Investigating Cosmological GAN Emulators Using Latent Space Interpolation

Abstract

Generative adversarial networks (GANs) have been recently applied as a novel emulation technique for large scale structure simulations. Recent results show that GANs can be used as a fast, efficient and computationally cheap emulator for producing novel weak lensing convergence maps as well as cosmic web data in 2-D and 3-D. However, like any algorithm, the GAN approach comes with a set of limitations, such as an unstable training procedure and the inherent randomness of the produced outputs. In this work we employ a number of techniques commonly used in the machine learning literature to address the mentioned limitations. In particular, we train a GAN to produce both weak lensing convergence maps and dark matter overdensity field data for multiple redshifts, cosmological parameters and modified gravity models. In addition, we train a GAN using the newest Illustris data to emulate dark matter, gas and internal energy distribution data simultaneously. Finally, we apply the technique of latent space interpolation to control which outputs the algorithm produces. Our results indicate a 1-20% difference between the power spectra of the GAN-produced and the training data samples depending on the dataset used and whether Gaussian smoothing was applied. Finally, recent research on generative models suggests that such algorithms can be treated as mappings from a lower-dimensional input (latent) space to a higher dimensional (data) manifold. We explore such a theoretical description as a tool for better understanding the latent space interpolation procedure.

20 pages, 23 figures, 3 tables

References in corpus (9)

• Conditional Generative Adversarial Nets
• Properties of galaxies reproduced by a hydrodynamic simulation
• Quantifying the Carbon Emissions of Machine Learning
• Constraining neutrino mass with weak lensing Minkowski Functionals
• Do dark matter halos explain lensing peaks?
• GANs May Have No Nash Equilibria
• Encoding large scale cosmological structure with Generative Adversarial Networks
• The Riemannian Geometry of Deep Generative Models
• Comparison of Generative Adversarial Networks Architectures Which Reduce Mode Collapse

Cited by in corpus (9)

• Large-scale dark matter simulations
• The CAMELS Multifield Dataset: Learning the Universe's Fundamental Parameters with Artificial Intelligence
• From EMBER to FIRE: predicting high resolution baryon fields from dark matter simulations with Deep Learning
• MillimeterDL: Deep Learning Simulations of the Microwave Sky
• Bayesian deep learning for cosmic volumes with modified gravity
• Neural style transfer of weak lensing mass maps
• Stochastic Super-resolution of Cosmological Simulations with Denoising Diffusion Models
• Reconstructing the mid-infrared spectra of galaxies using ultraviolet to submillimeter photometry and Deep Generative Networks
• nuGAN: Generative Adversarial Emulator for Cosmic Web with Neutrinos