API reference#

EROTICA’s public API — a curated selection, not an exhaustive one, and the difference is stated here rather than left for a reader to discover.

This page lists 32 names: the 12 in erotica.__all__ plus 20 more drawn from erotica.analysis.__all__, which itself exports 118. So roughly 90 public names in the analysis layer are importable and supported but not listed on this page — among them distance_model, fit_parallax_model, lambda_msr, half_mass_relaxation_time, coulomb_argument_from_mass_function and identifiability_report.

Warning

An earlier version of this paragraph called these “the classes and functions re-exported at the top level of the package (``erotica.__all__``)”. That was wrong in both directions at once: the page lists more than erotica.__all__ contains, and far less than the analysis layer exports. Do not read absence from this page as evidence that a name is private — check __all__. Genuinely private names are prefixed with an underscore.

Clustering & data#

`core.Clustering`	Perform HDBSCAN clustering with optional hyper-parameter search.
`core.HDBSCANEstimator`	Sklearn-compatible wrapper around `hdbscan.HDBSCAN`.
`io.DataLoader`	Load and filter EROTICA-compatible catalogues.
`preprocess.DataPreprocessor`	Wrap preprocessing steps commonly applied to EROTICA-ready tables.

utils.compare_datasets

Compare one, two, or four QTable datasets to identify overlaps, missing entries, and value discrepancies by source_id.

Analysis#

`ClusterAnalyzer`	High-level helper around `Clustering` and plotting utilities.
`ClusterDynamicsAnalyzer`	Stateful facade for cluster dynamical and Galactic-radius estimates.
`ClusterInferenceAnalyzer`	Probability-thresholded Bayesian summaries for a cluster source table.
`ClusterStructureAnalyzer`	Stateful structural analysis facade for a cluster source table.
`ClusterFigureBuilder`	Plotting facade around a cluster source table.
`IsochroneFitter`	Bayesian isochrone fitting using a MIST forward model and PyMC NUTS.
`PhotometricMassEstimator`	Mass-assignment facade for sampled isochrones and CMD source tables.

Reproducibility#

A result is reproducible only if you can say what code, what inputs and what randomness produced it. build_metadata() records all three; saving a trace with store_trace_results() writes it as a JSON sidecar automatically. See Decision log — why the code is the way it is for what each field is for and why.

`build_metadata`	Build a reproducibility record for a result.
`write_metadata`	Write `build_metadata()` to path as indented JSON.
`git_provenance`	Return the git commit and dirty flag for the tree containing path.
`file_checksum`	Content digest and size of one input file.
`dependency_versions`	Installed versions of the packages that can change a number.
`store_trace_results`	Append summarized trace statistics and optionally write the full trace to NetCDF.
`load_results`	Load summarized CSV results and optionally the associated NetCDF trace.
`summarize_trace`	Summarize all posterior variables in an ArviZ InferenceData object.
`posterior_mode`	Estimate posterior mode with ArviZ KDE if available, else histogram mode.

Radial structure#

Radial-profile fitting is an unbinned inhomogeneous Poisson point process: the likelihood is sum(log lambda(r_i)) - Lambda, with no binning step to choose and no Gaussian approximation to break in the sparse outer bins that set R_t. King and EFF share that machinery, so compare_radial_profiles() can put a Bayes factor between them. See Is a King profile the right model for NGC 6383? for what the fits can and cannot identify.

`king_unbinned`	Fit a King profile to stellar radii directly, without binning.
`eff_unbinned`	Fit an EFF profile to stellar radii, without binning.
`compare_radial_profiles`	Compare radial-profile families by Bayes factor.
`king_profile`	Projected King profile shape, plus an additive `background`.
`eff_surface_density`	EFF (Elson, Fall & Freeman 1987) surface density with a background.
`king_expected_count`	Expected number of stars inside a circular field, in closed form.
`king_expected_count_weighted`	Expected detected count when completeness varies with radius.
`eff_expected_count`	Expected count inside a circular field for the EFF profile, in closed form.
`radial_density_profile`	Return a typed radial-density profile for a cluster source table.

Synthetic clusters#

Validation data. Drawing from a smooth King or EFF profile begs the question when the thing being validated is a profile fit, so fractal_cluster() generates substructure by the Goodwin & Whitworth (2004) box-fractal construction instead.

`fractal_cluster`	Generate a box-fractal star cluster (Goodwin & Whitworth 2004).
`radial_profile_of`	Measure the radial surface (or volume) density of a realization.