This repository implements the eSCP algorithm given in ¹ for the purpose of running several experiments detailed in a paper² that studies the continuity of this algorithm. The three modules 'src/example_{one, two, three}.py' fully reproduce the plots from ². Details of the examples are explained in the paper.

The algorithm is currently written for 2-dimensions only, and a forthcoming repository called 'escp-constrained' will extend this code for arbitrary input and output dimensions among other things. If you need to re-use or adapt the code for the SCP, I recommend consulting the other repository instead.

Getting Started

1. Install uv, a package manager. See instructions here.

2. Clone the repository and navigate to it.

    git clone https://github.com/akprasadan/escp-continuity.git
   
    cd escp-continuity
   

3. Install the project and its dependencies.

    uv sync
   

Usage

Now you can run modules with uv in a virtual environment, e.g.,

    uv run python -m src.example_one

    uv run python -m tests.check_pushforward

You can also run all unit tests.

    uv run pytest

Directory Structure

• The core functions and examples are stored in src/. You only need to run each of the src/example_{i}.py files to obtain all the figures from the paper. Use the functions in src/estimate_functions.py to create your own examples. The rest of the modules in src/ just contain helper functions.
• Necessary datasets will be stored in data/ after running either src/example_three.py or tests/check_concrete_pushforward.py. These modules will download a concrete dataset and store the raw version in data/raw_concrete.csv and a processed version in data/processed_concrete.csv. The original dataset is due to ³ and can also be downloaded from the UC Irvine Machine Learning Repository.
• Results from running the examples or certain tests are stored in plots/.
• The code is tested in tests/. Three tests, tests/check_pushforward.py, tests/check_concrete_pushforward.py, tests/test_probs_to_mesh.py include visual tests and must be manually run. Their results are stored in plots/tests/.

License

MIT

Footnotes

1. Shi, H., Yang, L., Chi, J., Butler, T., Wang, H., Bingham, D., & Estep, D. (2026). Nonparametric Bayesian Calibration of Computer Models. arXiv preprint arXiv:2509.22597. ↩

2. Prasadan A., Bingham, D., & Estep, D. (2026). Continuity of the Solution of a Non-Parametric Bayesian Statistical Calibration Procedure. arXiv preprint arXiv:2603.20665. ↩ ↩²

3. Yeh, I. C. (1998). Modeling of strength of high-performance concrete using artificial neural networks. Cement and Concrete research, 28(12), 1797-1808. ↩