slv.inversion.sweep.SweepResults

class slv.inversion.sweep.SweepResults(path)

Wraps sweep_results.csv for post-sweep analysis.

Variables:

• path (pathlib.Path) – Path to the results CSV.

• df (pandas.DataFrame) – All completed rows, sorted by chi2_distance (closest to 1 first).

Examples

results = SweepResults("/scratch/my_sweep/sweep_results.csv")

# Best-fitting configs (chi² ∈ [0.9, 1.1])
results.best(tol=0.1)

# Which values of prior_base_std give a stable posterior flux?
results.stability("cfg_prior_base_std")

# Are results more sensitive to prior or MDM params?
results.sensitivity()

# 1-D scatter: chi² vs a single parameter
results.plot_chi2("cfg_prior_base_std")

# 2-D heatmap: chi² over two parameters
results.plot_heatmap("cfg_prior_base_std", "cfg_prior_std_frac")

# Rehydrate best config for a reproduction run
cfg, cls = results.to_config(results.best().iloc[0])

Methods

best([target, tol])	Return rows where |reduced_chi2 - target| < tol, best first.
failed()	Return rows that errored during the sweep.
plot_chi2(param[, ax, target_band])	Scatter plot of reduced chi² vs a swept parameter.
plot_heatmap(param_x, param_y[, metric, ax])	Heatmap of metric over two swept parameters.
plot_sensitivity([metric, ax])	Horizontal bar chart of parameter sensitivity (range of group means).
reload()	Re-read from disk (useful while a sweep is still running).
sensitivity([metric, target])	Rank swept parameters by their effect on metric.
stability(groupby_params[, metric])	Standard deviation of metric grouped by groupby_params.
to_config(row[, pipeline, cache])	Rehydrate an InversionConfig from a results row.