A Multi-fidelity Estimator of the Expected Information Gain for Bayesian Optimal Experimental Design

Thomas E. Coons, Xun Huan · 2025

Read A Multi-fidelity Estimator of the Expected Information Gain for Bayesian Optimal Experimental Design by Thomas E. Coons, Xun Huan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-10845.txt.

Description

Thomas Coons and Xun Huan built a multi-fidelity estimator in 2025 for the expected information gain in Bayesian experimental design. Good design picks the conditions that teach the most, and under a Bayesian view that gain is the design score. For nonlinear models the score cannot be computed exactly and must be sampled, which means solving the physics many times over.

With a high-fidelity model, that cost becomes impractical. Their estimator blends cheap low-fidelity runs with a few costly high-fidelity ones inside a control variate frame. It stays unbiased for the high-fidelity mean and cuts variance for a fixed computing budget, in the design reported. Readers get a way to plan better experiments on realistic models without an unrealistic computing bill.

Catalog record

CREATORThomas E. Coons, Xun Huan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE84 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-10845

Browse the full library · About the Tower of Babel