This PR outputs a map of the chi-values chi = sqrt(chi2) = |image - model|/sigma for each pixel as FITS extension 6 (name CHI).
This is useful as it is a statistically meaningful "relative error" map. Because each pixel follows a chi distribution with one degree of freedom, plotting the chi map with a scale [0, 2.57585] for example shows where residual pixels are excluded at the 99% confidence level.
This PR outputs a map of the chi-values
chi = sqrt(chi2) = |image - model|/sigma
for each pixel as FITS extension 6 (nameCHI
).This is useful as it is a statistically meaningful "relative error" map. Because each pixel follows a chi distribution with one degree of freedom, plotting the chi map with a scale [0, 2.57585] for example shows where residual pixels are excluded at the 99% confidence level.