Unable to simulate UV spectra with snlc_sim

[qnwang93]

I was trying to simulated some SNe Ia sample with UV spectra (wavelngth range ~ 1800-10000A) with snlc_sim.exe. However, the minimum wavelength of output spectra only goes down to 3600A. Is there a way to simulate the UV spectra?

I moved related files to $SNDATA/simlib/UVEXT/ . Sorry I don't have access to $SNANA_DEBUG so I made a new directory there. Feel free to move it to elsewhere applicable.

[qnwang93]

uploaded a new input file named SALT3_SIM_UVEXT.INPUT to $SNANA_DEBUG/qnwang/UVEXT_spectral_wavelength/ and it should be able to reproduce this issue now.

[RickKessler]

There are two issues with your spectra being truncated.
First, when using TAKE_SPECTRUM keys in the SIMLIB file, the output spectrum is constrained to be within the wavelengths provided in the SNR_LAMOBS key. The reasoning is that the sim should only generate within the wave range of the actual data that these TAKE_SPECTRUM keys represent. It doesn't make sense to simulate & analyze spectra that extend in wavelength beyond actual data.

Beyond the first issue, there is additional wavelength truncation due to invalid ZP_SPECTROGRAPH calculation (see manual screen shot below). The current sim code rejects wave bins with ZP<0 (i.e., "invalid"), but does not give any warning, which can result in wave bins mysteriously not appearing in the output spectra. For your spectrograph, about 1/3 of the wave bins have invalid ZP_SPECTROGRAPH < 0, and this is because of the values in
MAGREF_LIST: 17.0 32.0
The bright end of 17 is fine, but the faint end of 32 is far too faint, resulting in a poor estimate of ZP. Would suggest increasing the faint end to be closer to the faint end of SN that pass your selection criteria.

To avoid future confusion, I added a diagnostic to print ZP(min, max) and to abort if more than 2% of the wave bins fail the ZP>0 cut. Your test job now aborts with this error:
FATAL ERROR ABORT called by INIT_SPECTROGRAPH_SEDMODEL
Invalid ZP_SPECTROGRAPH frac = 0.335; exceeds FRAC_ZP_INVALID_ABORT=0.020
Check MAGREF_LIST in spectrograph table

I checked this diagnostic on the Roman PRISM, and ZPmin/max is about 10-11, well above the ZP>0 requirement. If this 2% abort threshold is too severe, I can loosen it or add another sim-input key allowing user-defined threshold, but was hoping to avoid yet another input key.

[qnwang93]

update: I tried two new MAGREF_LIST setting (17-27 and 17-22) with updated SNRs.
With 17-27 I got the same error as invalid ZP_SPECTROGRAPH frac = 0.335.
With 17-22 setting, program aborted when I try to make new kcor files with following message:
`PRE-ABORT DUMP from function check_SNR_SPECTROGRAPH :
MAGREF[0,1] = 17.000, 32.000
FLUXREF_RATIO = 1.000000e-06
SNR1/SNR0[real,Poisson] = 0.100010, 0.001000
LOGSNR_RATIO = 2.000045 (TOL=0.699)

|```````|
<| o\ /o |>
| ' ; ' |
| ___ | ABORT program on Fatal Error.
| |' '| |
| `---' |
_______/

FATAL ERROR ABORT called by check_SNR_SPECTROGRAPH
Possible crazy SNR1/SNR0 ratio (way off from Poisson)
LAM=1980.3 TEXPOSE=30.00 (l=0, t=0)`

I've uploaded the related files under $SNANA_DEBUG/qnwang/UVEXT_spectral_wavelength/ and the above issue can be reproduced by running kcor.exe kcor_uvext.input in the above folder.

[RickKessler]

Your kcor file is still pointing to a spectrograph file with
MAGREF_LIST: 17.0 32.0

[qnwang93]

Sorry for that. Now I corrected it and made a kcor file with MAGREF_LIST: 17.0 22.0 named as kcor_uvext.fits. Then I tried to run snlc_sim.exe SALT3_SIM_FORQINAN.INPUT again, and got the same error with previous 27/32 settings, i.e. invalid ZP_SPECTROGRAPH frac = 0.335.

[RickKessler]

sorry I lost track of this issue over the holidays. I just modified
sntools_spectrograph.h and set
ZPMIN_SPECTROGRAPH -10 (instead of zero) and now your simulation runs without aborting, but I have not checked the output.