A Practical Framework for Estimating the Repetition Likelihood of Fast Radio Bursts from Spectral Morphology

TL;DR

This work addresses the challenge of distinguishing FRB repeaters from apparent non-repeaters by introducing a physically interpretable probabilistic framework based on spectral morphology. Using t-SNE for nonlinear dimensionality reduction and HDBSCAN for clustering on an extended CHIME/FRB sample, the authors find that spectral running $r$ and spectral index $\gamma$ are the primary discriminants, achieving an AUC of $0.89$ in separating the two populations. They map repetition likelihood in the $\gamma$–$r$ plane, revealing a gradient from Region I (≈65% repeater probability) to Region IV (≈5%), and combine this with Gaussian Mixture Model posteriors to prioritize high-probability repeater candidates. The framework suggests FRBs may be manifestations of a single population under different geometric and propagation conditions, and it provides a practical, rapid tool for guiding follow-up observations and FRB progenitor studies, including identification of high-priority non-repeaters for monitoring.

Abstract

The repeating behavior of fast radio bursts (FRBs) is regarded as a key clue to understanding their physical origin, yet reliably distinguishing repeaters from apparent non-repeaters with current observations remains challenging. Here we propose a physically interpretable and practically quantifiable classification framework based on spectral morphology. Using dimensionality reduction, clustering, and feature-importance analysis, we identify the spectral running $r$ and spectral index $γ$ as the most critical parameters for distinguishing repeaters from apparent non-repeaters in the CHIME/FRB sample. In the $γ$-$r$ space, repeaters preferentially occupy regions with steeper, narrower-band spectra, whereas non-repeaters cluster in flatter, broader-band regions, resulting in a clear density separation. We further construct an empirical probability map in the $γ$-$r$ space, showing a clear gradient of repetition likelihood, from $\sim 65\%$ in the high-repetition region to $\sim 5\%$ in the low-repetition region. Combining this with Gaussian Mixture Model posterior analysis, we identify several apparent non-repeaters with high inferred repetition probability, recommending them as priority targets for future monitoring. This framework provides a simple and generalizable tool for assessing repeatability in the CHIME/FRB sample and highlights the diagnostic power of spectral morphology in unveiling FRB origins.

Figures (4)

• Figure 1: The embedding space of the t-SNE dimension reduction results and HDBSCAN clustering of FRBs in the CHIME/FRB Catalog 1 and Catalog 2023. Blue squares represent apparent non-repeaters, while red triangles denote repeaters. The blue and red contours indicate the clusters identified by HDBSCAN, corresponding to the Nonrepeater-like cluster and Repeater-like cluster, respectively.
• Figure 2: SHAP values for predictions in FRB classification. Features are ranked by overall importance, with the horizontal axis indicating the SHAP value, reflecting each feature’s impact on the model output.
• Figure 3: Distribution of repeaters (red) and apparent non-repeaters (blue) in the $\gamma$-$r$ parameter space, with red and blue contours overlaid to represent the density levels of each population. Marginal histograms show the distributions of bandwidth, pulse width, and burst rate, with error bars indicating the standard errors of the means.
• Figure 4: Distribution of repeaters (red) and apparent non-repeaters (blue) in the $\gamma$-$r$ parameter space. The background shading indicates regions of different inferred repetition probabilities, with annotated values representing the fraction of known repeaters in each quadrant. FRB 20201124A and FRB 20121102A are highlighted as cyan and yellow diamonds, respectively.