Bayesian optimization with Gaussian process surrogate model for geoacoustic inversion and parameter estimation

This repository contains code used to perform acoustic parameter estimation using Bayesian optimization with a Gaussian process surrogate model. The following papers use this code:

William Jenkins, Peter Gerstoft, and Yongsung Park, “Bayesian optimization with Gaussian process surrogate model for source localization,” J Acoust. Soc. Am., vol. 154, no. 3, pp. 1459–1470, Sep. 2023, doi: 10.1121/10.0020839.

W. F. Jenkins and P. Gerstoft, “Bayesian Optimization with Gaussian Processes for Robust Localization,” in 2024 IEEE International Conference on Acoustics, Speech and Signal Processing, Seoul, Republic of Korea, Apr. 2024, pp. 6010–6014. doi: 10.1109/ICASSP48485.2024.10445808.

W. F. Jenkins, P. Gerstoft, and Y. Park, “Geoacoustic inversion using Bayesian optimization with a Gaussian process surrogate model,” J. Acoust. Soc. Am., vol. 156, no. 2, pp. 812–822, Aug. 2024, doi: 10.1121/10.0028177.

Installation

In your desired target directory, run the following command:

git clone [email protected]:NeptuneProjects/BOGP.git

Once cloned, build the Conda environment. This may take a few minutes. Two dependencies, TritonOA and OAOptimization, are automatically installed via pip from their respective GitHub repositories.

conda env create -f gp_dev.yml

Activate the environment:

conda activate gp310

Usage

This workflow applies to multiple projects and data sets. Specific instructions for running the workflow on a particular data set are provided in the corresponding README.md files:

Application	Data	Instructions
Acoustic source localization	SWellEx-96	README.md
Source localization robust to array tilt	SWellEx-96	README.md
Geoacoustic inversion	SWellEx-96	README.md