integratedDelta_calibration

Scripts for calibrating simulation data to experimental observations using an embedded discrepancy Gaussian Process (GP) emulator.

---

Overview

This repository implements a Bayesian calibration framework that:

• Builds a GP emulator of simulator outputs
• Embeds a model discrepancy term
• Calibrates simulation parameters to observational data using MCMC

The workflow is driven by main.py and configured via config.json.

---

Environment Setup

This project uses a Python virtual environment.

1. Create a virtual environment

python3 -m venv venv

2. Activate the environment

macOS/Linux

source venv/bin/activate

Windows

venv\Scripts\activate

3. Install required packages

pip install -r requirements.txt

Core dependencies include:

• numpy
• scipy
• matplotlib
• pandas
• scikit-learn

---

Running the Calibration

To execute the integrated delta calibration routine:

python3 main.py

All runtime behavior is controlled through config.json.

---

Repository Structure

integratedDelta_calibration/
│
├── main.py
├── config.json
├── requirements.txt
│
├── modelData/
│   ├── appDomain.txt
│   ├── modelPredictions.txt
│   └── thetaVals.txt
│
└── observationData/
    ├── appDomain.txt
    └── observationData.txt

---

Input Data Format

Input data is organized into two subdirectories:

/modelData

Simulator outputs to be emulated and calibrated.

• appDomain.txt — Application domain locations
• modelPredictions.txt — Simulator outputs
• thetaVals.txt — Parameter values corresponding to simulations

/observationData

Experimental data to be matched.

• appDomain.txt — Observation domain locations
• observationData.txt — Measured responses

Formatting Requirements

All .txt files must:

• Be column-formatted
• Include a header row labeling each column
• Use the delimiter specified in config.json

The provided template files illustrate formatting for MD (observational) and DDD (simulator) predictions of
$\tau^{CRSS}$.

---

Configuration

All calibration settings are controlled via config.json.

calibration_settings

Controls MCMC behavior:

• N_mcmc — Total MCMC iterations
• N_burn — Burn-in samples
• N_samp_post — Posterior samples retained

input_settings

• Input file delimiter

output_settings

• Controls which plots and diagnostics are generated

results_path & results_options

(In development)
Controls output of serialized model objects for downstream analysis.

---

Notes

• The virtual environment directory (venv/) should not be committed to the repository.
• Only requirements.txt is required to reproduce the environment.

Citation

If you use this repository or the associated methodology in your research, please cite the following paper.

Bayesian Model Calibration with Integrated Discrepancy: Addressing Inexact Dislocation Dynamics Models Liam Myhill, Enrique Martinez Saez, Sez Russcher (2026)

Paper: https://arxiv.org/abs/2603.11960

@misc{myhill2026bayesianmodelcalibrationintegrated,
  title   = {Bayesian Model Calibration with Integrated Discrepancy: Addressing Inexact Dislocation Dynamics Models},
  author  = {Myhill, Liam and Martinez Saez, Enrique and Russcher, Sez},
  year    = {2026},
  eprint  = {2603.11960},
  archivePrefix = {arXiv},
  primaryClass  = {stat.ME},
  url     = {https://arxiv.org/abs/2603.11960}
}

---

Repository Citation

This repository includes a CITATION.cff file so that GitHub can generate a "Cite this repository" button automatically.