Quick start with the inversion software:
0) download/unzip IDL inversion routines inversion.zip Place all IDL routine so that they are seen by IDL program.
1) SPEX USERS: do thermal plus+broken-power law thin-target fit (f_vth_thin )
(for details see http://hesperia.gsfc.nasa.gov/rhessidatacenter/spectroscopy.html)
OSPEX USERS: start OSPEX defining OSPEX object name
or do thermal plus thin fit inside OSPEX (vth+thin), Press Accept the results and Continue
Currently inversion works for one time interval only.
(How to use OSPEX see http://hesperia.gsfc.nasa.gov/~dennis/OSPEX/The_Basics/index.htm)
1) run invesion program
SPEX> idl, inv_counts2electrons
If you want to find a photon flux replace electrons with photons :
SPEX> idl, inv_counts2photons
3) Look at the residuals you have after the inversion. Do you like the regularised solutions and the residuals? If not adjust reg_tweak parameter. If the resulting regularised solution is still oscillating (and the residuals are too good (CHI2<1)) you need to increase the regularisation parameter, if the residuals are bad (CHI2 >1) the solution is likely to be over-regularised then reduce the parameter. It is wise to decrease/increase reg_tweak bit-by-bit changing by 20 % or so. The default value of reg_tweak is 1.
SPEX users: SPEX> idl, inv_counts2electrons, reg_tweak=0.8
OSPEX users: IDL> inv_counts2electrons, obj=o, reg_tweak=0.8
4) Make sure that the results are correct. Play with different energy ranges in SPEX/OSPEX and regularisation order. Try different parameters in the inversion software. The output of the inversion will be written into your working directory. The code produces (model-independent (regularised) either electron flux <VnF> [electrons/sec/keV/cm^2]) or photon flux [photons/sec/cm^2/keV] saved as IDL save file, plotted into ps files.
; CALLING SEQUENCE:
; OPTIONAL INPUTS:
; version - string, version of the inversion code (currently only one version exist)
; order - integer, regularization order. Can be 0 (default), 1, or 2.
; Npass - integer, number of random noise realisations for the confidence strip.
; Default is 30.
; Guess - integer, 1 or 0 (default) – to use or NOT to use forward-fit solution in the constraint
; reg_tweak - float, parameter to adjust the regularization parameter (default is 1).