%0 Generic %1 bye202121cmvae %A Bye, Christian Hellum %A Portillo, Stephen K. N. %A Fialkov, Anastasia %D 2021 %K library %T 21cmVAE: A VAE-based Emulator of the 21-cm Global Signal %U http://arxiv.org/abs/2107.05581 %X Considerable observational efforts are being dedicated to measuring the sky-averaged (global) 21-cm signal of neutral hydrogen from Cosmic Dawn and the Epoch of Reionization. Deriving observational constraints on the astrophysics of this era requires modelling tools that can quickly and accurately generate theoretical signals across the wide astrophysical parameter space. For this purpose artificial neural networks were used to create the only two existing global signal emulators 21cmGEM and globalemu. In this paper we introduce 21cmVAE, a global signal emulator based on advanced machine learning methods such as variational autoencoder (VAE) and trained with the same dataset of ~ 30,000 global signals as the other two emulators. The VAE allows us to explore a low-dimensional representation of the dataset and establish the most important astrophysical processes that drive the global 21-cm signal at different epochs. 21cmVAE has a relative rms error of only 0.41\% -- equivalently 0.66 mK -- on average, which is a significant improvement compared to the existing emulators, and a run time of 0.04 seconds per parameter set. The emulator, the code, and the processed datasets are publicly available at https://github.com/christianhbye/21cmVAE and through http://doi.org/10.5281/zenodo.5085445.

@misc{bye202121cmvae, abstract = {Considerable observational efforts are being dedicated to measuring the sky-averaged (global) 21-cm signal of neutral hydrogen from Cosmic Dawn and the Epoch of Reionization. Deriving observational constraints on the astrophysics of this era requires modelling tools that can quickly and accurately generate theoretical signals across the wide astrophysical parameter space. For this purpose artificial neural networks were used to create the only two existing global signal emulators 21cmGEM and globalemu. In this paper we introduce 21cmVAE, a global signal emulator based on advanced machine learning methods such as variational autoencoder (VAE) and trained with the same dataset of ~ 30,000 global signals as the other two emulators. The VAE allows us to explore a low-dimensional representation of the dataset and establish the most important astrophysical processes that drive the global 21-cm signal at different epochs. 21cmVAE has a relative rms error of only 0.41\% -- equivalently 0.66 mK -- on average, which is a significant improvement compared to the existing emulators, and a run time of 0.04 seconds per parameter set. The emulator, the code, and the processed datasets are publicly available at https://github.com/christianhbye/21cmVAE and through http://doi.org/10.5281/zenodo.5085445.}, added-at = {2021-07-13T06:51:23.000+0200}, author = {Bye, Christian Hellum and Portillo, Stephen K. N. and Fialkov, Anastasia}, biburl = {https://www.bibsonomy.org/bibtex/2c2260e0f1b626fba7bd243425231dc10/gpkulkarni}, description = {21cmVAE: A VAE-based Emulator of the 21-cm Global Signal}, interhash = {c35482ee94d97e900987aadbf559117f}, intrahash = {c2260e0f1b626fba7bd243425231dc10}, keywords = {library}, note = {cite arxiv:2107.05581Comment: 10 pages, 5 figures, 2 tables (including appendices). Submitted to ApJ}, timestamp = {2021-07-13T06:51:23.000+0200}, title = {21cmVAE: A VAE-based Emulator of the 21-cm Global Signal}, url = {http://arxiv.org/abs/2107.05581}, year = 2021 }