From: Hironori Matsumoto <matumoto>
Date: Fri, 21 Dec 2001 16:58:33 -0500 (EST)
In article <200112192327.SAA18962_at_email.domain.hiddendavis writes:

>> I suspect that the treatment of the errors in the data are not the
>> same.  

I found that ISIS assinged the error of 1.866 to zero count
channels, while XSPEC assigned 1.0 to them. 

I corrected the ISIS error values of the zero counts
channels to 1.0 by put_data_counts, and then compered ISIS
and XSPEC again by fitting the same spetrum with the
absorbed brems model. Here are the best-fit values:

               ISIS                    XSPEC
NH(cm^-2)      10^22 (fix)             10^22 (fix)
kT(keV)        10 (fix)                10 (fix)
N(He)/N(H)     0.085 (fix)             0.085 (fix)
norm           4.30e-4                 4.13e-4
data bin       651                     651
chi^2          596.5                   607.86

If I use the XSPEC best-fit value for ISIS,
               ISIS                    XSPEC
norm           4.13e-4                 4.13e-4
chi^2          598.5                   607.86

If I use the ISIS best-fit value for XSPEC,
               ISIS                    XSPEC
norm           4.30e-4                 4.30e-4
chi^2          596.5                   609.92

Thus, there is still a difference in the best-fit chi^2
values. Why? I think the difference of ~10 may be serious. 


Center for Space Research
Massachusetts Institute of Technology
77 Massachusetts Avenue
Cambridge, MA02139-4307, USA
TEL: USA(1)-(617)253-8008
FAX: USA(1)-(617)253-8084
e-mail : matumoto_at_email.domain.hidden==============================================================
Received on Fri Dec 21 2001 - 16:58:33 EST

This archive was generated by hypermail 2.3.0 : Fri May 02 2014 - 08:35:44 EDT