RECAP: Representation-Based Calibration with Amortized Pooling for Frequentist Simulation-Based Inference
Abstract
Simulation-based inference enables uncertainty quantification in broad contexts when a simulation model is available but when a test statistic's sampling distribution is not tractable. But, frequentist calibration in this setting can be costly over high-dimensional parameter spaces. In this paper, we introduce RECAP (REpresentation-based Calibration with Amortized Pooling), a general framework for frequentist simulation-based inference. RECAP exploits low-dimensional structure in how the null distribution varies with parameters and is particularly useful when the parameter dimension is high or when different parameter configurations have redundant or compensating effects. RECAP pools simulated scores according to a representation that preserves this distribution exactly or approximately, allowing distant parameter values to share calibration information. Representations may be derived from model structure or learned independently. Empirical quantiles of the pooled scores then yield tests and confidence sets, for which we establish theoretical coverage guarantees. In the simulation studies, we show that RECAP has lower coverage error than existing approaches and power close to that of the point-wise oracle, while maintaining the computational benefits of amortized calibration. An application to population genomics illustrates learned low-dimensional calibration and interpretable uncertainty quantification.
Related
- Amortized ratio-estimation importance sampling and localized simulation-based calibration for intractable likelihoods
- Amortized Localized Posterior Inference for Model Calibration with Application to an Ocean Turbulence Closure
- Mitigating Representation Gaps in Amortized Bayesian Inference with Auxiliary Supervision
- Coverage is not enough: Frequentist tests of simulation-based inference for primordial non-Gaussianity
- ENFR: Evidence Networks with Flow-Matched References for Amortized Simulation-Based Model Inference
BibTeX
@misc{tang2026recap,
title = {{RECAP: Representation-Based Calibration with Amortized Pooling for Frequentist Simulation-Based Inference}},
author = {Runshi Tang and Genevera I. Allen},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.28332},
eprint = {2610.28332},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.28332},
}
RIS
TY - UNPB
TI - RECAP: Representation-Based Calibration with Amortized Pooling for Frequentist Simulation-Based Inference
AU - Runshi Tang
AU - Genevera I. Allen
PY - 2026
DA - 2026/10/20
PB - arXiv
DO - 10.48550/arxiv.2610.28332
UR - https://arxiv.org/abs/2610.28332
AB - Simulation-based inference enables uncertainty quantification in broad contexts when a simulation model is available but when a test statistic's sampling distribution is not tractable. But, frequentist calibration in this setting can be costly over high-dimensional parameter spaces. In this paper, we introduce RECAP (REpresentation-based Calibration with Amortized Pooling), a general framework for frequentist simulation-based inference. RECAP exploits low-dimensional structure in how the null distribution varies with parameters and is particularly useful when the parameter dimension is high or when different parameter configurations have redundant or compensating effects. RECAP pools simulated scores according to a representation that preserves this distribution exactly or approximately, allowing distant parameter values to share calibration information. Representations may be derived from model structure or learned independently. Empirical quantiles of the pooled scores then yield tests and confidence sets, for which we establish theoretical coverage guarantees. In the simulation studies, we show that RECAP has lower coverage error than existing approaches and power close to that of the point-wise oracle, while maintaining the computational benefits of amortized calibration. An application to population genomics illustrates learned low-dimensional calibration and interpretable uncertainty quantification.
ER -