NumCosmo: Firecrown Likelihood Best Fit#
This recipe outlines the process to compute the best fit for Supernova SNIa data using
the Firecrown example with NumCosmo. Ensure that you have already installed
Firecrown, CosmoSIS, and NumCosmo and set the environment variable FIRECROWN_DIR
to the location of the Firecrown installation.
Convert Configuration File#
Use the NumCosmo command-line tool to convert the sn_srd.ini
file into a NumCosmo
configuration in a yaml
file. Since this likelihood does not require a
power-spectrum likelihood, you can use the from-cosmosis
command without any extra
flags –matter-ps eisenstein_hu nor –nonlin-matter-ps halofit. To mute the output
of the consistency cosmosis module, use --mute-cosmosis
.
cd $FIRECROWN_DIR/examples/srd_sn
numcosmo from-cosmosis sn_srd.ini --mute-cosmosis
This command creates a file called sn_srd.yaml
, NumCosmo’s experiment file. It
contains cosmological and likelihood parameters, modeling choices, and chosen precision.
Run NumCosmo App#
Execute the NumCosmo app with the experiment file to compute the best fit. The
following command saves the output in a new yaml
file. Use the --help
flag for
available options.
cd $FIRECROWN_DIR/examples/srd_sn
numcosmo run fit sn_srd.yaml --output sn_srd-bestfit.yaml
The best fit is saved in sn_srd-bestfit.yaml
. Inspect the file to view the
best-fitting values. This file serves as the starting point for subsequent runs.
Restart Minimization Algorithm (Optional)#
If dealing with a high-dimensional parameter space or non-converging algorithms, consider using the –restart flag. This restarts the minimization algorithm from the best-fit found so far.
cd $FIRECROWN_DIR/examples/srd_sn
numcosmo run fit sn_srd.yaml --output sn_srd-bestfit.yaml --starting-point sn_srd-bestfit.yaml --restart 1.0e-3 0.0
The command restarts the minimization algorithm if the absolute change in the likelihood is less than 1.0e-3 and the relative change is less than 0.0.
Note: sn_srd-bestfit.yaml
cannot be used as the experiment file for the
numcosmo
command-line tool since it lacks data and model information.
Results Summary#
Starting Point:
m2lnL: 1215
Parameters:
sn_ddf_sample_M: -19.30
Omegac: 0.260
Best Fit:
m2lnL: 1.70934
Parameters:
sn_ddf_sample_M: -19.426
Omegac: 0.265