MFTT¶

Multifidelity Triangular Transport

Construct monotone triangular transport maps from multifidelity information.

MFTT follows the notation and method names of the manuscript Multifidelity Formulations for Triangular Transport (arXiv:2610.00698, 2026).

Methods¶

Single-fidelity

Learn one monotone triangular transport directly from samples of a target distribution.

Hierarchical multifidelity

Compose stage maps across an ordered fidelity hierarchy using changing- or fixed-reference training.

Nonhierarchical multifidelity

Combine multiple low-fidelity peer maps inside a coupled high-fidelity transport.

Clone the repository and install MFTT into an isolated Python 3.10 or newer environment:

git clone https://github.com/sandialabs/multifidelity-triangular-transport.git
cd multifidelity-triangular-transport
python -m pip install .

See the installation guide for environment setup and verification instructions.

Mathematical notation

Transport conventions and symbols used throughout MFTT.

Shared operations

Evaluation, inversion, sampling, and density operations.

API reference

Generated reference for the supported public interface.

Citation

Software authorship and the associated manuscript.