Subject: Re: JEM-X misalignment

Dear Carol Anne,

>  we're going to need a new misalignment matrix.

ok, but you'll have to do it yourself this time, don't worry it's not 
complicated.

> Basically: what is the data structure name? Where is it stored? How do 
> you
> create one? What do we need as input data to the creation? How can we 
> ensure
> that the correct misalignment matrix is used with the corresponding 
> version of
> the software? Why isn't it part of the IMOD group?

Thierry and I expected such questions, that's why a chapter of my phd is 
dedicated
to this topic. This chapter is now ready and should appear on the isdc 
doc web
page as a tech note soon. I'm attaching it to this email.

Section 8.9 will answer your questions, see also points 1 and 2 below.

Then to derive the new matrix, I suggest that you first skip sections 
8.2 to 8.5 and read section 8.6 to have an idea of what you'll have to 
do. Then jump to the cookbook, section  8.10.

Here are some advices:

1- if you recreate the aux branch as it is described in section 8.10, 
then you ensure that
the matrix used is the one in this new (aux branch) location. The file 
copied there can
include unity or non-unity matrices

2- when a reasonable matrix is obtained, I mean which improve the 
results with respect
to the previous one, it can be ingested in the archive. The file will 
then receive a time
stamp which ensure that this one ie the latest will be used in the 
future. Note that Roland has to agree with the results before allowing 
the ingestion

3- the very first thing to do is to select a bunch of pointings, from a 
source like grs1915+105 for example, as the Crab is not a point source 
in this X band. I took as much
as 300 pointings.

4- then set an osa environment, recreate the aux branch and copy there 
the last matrix from the archive. Set the jemx1 and jemx2 matrices to 
unity: 1 0 0 0 1 0 0 0 1 and run the osa whith your prefered set of 
parameters. Note that for jemx I was using only pointings with attitude 
stable within 2.5''. You'll need a few days to complete this task. 

5- once this is done, collect the results as stated in the cookbook and 
fill a "sources" file
with the following columns: ra_catalog, dec_cat, ra_measured, dec_mes, 
ra_scx, dec_scx, ra_scz, dec_scz. I attached an example.

6-run misfit.cpp in isdcroot and make sure the matrix produced is copied 
in the fits file.
The misfit.cpp attached is the one used with the files sources attached.

7-test it !!! Note, no need to run on 300 pointings this time. You can 
follow the steps
made in section 8.6.

The best is that you try and just ask your questions as they pop up, 
I'll try to react asap,

hope this answer all your questions !

We'll stay in touch,
Pascal 

>   Best wishes for the successful completion of your thesis,

thanks, it has been submitted now, I'm waiting on the Jury

note: in misfit.cpp, you just have to update "numlines", the name of the
fits file in DALobjectOpen("inst_misalign_20030701.fits[1]",&fitsfile,0);
and nrow according to the jemx unit you are working on. Now the code
starting after the comment "Display" can (and should) be customized 
according to
your need, wishes etc. You start from the code existing there but I cannot
garantee it will work on your data. In the end, I was always using other 
tools
to plot the results as root can be a bit heavy and slow sometimes for this.

-- 
Pascal Favre                       phone:  +41 22 950 91 20 (direct)
INTEGRAL Science Data Centre               +41 22 950 91 00 (switchb.)
Chemin d'Ecogia 16                 fax:    +41 22 950 91 33
CH-1290 Versoix                    e-mail: Pascal.Favre@obs.unige.ch          
Switzerland                        WWW:    http://isdc.unige.ch