Subject: Re: Normalizing binned fluxes

Hi Peter et.al.,

For the "trivial solution" the used area should maybe be
corrected with the open fraction so that it corresponds to
the area illuminated by a source. After normalization
to 100 cm2 the count rate for an on-axis source would be the
same as in source extraction. For other directions the normalized
flux will be lower by the vignetting factor. 

Regarding Q2 and pixels without detected events. To sum up the
used area correctly one has to take each detector pixel, check 
its status value, look up its corrected coordinates. From
all coordinates compute pixel areas, apply radiusLimit and sum up
(taking status flags into account).

This is what I do except that before the "sum up" I compute the
detector distribution for a particular source direction.
That step you don't do of course.

You can use my routine "j_src_pixel_area" for this. Given the 
corrected pixel coordinates (detCorX, detCorY, which you have to
read in from JMXi-CORX-MOD and JMXi-CORY-MOD) it returns for each 
pixel, its area and lower left coordinate.
You will only need the detPixArea (256x256 matix). 


      status = j_src_pixel_area( detPixArea,
			detPixLLX,
			detPixLLY,
			detCorX,
		        detCorY,
                        radiusLimit,
		        chatty,
		        status);


I'll send it to you with some comments if you think it is useful.

Cheers,

Stefan

> 
>   Dear friends,
> 
>   as you may remember one of the things we want to update for the
> detector spectra and lightcurves is to also normalize them to
> cts / 100 cm^2. Following a more eXtreme approach, this can be
> broken down into two steps, for both I have some questions to
> the team:
> 
> 1. The trivial solution
> 
>    Calculate used area (in cm^2) as circle.
> 
>    For non-imaging data:
>      area = full area  = pi *(12.5cm)^2
> 
>      Q1: shouldn't we have that area as constant somewhere (e.g. jemx.h)?
> 
>    For imaging data:
>      area = pi * radius_limit[cm]^2 
> 
> 
> 2. The elaborate solution:
> 
>    For non-imaging data:
>      As above. No way to correct.
> 
>    For imaging data:
>      Calculate area based on the effective pixel sizes of those pixels 
>      where events matching the GTI/STATUS/... selections could have been 
>      found.
> 
>      Q2: How does one correctly obtain that area?  
>          One cannot just use the events since sometimes there may be
>          pixel positions which could have been hit but weren't for a
>          given dataset.
>          Should we try to define a function calculating the effective
>          area based on radiusLimit and the event selections?
> 
> 
>   Your inputs are very welcome,
>   Peter