mlqmcpy 0.1.0

Multilevel quasi-Monte Carlo in Python.

mlqmcpy

Multilevel (quasi-)Monte Carlo methods in Python.

mlqmcpy provides iterators and example problems for estimating expectations with Monte Carlo, quasi-Monte Carlo, multilevel Monte Carlo, and multilevel quasi-Monte Carlo methods.

Installation

pip install mlqmcpy

Quick start

from mlqmcpy import GreedyMLQMCIterator
from mlqmcpy.problems.analytic import analytic

iterator = GreedyMLQMCIterator(2, error_tolerance=1e-2, seed=1234)

for new_samples in iterator:
    new_results = {
        level: analytic.ml(level, samples)
        for level, samples in new_samples.items()
    }
    iterator.update(new_results)

print(iterator.mean)
print(iterator.standard_error)

Development

This project uses uv for dependency management. As a library, mlqmcpy does not commit a lockfile; development and CI resolve from pyproject.toml.

uv sync --group dev --refresh
uv run pytest
uv run ruff check mlqmcpy tests
uv run black --check mlqmcpy tests
uv run isort --check-only mlqmcpy tests
uv build

Releases are published from GitHub Releases through PyPI Trusted Publishing. Configure the PyPI trusted publisher for repository PieterjanRobbe/mlqmcpy, workflow .github/workflows/release.yml, and environment pypi.