[MSLICE-tickets] [MSLICE] #21: detector trajectories mex files not working

[MSlice]

#21: detector trajectories mex files not working
-------------------------+----------------
Reporter:  Toby Perring  |     Owner:
    Type:  defect        |    Status:  new
Priority:  major         |  Keywords:
-------------------------+----------------
 Problem with detector trajectories calculation in mslice. This was
 discovered to be due to problem in mex file(s) [but maybe not
 exclusively]. E-mails forwarded to Alex.


 ================================================================
 Dear Toby,

 I have a problem with Mslice that I thought and hope that you might know
 the answer to.  The reason I ask you and not Radu, is that I think he once
 told me
 that you took over the maintenance (I realize of course that Horace is
 higher on your agenda).

 I am looking at CAPCC data from OSIRIS and trying to compare the H>H_sat
 data with the dispersion expected for the parameters obtained by Paul from
 the LET data.
 As a result I think Mslice does not compute the projections on  the
 viewing axes correctly for an indirect TOF spectrometer.

 In the attached figure, you can see that the detector trajectories look
 like those of a direct TOF and not an indirect spectrometer. Moreover, the
 constant energy contours actually
 extend beyond the detector trajectories. There is clearly something fishy
 going on.

 I tried a few sanity/insanity checks:

 •       First, I checked that if I manually extract H and K from the
 control window by projecting first on (H,Energy) and then on (K,Energy)
 and extracting H and K by using the “fromwindow” command in Mslice , I get
 the same – direct TOF – detector trajectories, which I believe to be
 wrong.
 •       I then traced back through the code (ms_calc_proj  calcprojb) to
 find that the H and K’s for an OSIRIS type spectrometer, are obtained by
 the executable mex file spe2proj_df, which I don’t immediately know how to
 change. It looks as if this code is doing something wrong in extracting H
 and K, which in turn means that when I try to overplot the dispersion
 computed for Pauls fit parameters, I get something which looks wrong.
 •       I also checked that the constant energy contours in the attached
 plot are correct, in the sense (at least) that they agree with some Mslice
 detector trajectory plots that I have on my computer from an earlier
 version of Mslice, where the detector trajectories also look different
 (and probably correct).
 •       Looking in ms_plot_traj, I can see that these contours are
 calculated directly, without resorting to spe2proj_df
 •       Spe2proj_df is being used by ms_plot_traj (through the pickvar.m
 file) to compute the momentum transfers H and K.

 In conclusion, it looks as if there is a bug in spe2proj_df.  In
 particular, it looks as if it assumes that the spectrometer is of the
 direct type.

 If I am right that you are officially maintaining Mslice now, then could I
 ask if you can check this?

 Best wishes

 Niels


 ===================================================================
 Dear Toby,

 Ignore my email from before. I just remembered that Radu once told me that
 Mslice does not need mex files any longer. So I removed the DLL folder
 from the path and now the detector trajectories look fine and Paul’s
 parameters seem to produce good agreement (at least with the first data
 set that I tried … let’s not get carried away).

 I still think that there is an error in the mex-file, but at least I seem
 to have found a fix.

 Best wishes

 Niels

-- 
Ticket URL: <http://trac.isis.rl.ac.uk/mslice/ticket/21>
MSlice <http://trac.isis.rl.ac.uk/mslice/>
MSLICE