3ptWL-mod: Weak-Lensing Three-Point Modeling

3ptWL-mod computes theoretical multipoles of the three-point correlation function (3PCF) of projected scalar fields, with a focus on weak-lensing convergence. The C pipeline projects matter-bispectrum models into (\zeta_m(\theta_1,\theta_2)) and supports perturbation-theory, effective-field-theory, and Takahashi/Halo-model inspired branches.

The repository provides:

• the wlcf command-line executable;
• the libwlcf.a static library;
• the wlcfpy Cython wrapper;
• notebooks for 3PCF visualization and neural-network emulation.

The executable and Python package retain their historical wlcf names for API compatibility. The repository name reflects the scientific role of the code.

Documentation: 3ptWL-mod on Read the Docs

Install the Python package

3ptWL-mod builds a native extension, so install a C compiler, GSL, FFTW3, make, and the Python headers first. On Debian or Ubuntu:

sudo apt-get update
sudo apt-get install build-essential libgsl-dev libfftw3-dev python3-dev pkg-config
python3 -m pip install 3ptWL-mod

The distribution is named 3ptWL-mod; the import remains wlcfpy for compatibility:

python3 -c "from wlcfpy import wlcf; print(wlcf)"

Build the command-line tools from source

To also build the wlcf executable and libwlcf.a static library, clone the repository and build all public interfaces:

git clone https://github.com/sadirs/3ptWL-mod.git
cd 3ptWL-mod
make clean
make PYTHON=python3 all

Run a compact validation calculation:

./wlcf rootDir=Output_quick prefix=quick_ \
  fnamePS=./input/linear_pk_Takahashi_z0.txt \
  numberThreads=1 verbose=0 verbose_log=0 \
  mMax=2 Nell=32 chiQuadSteps=40 GLpoints=24 writevectors=false

All generated files are written beneath rootDir. File names use prefix where applicable.

Python wrapper

After installing with pip or running make all:

from wlcfpy import wlcf

model = wlcf()
model.set({
    "rootDir": "Output_python",
    "prefix": "python_",
    "fnamePS": "./input/linear_pk_Takahashi_z0.txt",
    "tree_level": 4,
    "mMax": 2,
    "Nell": 32,
    "chiQuadSteps": 40,
    "GLpoints": 24,
    "numberThreads": 1,
    "verbose": 0,
    "verbose_log": 0,
    "writevectors": False,
})
model.Run()
model.clean_all()

Tutorials and emulator

The curated workflows live under tests/:

• example.ipynb runs the model and visualizes 3PCF multipoles;
• emulator.ipynb generates a cosmology grid and trains neural-network surrogates;
• use_wlcf_emulator.ipynb demonstrates accelerated inference;
• firecrown_emulator_likelihood.ipynb provides an experimental likelihood integration scaffold;
• emulator.py contains the reusable data-generation and training helpers.

The emulator is an optional research workflow, not a replacement for convergence testing against direct 3ptWL-mod calculations. Its design follows the broader neural-network acceleration strategy demonstrated by Sadi Ramirez et al. in Full shape cosmology analysis from BOSS in configuration space using neural network acceleration, JCAP 2024 (08) 049.

Repository layout

• source/, include/, main/: numerical C implementation and entry point.
• getparam/: command-line and parameter-file parser.
• general_lib/: shared low-level C utilities.
• python/: Cython wrapper and generated-declaration template.
• input/: example power spectra and lensing kernels.
• tests/: notebooks, emulator helpers, and validation inputs.
• docs/: Sphinx and manual-page sources.
• addons/: CLASS-style parser, Cython, and optional build components.

Documentation

python3 -m pip install -r docs/requirements.txt
make -C docs html

Open docs/_build/html/index.html. Read the Docs uses the repository-level .readthedocs.yaml configuration and treats Sphinx warnings as build errors.

Citation

For the 3PCF modeling framework, cite:

• Abraham Arvizu et al., Modeling the 3-point correlation function of projected scalar fields on the sphere, JCAP 12 (2024) 049.

When the neural-network workflow is material to the analysis, also cite:

• Sadi Ramirez, Miguel Icaza-Lizaola, Sebastien Fromenteau, Mariana Vargas-Magaña, and Alejandro Aviles, Full shape cosmology analysis from BOSS in configuration space using neural network acceleration, JCAP 2024 (08) 049.

Record the repository commit, compiler and dependency versions, and complete run parameters in scientific releases.

Contributors

The scientific code and documentation include contributions from Alejandro Aviles, Juan Carlos Hidalgo, Eladio Moreno, Gustavo Niz, Sadi Ramirez, Mario A. Rodriguez-Meza, Sofía Samario, and collaborators.

License and acknowledgements

3ptWL-mod is distributed under the MIT license. The project uses or builds on FFTLog, BiHaloFit/Takahashi models, Zeno, Numerical Recipes, GSL, and CLASS. Alejandro Aviles acknowledges support by grants UNAM PAPIIT IA101825 and SECIHTI CBF2023-2024-162.