EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

As Simple as Possible but No Simpler: Optimizing the Performance of Neural Net Emulators for Galaxy SED Fitting

Elijah P. Mathews, Joel Leja, Joshua S. 佳士 Speagle 沈, Benjamin D. Johnson, Justus Gibson, Erica J. Nelson, Katherine A. Suess, Sandro Tacchella, Katherine E. Whitaker, Bingjie 冰洁 Wang 王

The Astrophysical Journal · 01.09.2023 · 954 (2) · 132

DOI: 10.3847/1538-4357/ace720 ↗

Kurz gefasst

Fachartikel in der Zeitschrift The Astrophysical Journal, online erschienen am 29.08.2023. Verfasst von Elijah P. Mathews und 9 weiteren Autor:innen. Die Arbeit wurde bisher 23-mal zitiert (Crossref) und nennt 79 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Space Telescope Science Institute, National Aeronautics and Space Administration.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Artificial neural network emulators have been demonstrated to be a very computationally efficient method to rapidly generate galaxy spectral energy distributions, for parameter inference or otherwise. Using a highly flexible and fast mathematical structure, they can learn the nontrivial relationship between input galaxy parameters and output observables. However, they do so imperfectly, and small errors in flux prediction can yield large differences in recovered parameters. In this work, we investigate the relationship between an emulator’s execution time, uncertainties, correlated errors, and ability to recover accurate posteriors. We show that emulators can recover consistent results to traditional fits, with a precision of 25%–40% in posterior medians for stellar mass, stellar metallicity, star formation rate, and stellar age. We find that emulation uncertainties scale with an emulator’s width N as ∝ N −1 , while execution time scales as ∝ N 2 , resulting in an inherent tradeoff between execution time and emulation uncertainties. We also find that emulators with uncertainties smaller than observational uncertainties are able to recover accurate posteriors for most parameters without a significant increase in catastrophic outliers. Furthermore, we demonstrate that small architectures can produce flux residuals that have significant correlations, which can create dangerous systematic errors in colors. Finally, we show that the distributions chosen for generating training sets can have a large effect on an emulator’s ability to accurately fit rare objects. Selecting the optimal architecture and training set for an emulator will minimize the computational requirements for fitting near-future large-scale galaxy surveys. We release our emulators on GitHub ( http://github.com/elijahmathews/MathewsEtAl2023 ).

Förderung

  • Space Telescope Science Institute
  • National Aeronautics and Space Administration

Zitieren

APA

Mathews, E. P., Leja, J., 沈, J. S. 佳. S., Johnson, B. D., Gibson, J., Nelson, E. J., Suess, K. A., Tacchella, S., Whitaker, K. E., & 王, B. 冰. W. (2023). As Simple as Possible but No Simpler: Optimizing the Performance of Neural Net Emulators for Galaxy SED Fitting. The Astrophysical Journal, 954(2), 132. https://doi.org/10.3847/1538-4357/ace720

Vancouver

Mathews EP, Leja J, 沈 JS佳S, Johnson BD, Gibson J, Nelson EJ, et al. As Simple as Possible but No Simpler: Optimizing the Performance of Neural Net Emulators for Galaxy SED Fitting. The Astrophysical Journal. 2023;954(2):132. doi:10.3847/1538-4357/ace720

BibTeX
@article{Mathews2023,
  author = {Mathews, Elijah P. and Leja, Joel and 沈, Joshua S. 佳士 Speagle and Johnson, Benjamin D. and Gibson, Justus and Nelson, Erica J. and Suess, Katherine A. and Tacchella, Sandro and Whitaker, Katherine E. and 王, Bingjie 冰洁 Wang},
  title = {As Simple as Possible but No Simpler: Optimizing the Performance of Neural Net Emulators for Galaxy SED Fitting},
  journal = {The Astrophysical Journal},
  year = {2023},
  volume = {954},
  number = {2},
  pages = {132},
  doi = {10.3847/1538-4357/ace720}
}
Kostenlos starten →