Optimal Experiments for Partial Causal Effect Identification
Abstract
Causal queries are often only partially identifiable from observational data, and experiments that could tighten the resulting bounds are typically costly. We study the problem of selecting, prior to observing experimental outcomes, a cost-constrained subset of experiments that maximally tightens bounds on a target query. We formalize this as the max-potency problem, where epistemic potency measures the worst-case reduction in bound width guaranteed by an experiment, and show that this problem is NP-hard via a reduction from 0-1 knapsack. Building on the polynomial-programming framework of Duarte et al. (2023), we give a general procedure for evaluating epistemic potency in discrete settings. To control the super-exponential search space, we introduce two graphical pruning criteria that depend only on the causal graph and the query: a novel path-interception rule that exploits district structure to certify zero potency in linear time, and an identifiability check based on the ID algorithm. On Erdos-Renyi random graphs and 11 bnlearn benchmark networks, the two criteria together prune 50-88% of candidate experiments on average without solving a single polynomial program. For the general subset search, we show that ID-pruned experiments are combinatorially inert, yielding a super-exponential reduction in the number of subsets evaluated. We close with an end-to-end demonstration on observational NHANES data, selecting optimal experiments for estimating the effect of physical activity on diabetes.
Related
- Partial identification with entropy regularized optimal transport
- Identifying Causal Effects in Experiments with Latent Treatments: An Instrumental Variable Approach
- When AI Generates Covariates: Causal Typing and Estimand Drift in Sequential Experiments
- EARL: Exposure- and Allocation-Reweighted Linear Estimator for Bipartite Experiments with Partial Assignment
- Tracing the Loop: Non-Causal Computation, Partial Traces, & Postselected Entanglement
BibTeX
@misc{maringgele2026optimal,
title = {{Optimal Experiments for Partial Causal Effect Identification}},
author = {Tobias Maringgele and Jalal Etesami},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2605.06993},
eprint = {2605.06993},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.06993},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Optimal Experiments for Partial Causal Effect Identification
AU - Tobias Maringgele
AU - Jalal Etesami
PY - 2026
DA - 2026/10/27
PB - arXiv
DO - 10.48550/arxiv.2605.06993
UR - https://arxiv.org/abs/2605.06993
AB - Causal queries are often only partially identifiable from observational data, and experiments that could tighten the resulting bounds are typically costly. We study the problem of selecting, prior to observing experimental outcomes, a cost-constrained subset of experiments that maximally tightens bounds on a target query. We formalize this as the max-potency problem, where epistemic potency measures the worst-case reduction in bound width guaranteed by an experiment, and show that this problem is NP-hard via a reduction from 0-1 knapsack. Building on the polynomial-programming framework of Duarte et al. (2023), we give a general procedure for evaluating epistemic potency in discrete settings. To control the super-exponential search space, we introduce two graphical pruning criteria that depend only on the causal graph and the query: a novel path-interception rule that exploits district structure to certify zero potency in linear time, and an identifiability check based on the ID algorithm. On Erdos-Renyi random graphs and 11 bnlearn benchmark networks, the two criteria together prune 50-88% of candidate experiments on average without solving a single polynomial program. For the general subset search, we show that ID-pruned experiments are combinatorially inert, yielding a super-exponential reduction in the number of subsets evaluated. We close with an end-to-end demonstration on observational NHANES data, selecting optimal experiments for estimating the effect of physical activity on diabetes.
ER -