The pipeline is the core of the CosmoSIS system. Samplers drive the calculation but interact with all other parts of CosmoSIS, and in particular the physics and likelihood modules, through the pipeline. The following describes the configuration options used to configure the CosmoSIS pipeline so it can be used by samplers.

Pipeline

The [pipeline] section lists the modules to load and execute, defines the input parameters and their priors, and one or more output likelihoods.

Example pipeline section:

#!ini

[pipeline]
modules = camb planck  ; a list of modules in the order they will be executed
values = values.ini    ; a single ini file containing parameter values and ranges
priors = priors.ini    ; a list of files containing parameter priors
extra_output =         ; optional list of computed items to save (e.g. sigma8, power spectrum)
likelihoods = planck   ; a list of modules that will compute likelihood values
quiet=T                ; silence extra output, default T
debug=F                ; turn on debugging output, default F
timing=F               ; output timing information for each module, default F

Output

The output section is used to configure the output file format for samplers that support chain-style output. Currently there are two output types, text and cosmomc.

Modules

Each entry in the pipeline modules parameter has its own ini section with a single required parameter, file, which points to the module's location (typically a file ending in .so or .py). The remaining parameters allow the module to specify the name of each of its module functions (setup, execute, and cleanup), and then configure any custom parameters for that module, passed in the DataBlock to setup.

An example module section is as follows:

#!ini

[planck]
file = cosmosis-standard-library/likelihood/planck/planck_interface.so ; path to the Planck module
t_high_file = ups/planckdata/v1_1/data/CAMspec_v6.2TN_2013_02_26_dist.clik ; custom parameter to the Planck module pointing to the location of the high-ell temperature data