Given a data set and a cosmological model, CAMEL can be used to perform:

  • A best fit estimation
  • A profile likelihood for a given parameter
  • A MCMC (Markov Chain Monte Carlo)

For each case, you will find a different executable.
All executables share the same parameter file.

Write CMB Power Spectra

To write a CMB power spectrum given some parameters, use the executable writeSpectra

writeSpectra parfile(in) lmax(in) cl.fits(out) 

TODO: describe OUTPUT

Compute χ2 (and check your parmeter file)

To compute χ2 for a given set of parameters and combination of likelihoods, use the executable writeChi2

writeChi2 parfile(in) [clfile(out)]

TODO: describe OUTPUT

Parameter scan

To compute χ2 for several values of one parameter (all other parameters being fixed) based on a combination of likelihoods, use the executable ScanParam

ScanParam parfile(in) varname xmin xmax Npt

TODO: describe OUTPUT

The previous executables can be run interactively.

From now on all the followig ones should be run in batch mode

cc-in2p3 users: see some scripts for submitting the follwing jobs under work/batch_cc

Best-fit estimation

To perform a bestfit estimation, use the executable Minimize

Minimize parfile(in) bestfit(out) [covmat(out)]

TODO: describe OUTPUT


To run a profile likelihood, use the executable profile

Profile parfile(in) varname xmin xmax Npt 

In practice, it is usually less time consuming to perform Npt independent best-fit estimations (using Minimize) for each value of the parameter varname between xmin and xmax since all the fits are performed in parallel while the Profile code will perform them one after the other.

  • parfile is a parameter file (look here)
  • varname is the variable you want to profile with the same name as in the parameter file e.g. n_s
  • xmin is the minimum value of varname for the profile you want to build e.g. 0.8
  • xmax is the maximum value of varname for the profile you want to build e.g. 1.
  • Npt is the number of points between xmin and xmax for which you want to find the chi2min e.g. 5

you should keep back the printout file.

Monte Carlo Markov Chain (MCMC)

To run a MCMC chain, use the executable mcmc

mcmc parfile(in) samplesfile(out)

TODO: describe OUTPUT