Up to top level
AO15   AO16   AO17   AO18   AO19   Backgrounds   Calibration   Conference   Data   Docs   EPICMOS   EPICpn   Feedback   Gallery   Misc   OM   Pending   PhD_Theses   Publications   RGS   RadMonitor   SAS_Hardware   SAS_WS   SASv16.0   SASv16.0_Installation   SASv16.1   SASv16.1_Installation   SASv17.0   SASv17.0_Installation   SASv18.0   SASv18.0_Installation   SciSim   Simulators_other   Suggestions   Trash   Visibility   XMM-bouncing   XMM-news   XRPS   XSA   esas   incoming  

Logged in as guest

Viewing Misc/44247
Full headers

From: AMRUTA DESHPANDE <amrejd@physics.rutgers.edu>
Subject: OBSID 0112980201
Compose reply
Download message
Move To:
1 replies: 1
0 followups:

Private message: yes  no

Notes:

Notification:


Date: Wed, 23 Jun 2010 23:55:22 -0400 (EDT)
From: AMRUTA DESHPANDE <amrejd@physics.rutgers.edu>
To: xmmhelp@sciops.esa.int
Subject: OBSID 0112980201
HI,

I'm having trouble analyzing data in obsid 0112980201 (specifically the 
bullet cluster).

I want to analyze spectra.  My extractions seem to come out just fine, but 
when I go to fit them, there are major problems.  I am fitting using xspec 
phabs*mekal model.  I give initial guess of 4.8e-2 (Nh), 10 (kT), 1e-2 
(norm) and 0.296 (redshift).  Fit statistics that appear after I input 
these inital guess values are good.  But after applying the fit command, 
the norm drops by a third of the initial guess (while other parameters 
remain near by) and the reduced chi squared goes down to a few tenths, 
while the null hypothesis probability jumps to 1.

Another interesting problem is that I have extracted spectra for 
consecutive regions that are scaled multiples of one another.  As I fit 
data from larger regions, the fits get worse, the count rates decrease, 
and so does luminosity.  The regions increase 10* in size, and luminosity 
decreases by between 5 % and 14 % between consecutive regions.

Here's what I have confirmed so far:
I am using a calibration index file that I downloaded from the cifbuild 
online utility, and have made sure that I have all the calibration files 
necessary for an analysis date of "now"

I have also tried a different analysis date (sometime in the year 2006), 
and extracted arfs and rmfs using that ccf.  But i get teh same bad fit, 
with the bad null hypothesis probability.

I am using a detector map with the same bin size as the images provided by 
pps (80 physical units), which I think is the same spatial binning that 
appears when you view a filtered events list.  My detector map covers the 
source region just fine (although it does not cover the background 
region, which I believe it need not).

I was able to notice that fits improve as I group 2000 channels (minimum), 
but over multiple regions, I still see the opposite trend (of luminosity 
decreasing with increasing region size). (initially my minimum group size 
was 20, which is sufficient for almost every other XMM spectrum I've fit).

I have redownloaded the data, and run my extractions again, and the 
problem persists.

I thank you for your patience in reading my description above.  I am at a 
bit of a loss for ideas at the moment.

Could you please help?

Amruta


Reply 1

Resend
From: Pedro M. Rodriguez Pascual <xmmhelp@sciops.esa.int>
To: amrejd@physics.rutgers.edu
Subject: Re: OBSID 0112980201 (PR#44247)
Date: Thu Jun 24 10:16:45 2010
Hello,

> 
> I'm having trouble analyzing data in obsid 0112980201 (specifically the 
> bullet cluster).
> 
> I want to analyze spectra.  My extractions seem to come out just fine, but

> when I go to fit them, there are major problems.  I am fitting using xspec

> phabs*mekal model.  I give initial guess of 4.8e-2 (Nh), 10 (kT), 1e-2 
> (norm) and 0.296 (redshift).  Fit statistics that appear after I input 
> these inital guess values are good.  But after applying the fit command, 
> the norm drops by a third of the initial guess (while other parameters 
> remain near by) and the reduced chi squared goes down to a few tenths, 
> while the null hypothesis probability jumps to 1.
> 

I am afraid we cannot be of much help in this point since your question seems to
have to do with XSPEC and not with XMM-Newton data processing. However, it is
not surprising that chi2 decreases after the fitting process. Somewhat more
surprising is that a change in normalization of a factor 3 does not produce a
significant change in chi2. I suggest you change the energy range to fit (e.g.,
ignoring 0.0-0.3 and 6.-30.) and check if the same behavior is seen.


> Another interesting problem is that I have extracted spectra for 
> consecutive regions that are scaled multiples of one another.  As I fit 
> data from larger regions, the fits get worse, the count rates decrease, 
> and so does luminosity.  The regions increase 10* in size, and luminosity 
> decreases by between 5 % and 14 % between consecutive regions.
> 

I do not think I fully understand this: if the cluster is ~4arcmin in diameter
and the smallest extraction region already contains the whole cluster, a region
10 times larger is already larger than the EPIC field of view

> Here's what I have confirmed so far:
> I am using a calibration index file that I downloaded from the cifbuild 
> online utility, and have made sure that I have all the calibration files 
> necessary for an analysis date of "now"
> 
> I have also tried a different analysis date (sometime in the year 2006), 
> and extracted arfs and rmfs using that ccf.  But i get teh same bad fit, 
> with the bad null hypothesis probability.
> 
> I am using a detector map with the same bin size as the images provided by

> pps (80 physical units), which I think is the same spatial binning that 
> appears when you view a filtered events list.  My detector map covers the 
> source region just fine (although it does not cover the background 
> region, which I believe it need not).
> 

The detector map should cover the whole field of view of the detector.

> I was able to notice that fits improve as I group 2000 channels (minimum),

> but over multiple regions, I still see the opposite trend (of luminosity 
> decreasing with increasing region size). (initially my minimum group size 
> was 20, which is sufficient for almost every other XMM spectrum I've fit).
> 
> I have redownloaded the data, and run my extractions again, and the 
> problem persists.
> 
> I thank you for your patience in reading my description above.  I am at a 
> bit of a loss for ideas at the moment.

Please, note that no matter how detailed is your explanation of the case, we
need
to have the actual SAS tasks run, with all the parameters used and their output
log to properly understand how the XMM-Newton are processed.

Sincerely.

Pedro

Up to top level
AO15   AO16   AO17   AO18   AO19   Backgrounds   Calibration   Conference   Data   Docs   EPICMOS   EPICpn   Feedback   Gallery   Misc   OM   Pending   PhD_Theses   Publications   RGS   RadMonitor   SAS_Hardware   SAS_WS   SASv16.0   SASv16.0_Installation   SASv16.1   SASv16.1_Installation   SASv17.0   SASv17.0_Installation   SASv18.0   SASv18.0_Installation   SciSim   Simulators_other   Suggestions   Trash   Visibility   XMM-bouncing   XMM-news   XRPS   XSA   esas   incoming  

Logged in as guest


Please make your (short) question the subject of your request!


Web interface using JitterBug ... back to the XMM home page