Yabe-san's notebook produces the extraction residual image (calexp - image simulated from traces and pfsArm) and the residual/calexp image.

98042 r1_subtracted.pdf

98042 r1_subtracted-divided.pdf

Also, the stats (mean, median, and std dev) per 7x7 pix cell are plotted for the residual image.

98042 r1_mean.pdf

98042 r1_median.pdf

98042 r1_std.pdf

I'd like to add a function to calculate the rms of the extraction residual for each fiber and for the whole detector.

Residual/error image. np.sqrt(calexp.variance.array) is used as the error.

97821 b1_chi-2.pdf

The chi^2 and the calexp average of each fiber are plotted. The fibers of bright objects and its adjacent fibers tend to have large chi^2. The example images of the residual/error, calexp, and mask for the fiber of the bright object is attached. The bumpy pattern is created by rearranging the distorted fiber trace into a straight line.

97821 b1_chistats-13.pdf

Thank you, Hamano-san! Can I make a few suggestions?

• In 97821..chi-2.pdf, could you change the color scheme so that positive residual are red and negative residuals are blue?

• In addition to chi^2, I would like to see chi because we are interested in the sign.
• Could we show sigma(chi) for each fiber? We can get sigma from the interquartile range, |Q3-Q1| /1.354, for each fiber. If things behave, we expect sigma~1. Deviations from unity is a bad sign.

Thank you for your comments. I updated the figures.

97821 b1_chi_coolwarm.pdf

97821 b1_chistats_v2.pdf

The data points of different targetType are plotted with different colors.

97821 b1_chistats_v3.pdf