kinextract.correct_recovered_losvd

kinextract.correct_recovered_losvd(v_recovered: float, sigma_recovered: float, bias_table: dict) → dict[source]

Apply an empirical bias correction from an assess_recovery_bias() table.

Interpolates bias_table at (v_recovered, sigma_recovered) (linear interpolation within the grid’s convex hull, nearest-neighbor outside it or if fewer than 4 grid points are available) and returns bias-corrected point estimates plus an inflated uncertainty incorporating the bias table’s own seed-to-seed scatter at that location – an empirically-calibrated alternative to the retired analytic kinextract.errors.bias_corrected_losvd(), which is known to be unreliable near the instrumental resolution limit (see that function’s docstring).

Parameters:

v_recovered (float) – A real fit’s recovered velocity/dispersion (kinextract.losvd.fit_losvd_gauss_hermite()’s vherm/ sherm) to correct.
sigma_recovered (float) – A real fit’s recovered velocity/dispersion (kinextract.losvd.fit_losvd_gauss_hermite()’s vherm/ sherm) to correct.
bias_table (dict) – Output of assess_recovery_bias(), built from mocks matched to the same target/instrument/configuration as the fit being corrected.

Returns:

v_corrected, sigma_corrected (bias-subtracted point estimates) and v_uncertainty_inflation, sigma_uncertainty_inflation (the interpolated seed-to-seed standard deviation of the bias at this location, to be added in quadrature to whatever statistical uncertainty LOSVDErrorEstimator reports).

Return type:

dict