Gamma-Ray Bursts: When the Universe Makes Popcorn ✛ ✦

Original: "An internal shock model calibrated with real gamma-ray burst light curves using a genetic algorithm"
arXiv:2607.12731 · 2026-07-14 · CC BY 4.0 · 1 min
The brightest cosmic flashes erupt as chaotically as popping popcorn.

Gamma-ray bursts are cosmic popcorn. In a pot, kernels explode chaotically. In the universe, a dying supernova collapses into a black hole, a superdense sphere, or a supermagnetized object. An accretion disk forms around it, from which plasma jets shoot out, like kernels popping. Blobs crash into each other, creating gamma-ray light. Scientists used satellite data measuring brightness and colors of the rays, adjusted for redshift (the Hubble effect). A self-learning algorithm tried thousands of combinations, and through computer modeling produced bursts indistinguishable from real ones. Even a simple model where blobs fly out like popcorn accurately predicts observed data, if you factor in Einstein's theory. In a second, such a burst releases more energy than the Sun will in its entire lifetime, but due to the vast distances, only a harmless ripple reaches Earth.

The number of blobs in a gamma-ray burst follows the same rule as earthquake magnitudes: many weak ones and few strong ones.

\epsilon = 1 - \frac{(m_r + m_s)\gamma_m}{m_r\gamma_r + m_s\gamma_s}
ϵ is the fraction of energy converted into radiation upon collision of two shells; m_r, m_s are the masses of the fast and slow shells; γ_r, γ_s, γ_m are their Lorentz factors and the factor of the merged shell.
f(t) = \frac{1}{\tau} e^{-t/\tau}
τ is the characteristic decay time of the central engine's activity; the probability of ejecting a new shell drops exponentially over time.
Scientists
Georges LemaîtreFritz ZwickyJacob BekensteinKarl SchwarzschildRalph AlpherStephen Hawking
Concepts
central engine gamma-ray burst gamma ray burst merger genetic algorithm stochastic gene expression afterglow swift satellite xgboost outburst generative model gene expression noise Gaussian distribution random number generator CRISPR gibbs sampling log-normal distribution