
Delivered-To: oxborrow@dsri.dk
Date: Wed, 27 Mar 2002 18:45:27 +0100
From: Peter Kretschmar <peter.kretschmar@obs.unige.ch>
Subject: Automatic background handling - meeting results
To: Sami Maisala <Sami.Maisala@astro.Helsinki.fi>, Carol Anne Oxborrow <oxborrow@dsri.dk>, Juhani Huovelin <Juhani.Huovelin@Helsinki.fi>, Niels Jrgen Westergaard <njw@dsri.dk>, Stefan Larsson <stefan@astro.su.se>, Sami Maisala <Sami.Maisala@astro.Helsinki.fi>, Silvia Martinez <silvia@dsri.dk>
MIME-version: 1.0
Content-transfer-encoding: 8BIT
X-Accept-Language: de,fr
X-Comment: This message was scanned against viruses by mbx.unige.ch.


  Dear Sami and others,

  here's my attempt to write up my recollection of what we decided
upon, while the memory is still halfway fresh. 

  Sami - please do convert this into updates of your ADD descriptions 
as soon as possible (_before_ changing the code :-) ) and circulate
those for internal review.

  Stefan, Niels Jrgen - please read this and correct any mistakes
or misleading descriptions you may find.

  Hope you had a good flight back and will enjoy the days of Easter!
  Peter


* Spatial Model selection:  j_bkg_spat_sel + JMXi-SPAT-BKG/-DBG

  - Instrumental background 

    The overall Index JMXi-SPAT-BKG-IDX points to a list of
    individual models of normalized background shape shadowgrams.

    These models come in "blocks", where each "block" [feel free to 
    invent a better term] has a selection of shapes for a range of 
    values of _one_ decisive parameter, like orbital phase, sun aspect 
    angle, ... The selection keywords for the other parameters are
    set to appropriate "disregard" values. This scheme reduces the
    storage requirements compared to a multidimensional array of
    models where for each orbital phase we have all solar angles, etc.

    For each block j_bkg_spat_sel selects the two models closest
    to the corresponding parameter of the observation and makes a
    suitable weighed interpolation between them. The results are
    multiplied with each other, to create the overall normalized
    bkg shape shadowgram. It is TBD how the information which
    blocks to use is handed to the program (parameter?)

    Stefan commented that it could be useful to foresee further
    shadowgram manipulations before the models are added & multiplied.


  - Diffuse background

    The Index JMXi-SPAT-DBG-IDX contains (for a start) a _single_
    entry with a typical shadowgram for diffuse emission far off the
    Galaxy. 

    This shadowgram is selected (per default, but a real selection
    should be used to allow later extension) and then _manipulated_
    to produce the output JMXi-SPAT-DBG. Currently, manipulations
    include tilting the shadowgram and rotating it (two functions
    should be included in j_bkg_spat_sel to do this). The rotation
    is defined by the INTEGRAL roll angle. The tilt parameters
    (tilt strength, slope and direction) are read from a _new_ 
    datastructure, tentatively called JMXi-XTRA-DBG [Propose a better
    name and win one piece/bar/bag of the favorite snack of your 
    choice!] which contains the following information per row:

    gal. longitude, gal. latitude, tilt parameters, DGB spec. index

    j_bkg_spat_sel selects the row where l and b are closest to
    the pointing direction.


* Spectral Index map:  j_bkg_spec_sel + JMXi-SIDX-* + JMXi-SCAT-*

  - Instrumental background 

    Instead of trying to calculate JMXi-SIDX-BKG within j_bkg_spec_sel
    we have an Index of JMXi-SIDX-BKG, which is filled offline.
    The matching map is then selected, based on HK & AUX data similar
    to spatial models. 

    NOTE 1: Since it makes no real sense to interpolate spectrum index
    numbers, or to multiply them, the JMXi-SIDX-BKG must be created
    as truly N-dimensional array if we have N relevant parameters.
    The assumption is that we would have few such parameters (one could 
    well start out with a single map) and that the number of models per 
    parameter would not be large.

    NOTE 2: A given file with an JMXi-SIDX-BKG-IDX and children
    must come with a _matching_ JMXi-SCAT-BKG to make any sense.

    A first such map plus the corresponding JMXi-SCAT-BKG can and
    should be created based on the Soton simulations.

  - Diffuse background

    The JMXi-SIDX-DBG will have the _same_ spectrum index value for 
    the whole detector area. The spectrum index to choose is taken
    from JMXi-XTRA-DBG using, again, the row where l and b are closest
    to the pointing direction.


* Scale factors

  No actual change. It was pointed out that the factor to multiply
  with ACCEPTED_EVENTS should be <<1 normally.
