Working with Dan Scolnic, Dillon Brout, and Maria Vincenzi (and others) on regenerating DES simulations.
As a baseline, I am trying to regenerate DES5YR sims, but with one modification --- generating the RV and EBV distributions from the HOSTLIB rather than a GENPDF (I am aiming to do this so that I can update the HOSTLIB for intergalactic dust in another simulation); however, I am unable to obtain the same RV distributions when generating from the GENPDF vs. the HOSTLIB (the difference is most apparent for low mass hosts).
I have set up the commands to recreate this issue in $SNANA_DEBUG/epeterson3.
Outputs to the commands BIASCORSIM_DES.CMD (GENPDF) and BIASCORSIM_DESfromHOSTLIB.CMD (HOSTLIB) go here /scratch/midway2/rkessler/SNDATA_ROOT/SIM/TMP_epet*. Comparing the two commands (diff BIASCORSIM_DES.CMD BIASCORSIM_DESfromHOSTLIB.CMD), I observe that they are very similar. I shortened the sims by a factor of approximately 10 from what I have been doing for my analysis, and I added a SIMGEN_DUMP line to include RV and LOGMASS_TRUE in the output dump files.
The different low mass RV distributions are plotted with compare_dumps.py and observed in genpdf_hostlib_RV.png where the GENPDF method seems to experience significant selection effects while the HOSTLIB method does not.
Working with Dan Scolnic, Dillon Brout, and Maria Vincenzi (and others) on regenerating DES simulations.
As a baseline, I am trying to regenerate DES5YR sims, but with one modification --- generating the RV and EBV distributions from the HOSTLIB rather than a GENPDF (I am aiming to do this so that I can update the HOSTLIB for intergalactic dust in another simulation); however, I am unable to obtain the same RV distributions when generating from the GENPDF vs. the HOSTLIB (the difference is most apparent for low mass hosts).
I have set up the commands to recreate this issue in
$SNANA_DEBUG/epeterson3
.Outputs to the commands
BIASCORSIM_DES.CMD
(GENPDF) andBIASCORSIM_DESfromHOSTLIB.CMD
(HOSTLIB) go here/scratch/midway2/rkessler/SNDATA_ROOT/SIM/TMP_epet*
. Comparing the two commands (diff BIASCORSIM_DES.CMD BIASCORSIM_DESfromHOSTLIB.CMD
), I observe that they are very similar. I shortened the sims by a factor of approximately 10 from what I have been doing for my analysis, and I added a SIMGEN_DUMP line to include RV and LOGMASS_TRUE in the output dump files.The different low mass RV distributions are plotted with
compare_dumps.py
and observed ingenpdf_hostlib_RV.png
where the GENPDF method seems to experience significant selection effects while the HOSTLIB method does not.