Enabling Dark Energy Science with Deep Generative Models of Galaxy Images

TL;DR

This paper explores the use of deep generative models, including variational autoencoders and adversarial networks, to generate realistic galaxy images for calibrating shape measurements in dark energy research, reducing the need for costly observations.

Contribution

It introduces a new adversarial training objective for conditional generative models and demonstrates their effectiveness in producing realistic galaxy images for cosmological calibration.

Findings

Generative models can produce high-quality galaxy images.

The new adversarial objective improves image realism.

Models provide a cost-effective calibration data source.

Abstract

Understanding the nature of dark energy, the mysterious force driving the accelerated expansion of the Universe, is a major challenge of modern cosmology. The next generation of cosmological surveys, specifically designed to address this issue, rely on accurate measurements of the apparent shapes of distant galaxies. However, shape measurement methods suffer from various unavoidable biases and therefore will rely on a precise calibration to meet the accuracy requirements of the science analysis. This calibration process remains an open challenge as it requires large sets of high quality galaxy images. To this end, we study the application of deep conditional generative models in generating realistic galaxy images. In particular we consider variations on conditional variational autoencoder and introduce a new adversarial objective for training of conditional generative networks. Our results suggest a reliable alternative to the acquisition of expensive high quality observations for generating the calibration data needed by the next generation of cosmological surveys.