sum-stat 0.1.0

Cosmological summary statistics from galaxy catalogues and lensing data

sum_stat

Python package for measuring cosmological summary statistics from galaxy catalogues and lensing data, built on JAX for differentiable, JIT-compilable estimators.

Features

Module	Estimators
Two-point correlation functions	w_theta, xi_r, wp, xi_multipoles
Power spectra	cl_angular, pk3d, pk_multipoles
Galaxy–galaxy lensing	delta_sigma, delta_sigma_nfw_jax, delta_sigma_from_xi_jax
Luminosity / stellar-mass functions	luminosity_function_vmax, stellar_mass_function_vmax, luminosity_function_swml, cumulative_luminosity_function_cminus, fit_schechter_sty
Independence tests	efron_petrosian_tau (Efron & Petrosian 1992), rauzy_completeness (Rauzy 2001)
Redshift distributions	nz_histogram, nz_kde
Covariance	jackknife_covariance, bootstrap_covariance
I/O	SummaryStatWriter, SummaryStatReader (HDF5)
Catalogues	GalaxyCatalogue, ShapeCatalogue, PhotoZCalibTable
Cosmology	JAX-differentiable comoving distance, volume, Σ_crit

Installation

mamba env create -f environment.yml
mamba activate sum_stat
pip install -e .

Quick start

import numpy as np
import sum_stat as ss
from astropy.cosmology import FlatLambdaCDM

cosmo = FlatLambdaCDM(H0=67.36, Om0=0.3111)

# --- Catalogues ----------------------------------------------------------
cat  = ss.GalaxyCatalogue(ra=ra, dec=dec, redshift=z, weight=w, mag_limit=22.5)
rand = ss.GalaxyCatalogue(ra=ra_r, dec=dec_r, redshift=z_r)

# --- Two-point correlation functions -------------------------------------
import numpy as np
theta_bins = np.logspace(-2, 1, 20)          # degrees
theta, w, vw = ss.w_theta(cat, rand, theta_bins)

r_bins = np.logspace(-1, 2, 20)              # Mpc/h
r, xi, vxi = ss.xi_r(cat, rand, r_bins, cosmo)

r, wp_val, vwp = ss.wp(cat, rand, r_bins, pi_max=60.0, cosmo=cosmo)

s_bins = np.logspace(-1, 2, 20)
s, xi_dict = ss.xi_multipoles(cat, rand, s_bins, ells=(0, 2, 4), cosmo=cosmo)
# xi_dict[0], xi_dict[2], xi_dict[4] → monopole, quadrupole, hexadecapole

# --- Power spectra -------------------------------------------------------
ell_bins = np.geomspace(10, 3000, 30)
ell, cl, vcl = ss.cl_angular(cat, rand, ell_bins)

k_bins = np.logspace(-2, 0, 30)
k, pk, vpk = ss.pk3d(cat, rand, k_bins, cosmo=cosmo)

# --- Galaxy–galaxy lensing -----------------------------------------------
import dsigma.helpers as dh
from astropy.table import Table

# Build a dsigma-formatted source table (see dsigma docs for survey details)
table_s = dh.dsigma_table(Table({
    "ra": ra_s, "dec": dec_s, "z": z_s,
    "e_1": e1, "e_2": e2, "w": w_s, "m": m, "e_rms": e_rms,
    "R_11": R11, "R_22": R22, "R_12": R12, "R_21": R21,
}), "source")

rp_bins = np.logspace(-1, 1.5, 15)          # Mpc
corrections = {"scalar_shear_response_correction": True,
               "shear_responsivity_correction": True}
rp, ds, cov = ss.delta_sigma(cat, table_s, rp_bins, cosmo,
                              corrections=corrections)

# --- Luminosity / stellar-mass functions ---------------------------------
from sum_stat.lf_smf.swml import luminosity_function_swml
from sum_stat.lf_smf.cminus import cumulative_luminosity_function_cminus
from sum_stat.lf_smf.independence import efron_petrosian_tau, rauzy_completeness

M_bins = np.linspace(-24, -18, 20)
area_sr = 0.3

# 1/Vmax estimator (Schmidt 1968)
M, phi, phi_err = ss.luminosity_function_vmax(cat, 0.1, 0.5, 22.5, M_bins, area_sr, cosmo)

# SWML estimator (Efstathiou, Ellis & Peterson 1988)
M, phi_sw, phi_sw_err = luminosity_function_swml(cat, 0.1, 0.5, 22.5, M_bins, area_sr, cosmo)

# C⁻ non-parametric cumulative estimator (Lynden-Bell 1971)
M_sorted, Phi, Phi_err = cumulative_luminosity_function_cminus(cat, 0.1, 0.5, 22.5, area_sr, cosmo)

# STY parametric Schechter fit
result = ss.fit_schechter_sty(cat, 0.1, 0.5, 22.5, M_bins, area_sr, cosmo)

logM_bins = np.linspace(9, 12, 20)
logM, smf, smf_err = ss.stellar_mass_function_vmax(cat, 0.1, 0.5, logM_bins, area_sr, cosmo)

# Efron–Petrosian independence test (τ ~ N(0,1) under H₀)
ep = efron_petrosian_tau(cat.abs_mag, cat.redshift, 22.5, cosmo)
# Rauzy completeness tests
rz = rauzy_completeness(cat.abs_mag, cat.redshift, 22.5, cosmo)

# --- Covariance ----------------------------------------------------------
cov_jk = ss.jackknife_covariance(estimator_func, cat, n_patches=20)
cov_bt = ss.bootstrap_covariance(estimator_func, cat, n_bootstrap=200)

# --- I/O (HDF5) ----------------------------------------------------------
writer = ss.SummaryStatWriter("results.h5")
writer.write_twopcf("angular_2pcf/sample_A",
                    theta, w, cov_jk, theta_bins,
                    estimator="landy-szalay", cosmo=cosmo, meta={})

reader = ss.SummaryStatReader("results.h5")
data = reader.read_twopcf("angular_2pcf/sample_A")

JAX cosmology utilities

All functions in sum_stat.cosmology are JIT-compiled and differentiable with respect to cosmological parameters:

from sum_stat.cosmology import comoving_distance_jax, critical_surface_density_jax
import jax

chi  = comoving_distance_jax(z=0.5, h=0.6736, omega_m=0.3111)   # [Mpc]
dchi = jax.grad(comoving_distance_jax)(0.5, 0.6736, 0.3111)     # d chi / d z

sigma_crit = critical_surface_density_jax(z_l=0.3, z_s=0.8,
                                          h=0.6736, omega_m=0.3111)  # [M_sun/pc^2]

Development

# tests
pytest tests/

# formatting / linting
black sum_stat/
ruff check sum_stat/

# documentation
cd docs && make html

Citation

If you use sum_stat in your research, please cite it as:

@software{comparat2026sumstat,
  author  = {Comparat, Johan},
  title   = {{sum\_stat}: Cosmological summary statistics from galaxy catalogues and lensing data},
  year    = {2026},
  version = {0.1.0},
  url     = {https://github.com/JohanComparat/sum_stat},
  license = {MIT},
}

A CITATION.cff file is also provided for automated citation tools (GitHub "Cite this repository" button).

Contributing

Contributions are welcome! Please read CONTRIBUTING.md for development setup, coding conventions, and how to submit a pull request.

License

MIT — see LICENSE.

Package structure

sum_stat/
├── sum_stat/
│   ├── catalogue.py       # GalaxyCatalogue, ShapeCatalogue, PhotoZCalibTable
│   ├── cosmology.py       # JAX-differentiable distances and Σ_crit
│   ├── covariance/        # jackknife_covariance, bootstrap_covariance
│   ├── io/                # SummaryStatWriter, SummaryStatReader
│   ├── lensing/           # delta_sigma, NFW model
│   ├── lf_smf/            # 1/Vmax, SWML, C⁻, STY fit, EP τ, Rauzy
│   ├── nz/                # nz_histogram, nz_kde
│   ├── powspec/           # cl_angular, pk3d, pk_multipoles
│   └── twopcf/            # w_theta, xi_r, wp, xi_multipoles
├── tests/
├── docs/
├── environment.yml
└── pyproject.toml