Evaluating machine learning techniques for predicting power spectra from reionization simulations

Upcoming experiments such as the Square Kilometre Array will provide huge quantities of data. Fast modelling of the high-redshift 21 cm signal will be crucial for efficiently comparing these data sets with theory. The most detailed theoretical predictions currently come from numerical simulations and from faster but less accurate seminumerical simulations. Recently, machine learning techniques have been proposed to emulate the behaviour of these seminumerical simulations with drastically reduced time and computing cost. We compare the viability of five such machine learning techniques for emulating the 21 cm power spectrum of the publicly available code simfast21. Our best emulator is a multilayer perceptron with three hidden layers, reproducing simfast21 power spectra 108 times faster than the simulation with 4 per cent mean squared error averaged across all redshifts and input parameters. The other techniques (interpolation, Gaussian processes regression, and support vector machine) have slower prediction times and worse prediction accuracy than the multilayer perceptron. All our emulators can make predictions at any redshift and scale, which gives more flexible predictions but results in significantly worse prediction accuracy at lower redshifts. We then present a proof-of-concept technique for mapping between two different simulations, exploiting our best emulator’s fast prediction speed. We demonstrate this technique to find a mapping between simfast21 and another publicly available code 21cmfast. We observe a noticeable offset between the simulations for some regions of the input space. Such techniques could potentially be used as a bridge between fast seminumerical simulations and accurate numerical radiative transfer simulations.