X-ray transients in the Chandra archive: Introducing the cumulative distribution discriminator (CuDiDi)

TL;DR

This paper introduces CuDiDi, a new statistical classifier for systematically detecting diverse X-ray transients in Chandra data, significantly expanding the known sample of short-lived high-energy astrophysical events.

Contribution

The paper presents a novel detection pipeline incorporating CuDiDi, enabling the identification of a broad range of X-ray transients and extending the catalog to include shorter, previously undetected events.

Findings

Identified 765 high-confidence X-ray transients across the sky.

CuDiDi detects most known transients with stricter variability criteria.

Extended the catalog of Chandra transients to include shorter-duration events.

Abstract

X-ray transients on sub-observation timescales represent a diverse and underexplored class of astrophysical phenomena, from stellar flares and magnetar bursts to extragalactic fast transients and supernova shock breakouts. We present a systematic search for such events across 20,212 Chandra ACIS observations using a new detection pipeline that combines source identification, light-curve analysis, catalogue cross-matching, and a novel statistical classifier, the cumulative distribution discriminator (CuDiDi). From 1420 initial candidates, we identified a high-confidence golden sample of 765 transients spanning a broad range of timescales, fluxes, and spectral shapes. The candidates are distributed across the whole sky and show a wide range of durations with a median of 10 ks. A subset of fast events lasting < 30 s displays very soft spectra and is likely due to flaring dwarf stars, although extragalactic phenomena cannot be ruled out for all of them. The comparison with previously published samples showed that CuDiDi identifies most known transients while imposing somewhat stricter variability criteria, and it also extends the total sample of Chandra transients to include shorter events. We deliver a comprehensive catalogue of sub-observation Chandra X-ray transients and establish a general method for exploiting archival datasets to uncover rare short-lived high-energy phenomena.