Computer code validation via mixture model estimation
Abstract
When computer codes model complex physical systems, calibration alone is insufficient; one must also assess whether a discrepancy term is needed. In this paper, we study computer code validation through a Bayesian mixture approach that compares a pure-code model with a discrepancy-corrected model. The method relies on the posterior distribution of a mixture weight, which measures the relative support of the two competing distributions. Under the assumption that the code is linear in the calibration parameters, or can be well approximated by a linear surrogate, we show that mixture component-shared parameters can be used to combine flexible modeling, even when noninformative priors are assigned to some common parameters. Inference is performed using a Metropolis-within-Gibbs algorithm. In addition, we introduce a thresholded allocation rule that complements the global mixture weight by providing a local diagnostic of where the discrepancy-corrected component is truly needed along the input domain. Beyond global model comparison, the proposed approach is also able to perform local model discrimination by identifying where the discrepancy-corrected component is truly needed along the input domain.
Related
- Sandwich Variance Estimation in Scale Mixture of Skew-Normal Linear Mixed Models.
- Brown's loss function and estimation of the scale parameter and the entropy in scale mixture models with unknown location
- Mixture-based Nonparametric Estimation of Spatial Covariance Functions with Applications to HIV Key Population Size Estimation across Sub-Saharan Africa
- Joint Estimation and Feature Selection in Overspecified Finite Mixture of Regressions
- Weak Limits of Parameter Estimation in Strongly Identifiable Mixture Models
BibTeX
@misc{soleimani2026computer,
title = {{Computer code validation via mixture model estimation}},
author = {Negar Soleimani and Pierre Barbillon and Kaniav Kamary and Merlin Keller},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.27619},
eprint = {2609.27619},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.27619},
}
RIS
TY - UNPB
TI - Computer code validation via mixture model estimation
AU - Negar Soleimani
AU - Pierre Barbillon
AU - Kaniav Kamary
AU - Merlin Keller
PY - 2026
DA - 2026/09/23
PB - arXiv
DO - 10.48550/arxiv.2609.27619
UR - https://arxiv.org/abs/2609.27619
AB - When computer codes model complex physical systems, calibration alone is insufficient; one must also assess whether a discrepancy term is needed. In this paper, we study computer code validation through a Bayesian mixture approach that compares a pure-code model with a discrepancy-corrected model. The method relies on the posterior distribution of a mixture weight, which measures the relative support of the two competing distributions. Under the assumption that the code is linear in the calibration parameters, or can be well approximated by a linear surrogate, we show that mixture component-shared parameters can be used to combine flexible modeling, even when noninformative priors are assigned to some common parameters. Inference is performed using a Metropolis-within-Gibbs algorithm. In addition, we introduce a thresholded allocation rule that complements the global mixture weight by providing a local diagnostic of where the discrepancy-corrected component is truly needed along the input domain. Beyond global model comparison, the proposed approach is also able to perform local model discrimination by identifying where the discrepancy-corrected component is truly needed along the input domain.
ER -