arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2609.23982astro-ph.GAastro-ph.CO

CAMELS-CROCODILE 模拟套件:面向机器学习的新型宇宙学-天体物理学试验场

The CAMELS-CROCODILE Simulation Suite: A New Cosmology--Astrophysics Playground for Machine Learning

Kentaro Nagamine, Yuri Oku, Atsushi J. Nishizawa, Jun-Young Lee, Francisco Villaescusa-Navarro, Shy Genel, Daniel Anglés-Alcázar

首次发表
浏览论文内容

中文总结 AI 辅助

本文介绍 CAMELS-CROCODILE 模拟套件,扩展 CAMELS 框架以独立实现重子物理,并验证其与 IllustrisTNG 的差异,同时展示图神经网络跨模型推断的局限性,强调跨物理模型训练和测试 ML 的必要性。

中文摘要 AI 辅助

我们介绍 CAMELS-CROCODILE,这是一套宇宙学流体动力学模拟套件,它通过 \gadget{} 光滑粒子流体动力学代码和大阪反馈模型扩展了 CAMELS 框架。作为基于 IllustrisTNG 的第二代 CAMELS 套件的配套产品,它提供了对塑造星系、星系际介质和大尺度结构的重子物理的独立实现,拓宽了机器学习(ML)推断可以边缘化和测试的亚网格模型范围。该套件包含 $25$ 和 $50Mpc/h$ 的盒子,组织成单参数、宇宙方差和 Sobol 序列集合,这些集合围绕一个基准大阪模型变化 $26$ 个宇宙学和天体物理学参数。我们描述了模拟设计,并针对基准和其他 CAMELS 套件验证了该套件。与 IllustrisTNG 相比,基准模型仅将物质功率谱抑制了几个百分点,而在 $z=0$ 时最高可达 ${\sim}27\\%$,并且在低于 $10^{12.2} M_\odot/h$ 的质量范围内以相似的效率形成恒星,但在群规模暗晕中的恒星形成效率高出最多 $2.5$ 倍,这两者都与较弱的 AGN 反馈一致。我们还展示了恒星和黑洞偏置以及星系周围的 Ly$\alpha$ 吸收。作为第一个 ML 应用,我们将一个在 IllustrisTNG 星系目录上训练、权重冻结的图神经网络应用于 CAMELS-CROCODILE。它无法恢复 $\Omega_{\rm m}$ 和 $\sigma_8$,并返回过度自信的后验,主要针对包含比其训练集中任何模拟更多星系的模拟;在该范围内,$\Omega_{\rm m}$ 的恢复误差仅为分布内值的 $1.5$ 倍。这些结果强调了在物理上不同的星系形成模型上训练和测试 ML 推断的必要性。

英文摘要

We present CAMELS-CROCODILE, a suite of cosmological hydrodynamic simulations that extends the CAMELS framework with the \gadget{} smoothed particle hydrodynamics code and the Osaka feedback model. As a companion to the IllustrisTNG-based second-generation CAMELS suite, it provides an independent implementation of the baryonic physics that shapes galaxies, the intergalactic medium, and large-scale structure, broadening the range of subgrid models over which machine-learning (ML) inference can be marginalized and tested. The suite comprises $25$ and $50Mpc/h$ boxes organized into one-parameter, cosmic-variance, and Sobol-sequence sets that vary $26$ cosmological and astrophysical parameters around a fiducial Osaka model. We describe the simulation design and validate the suite against benchmarks and other CAMELS suites. Compared with IllustrisTNG, the fiducial model suppresses the matter power spectrum by only a few per cent, against up to ${\sim}27\%$ at $z=0$, and forms stars with similar efficiency below $10^{12.2} M_\odot/h$ but up to $2.5$ times more efficiently in group-scale halos, both consistent with weaker AGN feedback. We also present the stellar and black hole bias and the Ly$α$ absorption around galaxies. As a first ML application, we apply a graph neural network trained on IllustrisTNG galaxy catalogs, with frozen weights, to CAMELS-CROCODILE. It fails to recover $Ω_{\rm m}$ and $σ_8$ and returns overconfident posteriors, mainly for simulations containing more galaxies than any in its training set; within that range $Ω_{\rm m}$ is recovered with an error only $1.5$ times the in-distribution value. These results underline the need to train and test ML inference across physically distinct galaxy formation models.

发表机构

  • The University of Osaka(大阪大学)
  • The University of Tokyo(东京大学)
  • University of Nevada, Las Vegas(内华达大学拉斯维加斯分校)
  • Nevada Center for Astrophysics(内华达天体物理中心)
  • Nagoya University(名古屋大学)
  • Princeton University(普林斯顿大学)
  • Flatiron Institute(平顿研究所)

机构由 AI 辅助整理,请以论文原文为准。

补充信息

↑