each stating the bias and variance of recovered dv/v under a swept range of "reasonable" choices for that parameter against the known synthetic ground truth — reusing/extending the existing demo_4 through demo_7 synthetic sweeps. Closes when each parameter has its own quantified paragraph rather than sharing prose with the others.
Note (verbatim, from
paper/manuscript_marine.qmdaround line 350, under## Frequency band, coda window, reference and stacking {#sec:params}):Context: the section that follows (lines 355-380) currently covers all four parameters (frequency band, reference, stacking; coda window is implied via Fig.
demo_5_window_band.pngbut not discussed in its own paragraph) in a handful of shared paragraphs referencing Fig.fig:params, without a dedicated bias/variance quantification per parameter.Acceptance criterion: four distinct paragraphs, one per parameter (frequency band, coda window, reference, stacking length), each stating the bias and variance of recovered dv/v under a swept range of "reasonable" choices for that parameter against the known synthetic ground truth — reusing/extending the existing demo_4 through demo_7 synthetic sweeps. Closes when each parameter has its own quantified paragraph rather than sharing prose with the others.