Freedom251 › Physics Simulation Software

MParT

Free plan APILinuxMac

What MParT is

MParT is a free toolkit for constructing and using monotone functions in measure transport and regression. Its monotone triangular transport maps can represent conditional distributions and accelerate posterior sampling in Bayesian inference. The project is a C++ library with bindings for Python, Julia, and Matlab, and emphasizes fast execution with parsimonious parameterizations for near-real-time work on low- and moderate-dimensional problems. Implementations can be used in higher-level libraries such as TransportMaps and MUQ. The C++ library and Python wrapper install through conda-forge, while Julia bindings use Julia’s package manager. Basic source builds are intended for most operating systems, with documented environment instructions for macOS and Linux. Building requires CMake 3.13 or later and dependencies including Kokkos, Eigen, Cereal, Pybind11, and Catch2. CUDA requires compiling from source, and only Python bindings currently provide CUDA GPU acceleration. Tutorials cover map construction, monotone least squares, Bayesian inference, stochastic volatility, and map-based density estimation. MParT uses the BSD 3-Clause License.

Who it suits

It suits researchers and developers working with measure transport, regression, or Bayesian inference. Users can work through Python, Julia, Matlab, or C++ and contribute through project discussions and pull requests.

What it does well

Provides Python, Julia, and Matlab bindings
Supports monotone transport maps
Installs through conda-forge and Julia’s package manager
Includes Bayesian inference tutorials
Uses the BSD 3-Clause License

What to know first

CUDA requires compiling from source
CUDA acceleration is currently Python-only
Source builds require CMake 3.13 or later
Dependencies are not tested in every version

Verdict

MParT focuses on transport maps and monotone-function modeling with bindings across several technical languages. Its installation routes and CUDA limitations are important considerations for deployment planning.

Facts from MParT's own pages

Applications: Its monotone triangular transport maps can model conditional distributions and accelerate posterior sampling in Bayesian inference. · measuretransport.github.io, 5 Oct 2026
Build requirements: MParT uses CMake and requires version 3.13 or later; the documentation says it has not been tested with every version of its dependencies. · measuretransport.github.io, 12 Oct 2026
Dependencies: The source build requires CMake version 3.13 or later; the documentation lists Kokkos, Eigen, Cereal, Pybind11, and Catch2 among its dependencies. · measuretransport.github.io, 5 Oct 2026
Examples: The project provides tutorials for C++, Python, Julia, and Matlab, including examples for Bayesian inference and map-based density estimation. · measuretransport.github.io, 12 Oct 2026
Free plan: Yes · measuretransport.github.io, 23 Sept 2026
GPU support: CUDA support requires compiling from source with an externally installed Kokkos and NVIDIA CUDA libraries. · measuretransport.github.io, 12 Oct 2026
Installation: The main C++ library and Python wrapper can be installed with conda, and Julia bindings can be installed through Julia's package manager. · measuretransport.github.io, 12 Oct 2026
Integrations: The project describes MParT as suitable for use in higher-level libraries such as TransportMaps and MUQ. · measuretransport.github.io, 12 Oct 2026
Intended users: The project invites users to contribute to development and documentation, including by forking the repository and submitting pull requests. · measuretransport.github.io, 12 Oct 2026
Languages: MParT is a C++ library with bindings for Python, Julia, and Matlab. · measuretransport.github.io, 12 Oct 2026
License: MParT is released under the BSD 3-Clause License, which permits redistribution and use in source and binary forms subject to its stated conditions. · measuretransport.github.io, 12 Oct 2026
Performance: The project emphasizes fast execution and parsimonious parameterizations intended to permit near real-time computation on low and moderate dimensional problems. · measuretransport.github.io, 12 Oct 2026
Platform support: The installation guide provides MacOS and Linux environment path instructions and says the basic source build should work on most operating systems. · measuretransport.github.io, 5 Oct 2026
Purpose: MParT is a multi-language toolkit for constructing and using monotone functions for measure transport and regression. · measuretransport.github.io, 12 Oct 2026
Support: Users can ask questions, discuss enhancements, report bugs, or request features through GitHub issues and discussions or by emailing the developers. · measuretransport.github.io, 12 Oct 2026
Tutorials: The examples cover map construction from densities and samples, monotone least squares, Bayesian inference, and stochastic volatility. · measuretransport.github.io, 5 Oct 2026
Use cases: Its monotone triangular transport maps can model conditional distributions and accelerate posterior sampling in Bayesian inference. · measuretransport.github.io, 12 Oct 2026

Where MParT stands on Freedom251

Physics Simulation Software#68 of 89