Closed kevinhainline closed 7 years ago
jacopo-chevallard
commented
7 years ago Hi Kevin, that's interesting, we had exactly the same issue yesterday with @camipacifici and couldn't understand the origin of the problem... As you noticed, everything seems working fine when using the original filterfrm.res to create the binary FILTERBIN.RES, while using a different filterfrm.res seems to break down the things... I'll take a look asap
jacopo-chevallard
commented
7 years ago How many points do you have in the new MIRI filter curves?
kevinhainline
commented
7 years ago The MIRI filter curves have 2-3000 points. I can cut them down to smaller versions and try, if you think that would work.
jacopo-chevallard
commented
7 years ago I think it's worth trying, maybe resampling them to < 1000 points and see if this solves
kevinhainline
commented
7 years ago So, I went and just used three filters:
% ./add_filters
Filter response functions in (formatted) file = filterfrm.miri.cut.small.res
Normalize response functions to area = 1 (y or [n]) ? n
1 MIRI F560W 399
2 MIRI F770W 449
3 MIRI F1000W 399
Write binary file (y or [n]) ? y
Enter output (binary) file name = FILTERBIN.RES
Filter index in file: filters.log
STOP Binary file written.Here are the filter files:
And here is what happens when trying to use this filter file:
% % docker run --rm -it -v /Users/knh/Desktop/NIRCam/BEAGLE/BEAGLE-general-master:/BEAGLE_virtual_workdir --env-file env.list beagletool/beagle:0.8.2 --parameter-file /BEAGLE_virtual_workdir/params/BEAGLE_mock_Stacey.param --mock
Reading Filter File: /BEAGLE_virtual_workdir/filters/FILTERBIN.RES
3 filters defined, out of 500 maximum ...done
---> fixed sfh_type 0
---> gridded metallicity 0
---> gridded mass 0
---> dependent sfr 0
---> gridded redshift 0
---> fixed formation_redshift 0
---> dependent nebular_logU 0
---> fixed nebular_xi 0
---> dependent nebular_Z 0
---> fixed attenuation_type 0
---> fixed tauV_eff 0
---> fixed mu 0
n_fitted: 0
--- LINEAR: X0 = 6.020E+04 is outside X range --- 9.000E+02 6.010E+04 2362
--- Error reported only once. It may occur more than once. ---
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7F9A387F1E08
#1 0x7F9A387F0F90
#2 0x7F9A37F2349F
#3 0x4B6091 in sort2_ at sort2.f:13
#4 0x4B4E5B in build_filter_array_ at filter_n.f:189
#5 0x4B520F in filtr_master_ at filter_n.f:61
#6 0x4B566E in filtr_master_ at filter_n.f:17
#7 0x42F97C in __galaxy_sed_MOD_compute_sed at galaxy_SED.f90:486 (discriminator 14)
#8 0x406DC8 in MAIN__ at BEAGLE.f90:462And I get the same segmentation fault. I am going to try this with one of the original NIRCam filters and see what happens.
kevinhainline
commented
7 years ago Ok, perhaps I'm doing something very wrong. If I go and take just the F444W filter:
% ./add_filters
Filter response functions in (formatted) file = filterfrm.nircam.small_two.res
Normalize response functions to area = 1 (y or [n]) ? n
1 NIRCam F444W 1452
Write binary file (y or [n]) ? y
Enter output (binary) file name = FILTERBIN.RES
Filter index in file: filters.log
STOP Binary file written....and use it to create FILTERBIN.RES, and then update the filter file and the parameter file to reflect this:
# It contains the filters to be used when calculating only photometric models
# The possible units that can be used are: Jy, milliJy, microJy, nanoJy
# Alternatively one can pass a conversion factor to convert the fluxes in the
# catalogue into [erg s^-1 cm^-2 Hz^-1] -----> units:value:<conversion factor>
units:microJy
object_ID:colName:ID
#redshift:colName:redshift
#mass:colName:mass
index:1 min_rel_err:0.02 flux:colName:F444W fluxerr:colName:F444W_err label:F444WI end up with this:
% docker run --rm -it -v /Users/knh/Desktop/NIRCam/BEAGLE/BEAGLE-general-master:/BEAGLE_virtual_workdir --env-file env.list beagletool/beagle:0.8.2 --parameter-file /BEAGLE_virtual_workdir/params/BEAGLE_mock_Stacey.param --mock
Reading Filter File: /BEAGLE_virtual_workdir/filters/FILTERBIN.RES
1 filters defined, out of 500 maximum ...done
---> fixed sfh_type 0
---> gridded metallicity 0
---> gridded mass 0
---> dependent sfr 0
---> gridded redshift 0
---> fixed formation_redshift 0
---> dependent nebular_logU 0
---> fixed nebular_xi 0
---> dependent nebular_Z 0
---> fixed attenuation_type 0
---> fixed tauV_eff 0
---> fixed mu 0
n_fitted: 0
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7F03521B0E08
#1 0x7F03521AFF90
#2 0x7F03518E249F
#3 0x4B6091 in sort2_ at sort2.f:13
#4 0x4B4E5B in build_filter_array_ at filter_n.f:189
#5 0x4B520F in filtr_master_ at filter_n.f:61
#6 0x4B566E in filtr_master_ at filter_n.f:17
#7 0x42F97C in __galaxy_sed_MOD_compute_sed at galaxy_SED.f90:486 (discriminator 14)
#8 0x406DC8 in MAIN__ at BEAGLE.f90:462The same line worked when I used the full FILTERBIN.RES file from which F444W was taken. So it's not just the MIRI filters.
camipacifici
commented
7 years ago Hi Kevin,
I believe the code needs at least a standard optical filter to compute all the magnitudes. Can you please try using a filter file that includes the original filters down to at least SDSS (I believe SDSS z-band is number 119) and then the new filters you want to add?
Best wishes,
Cami
kevinhainline
commented
7 years ago So, if I go and add in the filters out to 119 (the SDSS z filter), and after that I add in some NIRCam and MIRI filters, I still get a segmentation fault. Even if I just add in one NIRCam filter, I get the segmentation fault:
% docker run --rm -it -v /Users/knh/Desktop/NIRCam/BEAGLE/BEAGLE-general-master:/BEAGLE_virtual_workdir --env-file env.list beagletool/beagle:0.8.2 --parameter-file /BEAGLE_virtual_workdir/params/BEAGLE_mock_Stacey.param --mock
Reading Filter File: /BEAGLE_virtual_workdir/filters/FILTERBIN.RES
120 filters defined, out of 500 maximum ...done
---> fixed sfh_type 0
---> gridded metallicity 0
---> gridded mass 0
---> dependent sfr 0
---> gridded redshift 0
---> fixed formation_redshift 0
---> dependent nebular_logU 0
---> fixed nebular_xi 0
---> dependent nebular_Z 0
---> fixed attenuation_type 0
---> fixed tauV_eff 0
---> fixed mu 0
n_fitted: 0
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7F97FC1C8E08
#1 0x7F97FC1C7F90
#2 0x7F97FB8FA49F
#3 0x4B6091 in sort2_ at sort2.f:13
#4 0x4B4E5B in build_filter_array_ at filter_n.f:189
#5 0x4B520F in filtr_master_ at filter_n.f:61
#6 0x4B566E in filtr_master_ at filter_n.f:17
#7 0x42F97C in __galaxy_sed_MOD_compute_sed at galaxy_SED.f90:486 (discriminator 14)
#8 0x406DC8 in MAIN__ at BEAGLE.f90:462@kevinhainline @camipacifici the error message that you were getting was quite confusing, and it took me some time to figure it out, but the problem is that you should always add filters at the end of the filterfrm.res file provided in this repository (here), and not creating a new filterfrm.res file (this point was clear in the manual, less on the wiki, that I've now modified to clarify this point).
Kevin, we should rather work on the segfault error that you were getting in the first place, and make the addition of MIRI curves at the of filterfrm.res work
jacopo-chevallard
commented
7 years ago @kevinhainline @camipacifici Closing this, I assume that the above solution has solved your issues. Feel free to re-open if not!
I have two issues. I was trying to add the MIRI filters to the BEAGLE filter list, and first, I found that _addfilters will spit out a segmentation fault if the filterfrm.res file is too long. So I took the NIRCam and MIRI filters and put them into a separate filterfrm.res file, and ran add_filters, which worked this time, creating a FILTERBIN.RES file that I then used to run BEAGLE (I updated the filters file in the parameter file).
Here is what happens:
Any idea what's going on?