Open phbruel opened 4 years ago
@phbruel Would you mind retesting this behavior with the 1.3.4 dev/beta build? I am unable to replicate the above differing event_type outputs.
I've just installed fermitools 1.3.4 and tested the command line gtlike and I see the same problem. All the files are in /afs/slac/g/glast/users/bruel/github_issue_56.
test with irfs=CALDB gtlike irfs=CALDB expcube=out_gtltcube.fits srcmdl=model_4FGL_edisp_test_wsrc.xml optimizer=NEWMINUIT cmap=out_gtsrcmaps_PSF3_test.fits bexpmap=out_gtexpcube2_PSF3_test.fits chatter=4 statistic=BINNED This is gtlike version HEAD ResponseFunctions::load: IRF used: P8R3_SOURCE_V3 (PSF) event_types: 5
test with irfs=P8R3_SOURCE_V3 gtlike irfs=P8R3_SOURCE_V3 expcube=out_gtltcube.fits srcmdl=model_4FGL_edisp_test_wsrc.xml optimizer=NEWMINUIT cmap=out_gtsrcmaps_PSF3_test.fits bexpmap=out_gtexpcube2_PSF3_test.fits chatter=4 statistic=BINNED This is gtlike version HEAD ResponseFunctions::load: IRF used: P8R3_SOURCE_V3 event_types: 0 1
@jasercion have you looked more into this issue? I see the same thing as @phbruel does with the latest betas 1.3.6 and 1.9.13. Specifying the IRF leads to different event type selection.
Using some test files produced with P8R2_SOURCE_V2 PSF3 evt type (evtype=32), I noticed a strange behavior when gtlike creates maps on the fly when irfs=P8R3_SOURCE_V2.
When running gtsrcmaps, irfs=CALDB and irfs=P8R3_SOURCE_V2 both select the same event class and type: ResponseFunctions::load: IRF used: P8R3_SOURCE_V2 (PSF) event_types: 5
But gtlike behaves differently:
The former is correct while the latter selects incorrectly FRONT+BACK.
This happens in ScienceTools 11-07-00 and in FermiTools 1.2.23.
It seems safer that gtlike should follow gtsrcmaps behavior.