Across five independent gamma-ray burst (GRB) catalogues (CGRO/BATSE, Swift/BAT, BeppoSAX/GRBM, Insight-HXMT/HE, and the full Fermi/GBM catalogue of about 3000 long GRBs), the distribution of the number of peaks per prompt-emission light curve is described only by a mixture of two exponentials (M2E): roughly 80% of bursts are 'peak-poor' (about 2.1 peaks on average) and 20% are 'peak-rich' (about 8 peaks on average), with parameters that are remarkably consistent across instruments of very different energy passbands and effective areas. The physical origin of this two-family structure is unknown: it may reflect two distinct classes of inner engines left behind by the collapsing star (for example a neutron star versus a black hole), a single engine class operating in two dynamical regimes (for example different jet magnetisation or dissipation mode), or an unidentified mixture of intrinsic and selection effects. The problem is to determine this origin: which physical distinction generates the two families, and what the dichotomy reveals about how GRB engines release their energy.

Background

Gamma-ray bursts (GRBs) are brief, extremely luminous flashes of gamma rays from cosmological distances. They are broadly divided by the duration statistic T90T_{90} (the time interval containing 5% to 95% of the total gamma-ray fluence) into short GRBs (T90<2T_{90} < 2 s), associated with compact-object mergers, and long GRBs, associated with the collapse of hydrogen-stripped massive stars (collapsars). In the collapsar picture, the collapsing core leaves behind a compact remnant — the 'inner engine', either a newborn neutron star (a highly magnetised one is called a magnetar) or a black hole — which launches an ultra-relativistic jet; the prompt emission, the initial gamma-ray phase lasting seconds to minutes, is thought to arise from dissipation inside the jet, e.g. via internal shocks (collisions between faster and slower shells of ejecta) or magnetic reconnection cascades such as in the ICMART model (internal-collision-induced magnetic reconnection and turbulence, where moderately magnetised shells collide and trigger runaway reconnection). The prompt light curve (LC, count rate versus time) is typically a sequence of peaks (pulses), ranging from single smooth pulses to highly complex spiky profiles.

A long-standing question is whether this variety of LCs reflects a common stochastic process, whose statistical characterisation would reveal how the engine releases energy. Peaks are identified with algorithms such as MEPSA (a peak-search algorithm for uniformly sampled time series affected by uncorrelated Gaussian noise, applied here at 64 ms resolution). Guidorzi et al. (2015) showed that the waiting times between prompt pulses and early X-ray flares follow a common stochastic process, and that long GRBs with many prompt peaks (N≥8N \geq 8) almost always lack later X-ray flares (23 of 25 bursts), suggesting bursts exhaust their episodic energy reservoir either early or late. Guidorzi et al. (2024, 'G24') then performed the first detailed modelling of the distribution of the number of peaks per long GRB across four catalogues (BATSE, Swift/BAT, BeppoSAX/GRBM, Insight-HXMT/HE), finding that only a mixture of two exponentials (M2E) fits: an exponential distribution describes the peak counts of each family, characterised by mean peak counts of 2.1±0.12.1 \pm 0.1 ('peak-poor', about 80% of bursts) and 8.3±1.08.3 \pm 1.0 ('peak-rich', about 20%). Maccary et al. (2024) confirmed this on the full Fermi/GBM catalogue (about 3000 long GRBs; peak-rich fraction 0.21±0.040.21 \pm 0.04, means 2.12±0.102.12 \pm 0.10 and 8±18 \pm 1), showed the peak count of a given burst is essentially detector-independent between GBM and Swift/BAT, and explored the Malmquist bias (the selection effect whereby, in a flux-limited sample, sources at larger distances are preferentially the intrinsically brighter and more variable ones, so a burst moved to higher redshift would be tagged with fewer peaks or missed altogether) by repeating the analysis at different peak signal-to-noise thresholds; a direct redshift-resolved test was impossible because only 154 bursts in the sample have spectroscopic redshifts.

The existence of two families is thus empirically robust, but its interpretation is not. G24 and Maccary et al. (2024) frame two readings: (a) two classes of inner engines — e.g. neutron-star versus black-hole remnants — with different intrinsic activity, or (b) one engine class in two dynamical regimes, such as jets of different magnetisation (the degree to which the outflow energy is carried by magnetic fields, which affects jet mixing and baryon loading) or different dissipation modes. G24 found that peak-rich bursts are on average more luminous (median isotropic-equivalent peak luminosity 6.8×10526.8 \times 10^{52} versus 1.5×10521.5 \times 10^{52} erg s−1^{-1}) but explicitly deferred a thorough mapping of peak-richness onto other burst properties. Long GRBs from compact-object mergers, such as GRB 211211A (48 peaks) and GRB 230307A (53 peaks), sit in the hard tail of the peak-rich family, hinting at an engine connection, but only a handful of such events are known. No model currently predicts why dissipation should yield exactly two exponential components with the measured, instrument-independent parameters.

Problem Statement

Determine the physical origin of the two-family (peak-poor/peak-rich) structure in the distribution of the number of peaks within long gamma-ray bursts. The phenomenology to be explained is pinned by observation: across five independent catalogues (CGRO/BATSE, Swift/BAT, BeppoSAX/GRBM, Insight-HXMT/HE, Fermi/GBM), the peak-count distribution is satisfactorily modelled only by a mixture of two exponentials with component means of about 2.1 and 8 peaks per burst and a peak-rich fraction of about 20%, with parameters apparently independent of the instrument. The origin to be identified, at the population level, is which of the following generates the two families: (a) two distinct classes of inner engines left by the stellar collapse (for example neutron-star versus black-hole remnants), (b) a single engine class operating in two dynamical regimes (for example differing jet magnetisation or dissipation mode), or (c) another identified cause, such as a selection or detection effect. Resolving the origin requires discriminating evidence: showing that more than one reading is compatible with the data does not settle it.

References

  1. Maccary, R., Maistrello, M., Guidorzi, C., Sartori, M., Amati, L., Bazzanini, L., Bulla, M., Camisasca, A. E., Ferro, L., Frontera, F., & Tsvetkova, A. 2024, 'Distribution of the number of peaks within a long gamma-ray burst: The full Fermi/GBM catalogue', Astronomy & Astrophysics 688, L8. DOI: 10.1051/0004-6361/202450666. https://doi.org/10.1051/0004-6361/202450666
  2. Guidorzi, C., Sartori, M., Maccary, R., Tsvetkova, A., Amati, L., Bazzanini, L., Bulla, M., Camisasca, A. E., Ferro, L., Frontera, F., Li, C. K., Xiong, S. L., & Zhang, S. N. 2024, 'Distribution of the number of peaks within a long gamma-ray burst', Astronomy & Astrophysics 685, A34. DOI: 10.1051/0004-6361/202449200. https://doi.org/10.1051/0004-6361/202449200
  3. Guidorzi, C., Dichiara, S., Frontera, F., Margutti, R., Baldeschi, A., & Amati, L. 2015, 'A Common Stochastic Process Rules Gamma-Ray Burst Prompt Emission and X-Ray Flares', The Astrophysical Journal 801, 57. DOI: 10.1088/0004-637X/801/1/57. https://doi.org/10.1088/0004-637X/801/1/57
  4. Maccary, R., Guidorzi, C., Amati, L., Bazzanini, L., Bulla, M., Camisasca, A. E., Ferro, L., Frontera, F., & Tsvetkova, A. 2024, 'Distributions of Energy, Luminosity, Duration, and Waiting Times of Gamma-Ray Burst Pulses with Known Redshift Detected by Fermi/GBM', The Astrophysical Journal 965(1), 72. DOI: 10.3847/1538-4357/ad26f4. https://doi.org/10.3847/1538-4357/ad26f4
  5. Bazzanini, L., Ferro, L., Guidorzi, C., Angora, G., Amati, L., Brescia, M., Bulla, M., Frontera, F., Maccary, R., Maistrello, M., Rosati, P., & Tsvetkova, A. 2024, 'Long gamma-ray burst light curves as the result of a common stochastic pulse–avalanche process', Astronomy & Astrophysics 689, A266. DOI: 10.1051/0004-6361/202450150. https://doi.org/10.1051/0004-6361/202450150
  6. Maistrello, M., Ferro, L., Bazzanini, L., Maccary, R., & Guidorzi, C. 2025, 'An advanced pulse-avalanche stochastic model of long gamma-ray burst light curves', Astronomy & Astrophysics 697, A76. DOI: 10.1051/0004-6361/202553821. https://doi.org/10.1051/0004-6361/202553821
  7. Guidorzi, C., Orlandini, M., Maccary, R., Amati, L., Bazzanini, L., Bulla, M., Ferro, L., Frontera, F., Maistrello, M., Tan, W.-J., Tsvetkova, A., Xiong, S. L., & Zhang, S. N. 2025, 'A search for periodic activity in multi-peaked long gamma-ray bursts', Astronomy & Astrophysics 697, A228. DOI: 10.1051/0004-6361/202452326. https://doi.org/10.1051/0004-6361/202452326
  8. Maccary, R., Guidorzi, C., Maistrello, M., Kobayashi, S., Bulla, M., Moradi, R., Yi, S.-X., Wang, C. W., Zhang, W. L., Tan, W.-J., Xiong, S.-L., & Zhang, S.-N. 2026, 'A set of distinctive properties ruling the prompt emission of GRB 230307A and other long γ-ray bursts from compact object mergers', Journal of High Energy Astrophysics 49, 100456. DOI: 10.1016/j.jheap.2025.100456. https://doi.org/10.1016/j.jheap.2025.100456
  9. Maccary, R., Guidorzi, C., Amati, L., Bulla, M., Kobayashi, S., Maistrello, M., Rossi, A., Stratta, G., & Tsvetkova, A. 2026, 'The hidden population of long gamma-ray bursts from compact object mergers', Astronomy & Astrophysics 708, L18. DOI: 10.1051/0004-6361/202659083. https://doi.org/10.1051/0004-6361/202659083
  10. Maistrello, M., Guidorzi, C., Kobayashi, S., & Maccary, R. 2026, 'An internal shock model calibrated with real gamma-ray burst light curves using a genetic algorithm', Astronomy & Astrophysics. DOI: 10.1051/0004-6361/202661008. https://doi.org/10.1051/0004-6361/202661008 (arXiv:2607.12731)