
Date: Thu, 22 Feb 2001 16:53:21 +0100
From: Peter Kretschmar <peter.kretschmar@obs.unige.ch>
Subject: Re: parameters
To: Stefan Larsson <stefan@astro.su.se>
Cc: allan@dsri.dk, njw@dsri.dk, smaisala@astro.helsinki.fi, oxborrow@dsri.dk, stefan@des.astro.su.se
MIME-version: 1.0
Content-transfer-encoding: 7bit
X-Accept-Language: de,fr

Stefan Larsson wrote:
> 
> When you have contemplated all of it, I would like your help to
> "straighten out" the ?-marks on data formats. OBTime-format can
> not be read through par-files as far as I understand(?).

  It could be, but it would very akward to use, as OBTime is in units
of 2^(-20) (~ 1 microsec) counted from 1.1.2000 (plus corrections).
So typical OBTime numbers would look like 65229815808000, too easy to
make a typo, in my opinion.

> When defining timeZero some months ago we did not specify its format.
> 
> If timeZero & timeEnd are given as par-file reals
> (in seconds) with respect to the start of the Science Window
> (OBTSTART in JMX1-FULL-PRP and JMX1-REST-PRP) can that create
> problems? My executables only work on one SW but what about Peter
> and NJW who also use timeZero (or anyone else)?
>
  I would prefer to be able to give an absolute time, not relative
to the start of a ScW or an observation. Therefore my preference would
be to have timeZero in IJD (INTEGRAL Julian Date, which can include
fractions of the day).
 
>[...] 
> 1)  timeZeroSet, timeZero       As before, starting time of first bin.
> 
> 2)  timeEndSet, timeEnd         Similar to timeZero, but end point
>                                 (no data used after this point!)
> 
> 3)  timeStepSet, timeStep       timeStep now also has a Set-parameter.
>                                 (Default: timeStep = integration time
>                                                      for spectra).
> 
> 4)  nPhaseBins, phaseBins       Only used for cyclic folding in phase bins
>                                 of unequal length.
> 
> 
> 1) is as before. 2) is just a generalization of 1)
> 3) Is used to get time resolved or phase folded spectra.
> 4) Is used to specify phase folding bin of unequal sizes. (Implemented later).
> 
> [...]

  To me this sounds slightly confusing, especially the logic for
timeStepSet. How about saying that if nPhasebins > 0 then we do cyclical 
folding with nPhasebins bins and a period of timeStep?

  I would also still prefer if we could do away with time[Zero|End]Set.
I know we have had the discussion before and the problem is that both 0
and negative numbers would be in principle possible. Could we agree on
a large negative value (e.g. -9999) that would be the default and mean
"ignore" or is that to messy?

  Regards, 
  Peter
