Optimal detection of multi-sample aligned sparse signals

TL;DR

This paper establishes the fundamental detection boundary for multi-sample aligned sparse signals and develops optimal tests that adapt to the signal's sparsity and scale, unifying several detection problems.

Contribution

It introduces optimal detection methods for aligned sparse signals, characterizing the boundary between detectability and nondetectability, and unifies multiscale and sparse mixture testing.

Findings

Detection boundary depends on sequence length to signal length ratio.

Optimal tests achieve detectability at the critical boundary.

Detection difficulty remains stable unless the ratio grows exponentially.

Abstract

We describe, in the detection of multi-sample aligned sparse signals, the critical boundary separating detectable from nondetectable signals, and construct tests that achieve optimal detectability: penalized versions of the Berk-Jones and the higher-criticism test statistics evaluated over pooled scans, and an average likelihood ratio over the critical boundary. We show in our results an inter-play between the scale of the sequence length to signal length ratio, and the sparseness of the signals. In particular the difficulty of the detection problem is not noticeably affected unless this ratio grows exponentially with the number of sequences. We also recover the multiscale and sparse mixture testing problems as illustrative special cases.