API Reference

Pipeline stages

class clpipe.clp_covariance.CLPCovariance(*args, **kwargs)[source]

Bases: PipelineStage

TJPCov pipeline stage for covariance computation.

This stage: - Reads an input SACC file (data vector) - Reads the fiducial cosmology (shared with TXPipe) - Computes covariance terms using TJPCov - Optionally merges in non-cluster covariance blocks from the input file - Optionally replaces cluster-count covariance using CROW - Writes a single output SACC file with the final covariance

Key configuration groups: - Covariance selection (cov_type) - Mass–observable relation (mor_parameters) - Pipeline behavior (replace_tjpcov_cov)

run()[source]

Main execution:

  • Load input SACC file and fiducial cosmology

  • Compute covariance via TJPCov

  • Merge with existing non-cluster covariance blocks (if needed)

  • Optionally replace cluster-count covariance

  • Save the final result exactly once

extract_and_save_cluster_counts(input_sacc_file, output_sacc_file)[source]

Reads a SACC file, extracts only the cluster counts data (without covariance), and saves it to a new SACC file.

Parameters:

  • input_sacc_file (str) – Path to the input SACC file containing full data.

  • output_sacc_file (str) – Path where the new SACC file with only cluster counts will be saved.

merge_data_covariance(sacc_obj, full_cov, diagonal_only=True)[source]

Copy non-cluster-count covariance blocks from the original input SACC file’s covariance into the newly computed full covariance.

TJPCov only computes cluster-related covariance terms in this pipeline (cov_type is restricted to Cluster* classes), so any other data type (e.g. cluster_delta_sigma) that was already present with a covariance in the input file needs to be preserved here rather than left at TJPCov’s placeholder value.

Parameters:

  • sacc_obj (sacc.Sacc) – The original input SACC file, still carrying its original covariance matrix.

  • full_cov (np.ndarray) – The newly computed full covariance array to merge non-cluster blocks into (modified in place and returned).

  • diagonal_only (bool) – If True, only copy the diagonal (per-point variance) for each non-cluster-count data type, dropping any off-diagonal correlation between data points of that type.

Returns:

full_cov with non-cluster-count blocks copied over from sacc_obj, either diagonal-only or full dense blocks depending on diagonal_only.

Return type:

np.ndarray

replace_crow_counts(config_dict, sacc_full, cov_terms, full_cov, cosmo)[source]

Replace TJPCov cluster-count covariance using CROW predictions.

This is a temporary workaround.

Steps: - Construct mass–observable relation (mor_parameters) - Compute theoretical counts - Replace covariance elements using SSC scaling

WARNING: - Hardcoded modeling choices (mass function, grids) - Should eventually be implemented inside TJPCov

Parameters:

  • config_dict (dict) – Stage configuration.

  • sacc_full (sacc.Sacc) – SACC object with tracers/data points for the full covariance (used to look up tracer metadata).

  • cov_terms (dict) – {‘gauss’: array, ‘SSC’: array} raw covariance term arrays from TJPCov.

  • full_cov (np.ndarray) – Full covariance array to modify in place.

  • cosmo (pyccl.Cosmology) – fiducial cosmology, the same object used for TJPCov’s own covariance computation in run().

Returns:

full_cov with cluster-count blocks replaced.

Return type:

np.ndarray

class clpipe.clp_firecrown.CLPFirecrown(*args, **kwargs)[source]

Bases: PipelineStage

Firecrown pipeline stage for cluster cosmology analysis.

This stage:

  • Builds a Firecrown likelihood from a SACC file

  • Generates the corresponding CosmoSIS configuration

  • Writes parameter files for sampling, defaulting the cosmological block to the fiducial cosmology (shared with TXPipe/TJPCov); any entry in cosmological_parameters overrides that default, which is how a parameter gets sampled instead of held fixed

Key configuration groups:

  • Modeling options (hmf, mass range, redshift range)

  • Observable selection (cluster counts, shear)

  • Systematics (purity, completeness)

  • Sampling configuration (emcee, polychord)

Full configuration documentation: See docs/clp_firecrown.txt

run()[source]

Run the analysis for this stage. Generates Firecrown likelihood and cosmosis ini files.

generate_python_file(path_name)[source]

Generates a Python file based on the configuration dictionary.

Parameters:

path_name (str) – Path to save the generated Python file.

generate_ini_file(output_ini_path, likelihood_source_name, values_file_name)[source]

Generates an .ini file.

Parameters:

  • output_ini_path (str) – Path where the generated .ini file will be saved.

  • likelihood_source_name (str) – Basename of the generated Firecrown likelihood Python file (the likelihood_file output), written alongside output_ini_path.

  • values_file_name (str) – Basename of the generated CosmoSIS values file (the priors_file output), written alongside output_ini_path.

File types

CLPipe-specific file types, subclassing base types from ceci. SACCFile and FiducialCosmology adapted from LSSTDESC/TXPipe (file_types.py). CosmosisFile and PythonFile are CLPipe additions, not present in core ceci.

class clpipe.file_types.SACCFile(*args, **kwargs)[source]

Bases: DataFile

Adapted from TXPipe.

class clpipe.file_types.CosmosisFile(*args, **kwargs)[source]

Bases: DataFile

A CosmoSIS .ini file. Not present in core ceci.

class clpipe.file_types.PythonFile(*args, **kwargs)[source]

Bases: DataFile

A plain-text Python source file. Not present in core ceci.

class clpipe.file_types.FiducialCosmology(*args, **kwargs)[source]

Bases: YamlFile

Adapted from TXPipe (LSSTDESC/TXPipe, file_types.py). TODO replace when CCL has more complete serialization tools.

Cluster theory predictions

clpipe.cluster_theory_pred.build_cluster_recipes_from_config(yml_file, sacc_file, set_params, **kwargs)[source]

Build cluster counts and (optionally) shear recipes from a YAML configuration and return theory/data vectors.

This function:

  • Loads configuration from yml_file (supports a top-level CLPFirecrown key).

  • Builds a pyccl halo mass function and pyccl Cosmology (via build_ccl_cosmology_from_config).

  • Constructs completeness/purity models if requested.

  • Configures a mass-proxy model and fills its parameters from the firecrown_parameters block or from runtime set_params.

  • Builds either GridBinnedClusterRecipe or ExactBinnedClusterRecipe for counts and (optionally) for shear.

  • Reads the SACC file and returns theory predictions and data vectors.

Parameters:

  • yml_file (str) – Path to the YAML configuration.

  • sacc_file (str) – Path to the SACC file used for data vectors.

  • set_params (Dict[str, Any]) – Dictionary with runtime-supplied sampled parameters (e.g. from Firecrown/Cosmosis).

  • **kwargs – Currently unused; accepted for forward compatibility.

Returns:

(cluster_counts_theory, cluster_counts_data, cluster_shear_profile_theory, cluster_shear_data) cluster_shear_profile_theory and cluster_shear_data can be None if shear is not requested.

Return type:

Tuple[ndarray, ndarray, Optional[ndarray], Optional[ndarray]]

clpipe.cluster_theory_pred.build_ccl_cosmology_from_config(yml_config, set_params=None)[source]

Build a pyccl.Cosmology using config values + runtime sampled parameters.

The configuration is expected to contain a “cosmological_parameters” block where each parameter is either marked as sampled (sample: true) in which case the value must be present in set_params, or fixed with a numeric “values” entry.

Parameters:

  • yml_config (Dict[str, Any]) – Configuration dictionary (the CLPFirecrown section).

  • set_params (Optional[Dict[str, Any]]) – Runtime-supplied sampled parameters (may be None).

Return type:

Cosmology