<tt>PINION</tt>: physics-informed neural network for accelerating radiative transfer simulations for cosmic reionization

With the advent of Square Kilometre Array Observatory (SKAO), scientists will be able to directly observe Epoch Reionization by mapping distribution neutral hydrogen at different redshifts. While physically motivated results can simulated with radiative transfer codes, these simulations are computationally expensive and not readily produce required scale resolution simultaneously. Here we introduce Physics-Informed neural Network for reIONization (PINION), which accurately swiftly predict complete 4-D fraction evolution from smoothed gas mass density fields pre-computed N-body simulation. We trained PINION on C$^2$-Ray simulation outputs a physics constraint reionization chemistry equation is enforced. only five redshift snapshots propagation mask as simplistic approximation ionizing photon mean free path, entire history between $z=6$ $12$. evaluate accuracy our predictions analysing dimensionless power spectra morphology statistics estimations against results. show that while network's in good agreement $z>7$, suffers $z<7$ primarily due oversimplified mask. motivate how performance drastically improved potentially generalized large-scale simulations.