FastRerandomize: Fast Rerandomization Using Accelerated Computing

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Goldstein, Rebecca, Jerzak, Connor T., Kamat, Aniket, Zhu, Fucheng Warren
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909983906463744
author Goldstein, Rebecca
Jerzak, Connor T.
Kamat, Aniket
Zhu, Fucheng Warren
author_facet Goldstein, Rebecca
Jerzak, Connor T.
Kamat, Aniket
Zhu, Fucheng Warren
contents We present fastrerandomize, an R package for fast, scalable rerandomization in experimental design. Rerandomization improves precision by discarding treatment assignments that fail a prespecified covariate-balance criterion, but existing implementations can become computationally prohibitive as the number of units or covariates grows. fastrerandomize introduces three complementary advances: (i) optional GPU/TPU acceleration to parallelize balance checks, (ii) memory-efficient key-only storage that avoids retaining full assignment matrices, and (iii) auto-vectorized, just-in-time compiled kernels for batched candidate generation and inference. This approach enables exact or Monte Carlo rerandomization at previously intractable scales, making it practical to adopt the tighter balance thresholds required in modern high-dimensional experiments while simultaneously quantifying the resulting gains in precision and power for a given covariate set. Our approach also supports randomization-based testing conditioned on acceptance. In controlled benchmarks, we observe order-of-magnitude speedups over baseline workflows, with larger gains as the sample size or dimensionality grows, translating into improved precision of causal estimates.
format Preprint
id arxiv_https___arxiv_org_abs_2501_07642
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FastRerandomize: Fast Rerandomization Using Accelerated Computing
Goldstein, Rebecca
Jerzak, Connor T.
Kamat, Aniket
Zhu, Fucheng Warren
Computation
62K10, 65C60
G.3; G.4
We present fastrerandomize, an R package for fast, scalable rerandomization in experimental design. Rerandomization improves precision by discarding treatment assignments that fail a prespecified covariate-balance criterion, but existing implementations can become computationally prohibitive as the number of units or covariates grows. fastrerandomize introduces three complementary advances: (i) optional GPU/TPU acceleration to parallelize balance checks, (ii) memory-efficient key-only storage that avoids retaining full assignment matrices, and (iii) auto-vectorized, just-in-time compiled kernels for batched candidate generation and inference. This approach enables exact or Monte Carlo rerandomization at previously intractable scales, making it practical to adopt the tighter balance thresholds required in modern high-dimensional experiments while simultaneously quantifying the resulting gains in precision and power for a given covariate set. Our approach also supports randomization-based testing conditioned on acceptance. In controlled benchmarks, we observe order-of-magnitude speedups over baseline workflows, with larger gains as the sample size or dimensionality grows, translating into improved precision of causal estimates.
title FastRerandomize: Fast Rerandomization Using Accelerated Computing
topic Computation
62K10, 65C60
G.3; G.4
url https://arxiv.org/abs/2501.07642