By J.McTiernan, 25-June-1998
This is a comparison of the IDL and FORTRAN versions of the new HXT MEM imaging routines, with the Fortran version run using HXTPRO_NMP, and the IDL version using HXT_MULTIMG. The purpose here is to show that translating the routines into IDL does not have a significant effect on the reconstructed images.
There are two major differences between these and the previous versions: (1) new modulation parameters have been established via on-board calibration of HXT by Dr. Jun Sato as part of his Ph.D. thesis study. (2) The total number of counts in the image is now a variable, instead of a constant as in the original versions. The images are significantly improved using the new versions, as detailed in Sato-san's thesis.
Here we compare the IDL and FORTRAN versions for two flares. The results from the two versions are not identical, but are close. The differences result from a different ordering of the calculations, and, in particular, are mostly due to round-off errors in the calculation of the modulation pattern matrix. In each figure, the left-hand image is that obtained using IDL, the right hand image is the FORTRAN version. The scales on each side are counts/SubCollimator. The iteration parameters are at the top; lmda refers to a Lagrange multiplier obtained in the image recontructed images, it is a measure of image contrast. Ch2 is reduced Chi-squared, and Bt is the total brightness in the image in Counts/SC.
Flare of 2-Nov-1991, 06:45:44.5 to 06:45:48
Flare of 13-Nov-1992 (The Masuda Flare), 17:27:00 to 17:27:40
Here is the IDL command used to generate the images for the 2-nov-91 flare, sindex is the SXT index, for Be119 images, used for obtaining the HXT coordinates, index and data are the hxt index and data. All of the stuff that is passed in as keywords can be set interactively, of course.
hxt_multimg, index, data, index1, data1, p, xy = sindex(0), dt_r = -1, bck_tr = ['2-nov-91 07:15', '2-nov-91 07:16:36'], img_tr = ['2-nov-91 06:45:44.5', '2-nov-91 06:45:47.5'], outfile = 'hxi911102.test'
Comments to: jimm@ssl.berkeley.edu