Instance-dependent sequential limits as an exact information ledger
Abstract
Instance-dependent limits for bandit regret, sequential tests, best-arm identification and Markov decision processes are proved by one chain. It runs from a divergence decomposition through data processing to a statistic, a testing step that turns a class guarantee into a requirement on the history, and optimization over alternatives and allocations. Only the first step is an identity. This work writes the chain as an exact ledger. By the chain rule, data processing loses exactly a conditional divergence, which vanishes for arms in natural exponential families at the per-arm counts and reward sums. A one-parameter family of martingale identities extends the decomposition to every tilt between two instances. The classical limits follow by relaxations with named defects. All defects but the asymptotic step's can be computed from an algorithm's own runs, the uniform quantification's only when the class guarantee is a number. On seven published sources, from movie ratings to engine replacement, the ledger locates the difference between standard algorithms and the classical limits. For identification among trained models and for Wald's test the largest term is usually the uniform quantification's.
Related
- The Certification Exponent: Instance-Optimal Sequential Certification of Policy Improvement from History Dependent Logs
- Sequential Certification of Partially Identified Product Targets: Information Limits and Adaptive Sampling
- Towards instance-dependent regret optimality in Episodic MDPs with Posterior Sampling
- Instance-dependent Stochastic Lipschitz bandit
- Randomized Adaptive Sequential Allocation and the Limits of Bounded Exposure to Inferior Treatments
BibTeX
@misc{balsubramani2026instancedependent,
title = {{Instance-dependent sequential limits as an exact information ledger}},
author = {Akshay Balsubramani},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.42133},
eprint = {2610.42133},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.42133},
}
RIS
TY - UNPB
TI - Instance-dependent sequential limits as an exact information ledger
AU - Akshay Balsubramani
PY - 2026
DA - 2026/10/30
PB - arXiv
DO - 10.48550/arxiv.2610.42133
UR - https://arxiv.org/abs/2610.42133
AB - Instance-dependent limits for bandit regret, sequential tests, best-arm identification and Markov decision processes are proved by one chain. It runs from a divergence decomposition through data processing to a statistic, a testing step that turns a class guarantee into a requirement on the history, and optimization over alternatives and allocations. Only the first step is an identity. This work writes the chain as an exact ledger. By the chain rule, data processing loses exactly a conditional divergence, which vanishes for arms in natural exponential families at the per-arm counts and reward sums. A one-parameter family of martingale identities extends the decomposition to every tilt between two instances. The classical limits follow by relaxations with named defects. All defects but the asymptotic step's can be computed from an algorithm's own runs, the uniform quantification's only when the class guarantee is a number. On seven published sources, from movie ratings to engine replacement, the ledger locates the difference between standard algorithms and the classical limits. For identification among trained models and for Wald's test the largest term is usually the uniform quantification's.
ER -