MEPSA: a flexible peak search algorithm designed for uniformly spaced time series

TL;DR

MEPSA is a flexible, noise-robust peak detection algorithm for uniformly spaced time series, adaptable to various astrophysical data and validated against simulations and existing methods.

Contribution

It introduces a novel, adaptable peak search algorithm that improves detection flexibility and can be tailored to different data types without code modification.

Findings

Validated with simulated GRB profiles and featureless data

Compared favorably with Li and Fenimore's algorithm

C code implementation is publicly available

Abstract

We present a novel algorithm aimed at identifying peaks within a uniformly sampled time series affected by uncorrelated Gaussian noise. The algorithm, called "MEPSA" (multiple excess peak search algorithm), essentially scans the time series at different timescales by comparing a given peak candidate with a variable number of adjacent bins. While this has originally been conceived for the analysis of gamma-ray burst light (GRB) curves, its usage can be readily extended to other astrophysical transient phenomena, whose activity is recorded through different surveys. We tested and validated it through simulated featureless profiles as well as simulated GRB time profiles. We showcase the algorithm's potential by comparing with the popular algorithm by Li and Fenimore, that is frequently adopted in the literature. Thanks to its high flexibility, the mask of excess patterns used by MEPSA can be tailored and optimised to the kind of data to be analysed without modifying the code. The C code is made publicly available.