Logged in as guest
Viewing esas/43944 Full headers
Private message: yes no
Notes: Notification:
Date: Mon, 17 May 2010 14:04:15 GMT From: gaetz@cfa.harvard.edu To: xmmhelp@sciops.esa.int Cc: gaetz@cfa.harvard.edu Subject: SASv10.0, mos_back problem
Full_Name: Terry Gaetz Submission from: (NULL) (131.142.42.182) In running mos_back, I consistently get an error: mos1S023-im1-400-2000.fits RETFIT POSITION 1 STATUS FTGIOU = 9367872 Reading input file mos1S023-im1-400-2000.fits Input array size not equal to image I'm confused as to what is happening here. The file exists and has reasonable dimensions: XMM-SAS:lhea:geri% funhead mos1S023-im1-400-2000.fits | grep NAXIS NAXIS = 2 / number of data axes NAXIS1 = 780 / length of data axis 1 NAXIS2 = 780 / length of data axis 2 I'm using the current HEASOFT 6.9 release, but the same thing happens Looking at the source, the error is coming from "retfit" in esas/subs.f; retfit is called two places in mos_back_sub.f, once for the fwc data, and once for the observation image in detector coordinates. There is an appropriate -obj-im-det-400-2000, so I' SAS:lhea:geri% funhead mos1S023-obj-im-det-400-2000.fits | grep NAXIS NAXIS = 2 / number of data axes NAXIS1 = 780 / length of data axis 1 NAXIS2 = 780 / length of data axis 2 XMM-SAS:lhea:geri% sasversion sasversion:- XMM-Newton SAS release and build information: SAS release: xmmsas_20100423_1801-10.0.0 Compiled on: Fri Apr 23 18:17:14 GMT 2010 Compiled by: sasbuild@sasbld02.net4.lan Platform : RHEL5 32 SAS-related environment variables that are set: SAS_DIR = /home/gaetz/local/pkgs/x86_64-linux_CentOS-5.4/xmmsas SAS_PATH = /home/gaetz/local/pkgs/x86_64-linux_CentOS-5.4/xmmsas SAS_CCFPATH = /home/gaetz/data/caldb/data/xmm/ccf SAS_CCF = /data/geri3/gaetz/data.xmm.n132d/0129341501/odf/ccf.cif SAS_ODF = /data/geri3/gaetz/data.xmm.n132d/0129341501/odf I'm using the current HEASOFT 6.9-release, but the same thing happens with HEASOFT-6.8
From: Pedro M. Rodriguez Pascual <xmmhelp@sciops.esa.int> To: gaetz@cfa.harvard.edu Subject: Re: SASv10.0, mos_back problem (PR#43944) Date: Tue May 18 09:03:55 2010
Hello Terry, Your question has been forwarded to SAS developers for analysis. We will come back with the results of their analysis as soon as it is available. Sincerely, Pedro
From: Pedro M. Rodriguez Pascual <xmmhelp@sciops.esa.int> To: gaetz@cfa.harvard.edu Subject: Re: SASv10.0, mos_back problem (PR#43944) Date: Mon May 24 11:29:55 2010
Terry, According to developers, the problem is related to parameters initialization in calls to fitsio library; Linux is not affected, but Macs are. It will be fixed in the next release. Sincerely, Pedro
Date: Sat, 29 May 2010 14:04:23 -0400 From: Terry Gaetz <gaetz@cfa.harvard.edu> To: "Pedro M. Rodriguez Pascual" <xmmhelp@sciops.esa.int> Cc: Terry Gaetz <gaetz@cfa.harvard.edu> Subject: Re: SASv10.0, mos_back problem (PR#43944)
Pedro M. Rodriguez Pascual wrote: > Terry, > > According to developers, the problem is related to parameters initialization in > calls to fitsio library; Linux is not affected, but Macs are. It will be fixed > in the next release. Hi Pedro. NOTE: The problem *DOES* affect linux - I was in fact reporting the problem for linux boxes (x86_64) running CentOS 5.4 (& 5.5), (and CentOS is a repackaging of RHEL 5.4 (and 5.5)). I haven't actually tried XMMSAS or XMMESAS on the intel Mac (Snow Leopard, 10.6.3) yet, since my main machines are the *linux* boxes running CentOS. When will a patch be available? Since all my observations are of diffuse objects, I really need the xmmesas software running. Now that it is integrated into SAS, I no longer have the option of debugging and recompiling my own version... Thanks. Terry > Sincerely, > > Pedro > > > > ================================================================================================ > This message and any attachments are intended for the use of the addressee or addressees only. > > The unauthorised disclosure, use, dissemination or copying (either in whole or in part) of its content is prohibited. > > If you received this message in error, please delete it from your system and notify the sender. > > E-mails can be altered and their integrity cannot be guaranteed. > > ESA shall not be liable for any e-mail if modified. > =================================================================================================
From: Pedro M. Rodriguez Pascual <xmmhelp@sciops.esa.int> To: gaetz@cfa.harvard.edu Subject: Re: SASv10.0, mos_back problem (PR#43944) Date: Tue Jun 8 12:01:30 2010
Hello Terry, My apologies for this late reply, but it has taken me a while to run esas tasks on the 0129341501 data-set. Unfortunately, it is no easy for me to test exactly the same configuration you are using: SAS built on a 32 bits running on a 64 bits computer. However, from the previous analysis, the fact that the tasks run on 32bits machines and the status value returned by the cfitsio function ftgiou in your email, the initialization of the fitsio library seems very likely to be in the origin of the problem. SAS developers at ESAC are currently testing a patch provided by external developers. Upon success of these tests, the patch will be made public, but I cannot give now an estimate other than "some weeks" on when it will happen. Sorry, Pedro
Date: Tue, 15 Jun 2010 23:38:36 -0400 From: Terry Gaetz <gaetz@cfa.harvard.edu> To: "Pedro M. Rodriguez Pascual" <xmmhelp@sciops.esa.int> Cc: Terry Gaetz <gaetz@cfa.harvard.edu> Subject: Re: SASv10.0, mos_back problem (PR#43944)
Pedro M. Rodriguez Pascual wrote: > Hello Terry, > > Your question has been forwarded to SAS developers for analysis. We will come > back with the results of their analysis as soon as it is available. > > Sincerely, > > Pedro Hello Pedro I am sorry for the delay in replying -- a family matter has kept me out of touch. The patched version fixes the previous problem, but now I run into another issue. In processing the corner data, the exposure-weighted count rate ends up with NaN's. This propagates through the processing, and the task dies in POLY_FIT, with a floating point error. The problem seems to start while processing mos1-qpb.fits.gz: /home/gaetz/data/caldb/data/xmm/esas_ccf/mos1-qpb.fits.gz Rate: 3.1818010E-02 Rate error: NaN Hard: 3.7560976 Hard error: NaN The end of the log looks like: Reading input file mos1S001-im5-400-2000.fits Average non-zero value of FWC image: 1.4981754 FWC image scaling for chip: 5 NaN 3.8590000E+04 NaN mos1S001-im7-400-2000.fits Reading input file mos1S001-im7-400-2000.fits Average non-zero value of FWC image: 1.4894192 FWC image scaling for chip: 7 NaN 3.7945492E+04 NaN Reading input file mos1S001-obj-im-det-400-2000.fits Total observation counts in band 8.0534000E+04 Total background counts in band NaN DGAUSJ, singular matrix, pos 2 MQM ==> Bad return from DGAUSJ Bad return from MQM Program stopped in POLY_FIT Position 01 I can send the rest of the log if you need it. It showed up in my test case (the observation id I sent before), and also the science case I'm working on. Both cases died the same way, suggesting that there is a deeper problem. Please forward this to the appropriate people, and let me know if further information is needed. Thank you. Terry
From: Pedro M. Rodriguez Pascual <xmmhelp@sciops.esa.int> To: gaetz@cfa.harvard.edu Subject: Re: SASv10.0, mos_back problem (PR#43944) Date: Tue Jun 22 08:05:53 2010
Hello Terry, sorry for this long silence. Your question was immediately forwarded to SAS developers and I have been waiting for their comments to reply. Unfortunately, I have not got them yet and this message is just to let you know that your question has not been forgotten. with our apologies, Pedro
From: Pedro M. Rodriguez Pascual <xmmhelp@sciops.esa.int> To: gaetz@cfa.harvard.edu Subject: Re: SASv10.0, mos_back problem (PR#43944) Date: Fri Jul 2 09:17:45 2010
Hello Terry, SAS developers do not reproduce the last problem reported, on obsid 0129341501. Would it be possible you send to us the complete log of mos-back when run on 0129341501 ? Alternatively, the complete log and obsid corresponding to your last message (Jun 15) would be fine. Thanks, Pedro