Bug #2660

csphagen "No point spread function information found in response." for nthreads<>1

Added by Specovius Andreas over 6 years ago. Updated about 6 years ago.

Status:ClosedStart date:08/06/2018
Priority:HighDue date:
Assigned To:Knödlseder Jürgen% Done:

100%

Category:-
Target version:1.6.0
Duration:

Description

When executing csphagen with parameter nthreads=1 everything seems to work fine.
When executing csphagen with parameter nthreads not set (default 0) or set to 0 or >1 an error occurs (see below!).
This looks like something does not work properly in the multithreaded case!

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-20-1ece58d054b9> in <module>()
     50 # phagen['nthreads']    = 1
     51 phagen['use_irf_bkg'] = False
---> 52 phagen.execute()
     53 
     54 print("- %s" % phagen.obs())

/home/woody/caph/sn0533/git/ctools_install/lib/python2.7/site-packages/ctools/tools.pyc in _execute(self)
   1539     self.logFileOpen()
   1540     self._read_ahead(True)
-> 1541     self.run()
   1542     self.save()
   1543 csobservation.execute = _execute

/home/woody/caph/sn0533/git/ctools_install/lib/python2.7/site-packages/cscripts/csphagen.pyc in run(self)
    587             args        = [(self, '_process_observation', i)
    588                            for i in range(self.obs().size())]
--> 589             poolresults = mputils.process(self._nthreads, mputils.mpfunc, args)
    590 
    591             # Construct results

/home/woody/caph/sn0533/git/ctools_install/lib/python2.7/site-packages/cscripts/mputils.pyc in process(nthreads, function, args)
    100 
    101     # Run function in parallel
--> 102     results = pool.map(function, args)
    103 
    104     # Close pool and join

/apps/python/2.7-anaconda/lib/python2.7/multiprocessing/pool.pyc in map(self, func, iterable, chunksize)
    251         '''
    252         assert self._state == RUN
--> 253         return self.map_async(func, iterable, chunksize).get()
    254 
    255     def imap(self, func, iterable, chunksize=1):

/apps/python/2.7-anaconda/lib/python2.7/multiprocessing/pool.pyc in get(self, timeout)
    570             return self._value
    571         else:
--> 572             raise self._value
    573 
    574     def _set(self, i, obj):

ValueError: *** ERROR in GCTAResponseIrf::psf_delta_max(double&, double&, double&, double&, double&): Invalid value. No point spread function information found in response.
Please make sure that the instrument response is properly defined.

Recurrence

No recurrence.

History

#1 Updated by Knödlseder Jürgen about 6 years ago

  • Assigned To set to Knödlseder Jürgen
  • Target version set to 1.6.0

Can you tell me what type of PSF you are using? Best would be to have some test data since I suspect that something goes wrong with the pickling.

#2 Updated by Knödlseder Jürgen about 6 years ago

  • Project changed from GammaLib to ctools
  • Target version deleted (1.6.0)

#3 Updated by Specovius Andreas about 6 years ago

Knödlseder Jürgen wrote:

Can you tell me what type of PSF you are using? Best would be to have some test data since I suspect that something goes wrong with the pickling.

I am using a table psf (→ loaded as GCTAPsfTable object).

What data do you need exactly for testing? Only the psf response file(s)?

#4 Updated by Knödlseder Jürgen about 6 years ago

  • Target version set to 1.6.0

I would need of one the PSF response files to make sure that the pickling works as expected.

#5 Updated by Specovius Andreas about 6 years ago

I just sent you one of the H.E.S.S. table psf response files via email!

#6 Updated by Knödlseder Jürgen about 6 years ago

  • Status changed from New to In Progress
  • % Done changed from 0 to 50

Looks like the issue was that the default PSF filename extension that is used if no extension name is specified was set to "POINT SPREAD FUNCTION" instead of "PSF_2D_TABLE" in GCTAPsfTable.hpp.

In addition, the pickling of @@ does not work for a PSF table, which is likely the origin of the problem:

>>> irf=gammalib.GCTAResponseIrf()
>>> irf.load_psf('hess_psf_table_047802.fits')
>>> print irf
=== GCTAResponseIrf ===
 Caldb mission .............: 
 Caldb instrument ..........: 
 Response name .............: 
 Energy dispersion .........: Not used
 Save energy range .........: undefined
=== GCaldb ===
 Database root .............: /usr/local/gamma/share/caldb
=== GCTAPsfTable ===
 Filename ..................: hess_psf_table_047802.fits
 Number of energy bins .....: 32
 Number of offset bins .....: 6
 Number of delta bins ......: 144
 Log10(Energy) range .......: 0.00999999977648258 - 100 TeV
 Offset angle range ........: 0 - 2.5 deg
 Delta angle range .........: 0 - 0.665605783462524 deg
>>> dump=pickle.dumps(irf)
>>> obj=pickle.loads(dump)
>>> print obj
=== GCTAResponseIrf ===
 Caldb mission .............: 
 Caldb instrument ..........: 
 Response name .............: 
 Energy dispersion .........: Not used
 Save energy range .........: undefined
=== GCaldb ===
 Database root .............: /usr/local/gamma/share/caldb

#7 Updated by Knödlseder Jürgen about 6 years ago

The pickling method was only implemented for the case of a calibration database. I’ll change that.

%pythoncode {
    def __getstate__(self):
        state = (self.rspname(), self.caldb())
        return state
    def __setstate__(self, state):
        if state[0]:
            self.__init__(state[0], state[1])
        else:
            self.__init__()
}

#8 Updated by Knödlseder Jürgen about 6 years ago

  • % Done changed from 50 to 60

I’m making progress on the issue. To be sure that all other IRF components can be correctly digested by GammaLib, could you please also send an effective area file and a background file? Thanks in advance.

#9 Updated by Specovius Andreas about 6 years ago

I just sent you one of the H.E.S.S. effective area and background response files via email!

#10 Updated by Knödlseder Jürgen about 6 years ago

  • Status changed from In Progress to Pull request
  • % Done changed from 60 to 100

Thanks. I changed a bit the code so that all files are now digested correctly. I also added a pickling test.

I’m currently integrating the code.

#11 Updated by Knödlseder Jürgen about 6 years ago

  • Status changed from Pull request to Feedback

The code is now merged into devel (changes only in GammaLib). Could you please check if csphagen now also works for nthreads!=1?

#12 Updated by Specovius Andreas about 6 years ago

For me csphagen runs now with nthreads<>1 without problems.
Good job!

#13 Updated by Knödlseder Jürgen about 6 years ago

  • Status changed from Feedback to Closed

Also available in: Atom PDF