MCEvidence

A python package implementing the MARGINAL LIKELIHOODS FROM MONTE CARLO MARKOV CHAINS algorithm described in Heavens et. al. (2017)

This code is tested in Python 2 version 2.7.12 and Python 3 version 3.5.2.

Notes

The MCEvidence algorithm is implemented using scikit nearest neighbour code.

Installation

To install this project into your machine using pip, do the following

 $ git clone https://github.com/yabebalFantaye/MCEvidence
 $ cd MCEvidence
 $ pip install . --editable

The "--editable" or "-e" extension in the last command is to install the project in the editable mode.

To install this project with pip without clonning

 $ pip install git+https://github.com/yabebalFantaye/MCEvidence

Examples

To run the evidence estimation from an ipython terminal or notebook

>> from MCEvidence import MCEvidence
>> MLE = MCEvidence('/path/to/chain').evidence()

You can find a more advanced example that uses MCEvidence to analyse a set of MCMC chains in planck_mcevidence.py. The result of our companion paper No evidence for extensions to the standard cosmological model is obtained using this code.

To run MCEvidence from shell

$ python MCEvidence.py </path/to/chain> [optional arguments]

You can check the allowed parameters by doing $ python MCEvidence.py -h

The output is:

usage: MCEvidence.py [-h] [-k KMAX] [-ic IDCHAIN] [-np NDIM] [-b BURNFRAC]
                     [-t THINFRAC] [-v VERBOSE] [--cosmo]
                              root_name

Planck Chains MCEvidence. Returns the log Bayesian Evidence computed using the
kth NN.

 positional arguments:
   root_name             Root filename for MCMC chains or python class filename

 optional arguments:
   -h, --help            show this help message and exit
   -k KMAX, --kmax KMAX  scikit maximum K-NN
   -ic IDCHAIN, --idchain IDCHAIN
                    Which chains to use - the id e.g 1 means read only
                    *_1.txt (default=None - use all available)
   -np NDIM, --ndim NDIM
                    How many parameters to use (default=None - use all
                    params)             
   -b BURNFRAC, --burnfrac BURNFRAC, --burnin BURNFRAC, --remove BURNFRAC
                    Burn-in fraction
   -t THINFRAC, --thin THINFRAC, --thinfrac THINFRAC
                    Thinning fraction
   -vb VERBOSE, --verbose VERBOSE
                    Verbosity of the code while running: The mapping between verbose number
                    and the logging module levels are: 0: WARNNING, 1: INFO, 2: DEBUG
                    setting verbose>2 outputs EVERYTHING
   --paramsfile
                    text file name that contains additional parameter names to be consider as cosmological parameters
   --allparams              
                    Flag to consider all parameters - prior_volume is computed using all parameters. By default only 
                    cosmological parameters listed in iscosmo_params function is considered.
                    For arbitrary parameter names, please modify the code accordingly.
                    If allparams is passed, consider using ndim<20 to get resonable evidence estimation.

   -np, --pvolume PRIORVOLUME
                    The prior volume to use if it can not be computed internally using cosmomc *.ranges or
                    montepython log.params files.

   --cross
                    Flag to split chain (s) into two sets to estimate cross Evidence. The DEFAULT is auto Evidence.
                    Cross EVIDENCE is computed using two independent chains. This means Nearest Neighbour of
                    a point  "A" in MCMC sample MC1 is searched in MCMC sample MC2.
                    THE ERROR ON THE EVIDENCE FROM (AUTO) EVIDENCE IS LARGER THAN THE CROSS EVIDENCE BY ~SQRT(2).
                    This is the result of : if the nearest neighbour of A is B, then the NN to B is LIKELY to be A.

If you use the code, please cite the following paper

.. [1] Heavens et. al. (2017)