Project status Last update: 2011/10/17 x = done ------------------------- Hans x draft forward model script & demonstrate test of correction subroutine Tom x draft GOF script & demonstrate test of GOF subroutine Jake x draft correction subroutine script & demonstrate test of forward model Shane ? draft PSF equalizer script & demonstrate test of PSF equalizer Charles x draft main program fake data for a test (512x256 = 1/16 size) script & demonstrate test of main program Looking ahead: main program test using fake data data reduction script to move flight data through the pipeline batch process all flight data how can we validate results?!?! publish ??? enjoy the fame ??? profit Update 3/12/2012: One thing Lewis had noticed, but I nearly forgot, is that there's a gradient in sensitivity across our m=-1 detector. Attached is a plot that shows variation across the CCD (ratio of outboard orders to m=0), along with fits for each trend. That's why we kept getting poor chisquared in that channel. Recent example after 30 iterations, for m=-1, 0, +1: 14.0837 6.33796 3.93608 I've now applied a correction to the data (for now, it is done in moses_invert_reverse.pro, but should eventually go into MOSES_PREP), and propagated that through the inversion code. Chisquared after 30 iterations is now markedly improved in all channels: 3.54095 2.76951 2.50088 Also have introduced keyword Iinf_emphasis to smart2.pro. This allows something other than the 1/e rule of Lewis for setting gamma for the infinite order. I am hoping this will give us greater flexibility to vary from the average line profile (Iinf).