发表机构
Max-Planck-Institut für extraterrestrische Physik; Ludwig-Maximilians-Universität München; Instituto de Astronomía Teórica y Experimental, CONICET-UNC; Universidad Nacional de Córdoba（马克斯·普朗克地外物理研究所; 慕尼黑大学; 理论与实验天文学研究所，阿根廷国家科学研究委员会-科尔多瓦大学; 科尔多瓦国立大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出基于演化映射和高斯过程回归的晕质量函数仿真器Aletheia，通过两步仿真实现百分之一精度，适用于多种扩展宇宙学，并公开提供。
AI 中文摘要
我们提出了一种基于高斯过程回归的晕质量函数（HMF）仿真器，该仿真器利用了演化映射框架。该框架将宇宙学参数分为形状参数（$\mathbf{\Theta}_\mathrm{s}$），其决定线性功率谱的形状，以及演化参数（$\mathbf{\Theta}_\mathrm{e}$），其仅重新缩放功率谱的振幅，其净效应完全由聚类振幅 $\sigma_{12}$ 捕获。仿真分两步进行：首先，一个仿真器根据 $\mathbf{\Theta}_\mathrm{s}$ 和 $\sigma_{12}$ 预测 HMF 的整体形状；然后，第二个仿真器应用一个小修正，以考虑对密度涨落积分增长历史的残余依赖。该仿真器在 100 个平坦 $\Lambda$CDM 宇宙学模拟套件上训练，但由于演化映射，可应用于预测具有任何扩展 $\mathbf{\Theta}_\mathrm{e}$ 空间（如曲率、quintessence 和动力学暗能量）的宇宙学中的晕丰度。我们针对独立的 $\Lambda$CDM 运行以及额外的动力学暗能量模拟验证套件对仿真器进行了测试，在广泛的质量和红移范围内实现了百分之一级别的精度，性能仅在早期时刻略有下降，此时训练数据点较少。与其他最先进的 HMF 仿真器相比，其精度在几乎所有测试案例中都更高。我们的结果证明了演化映射框架在构建密度场统计量的高精度仿真器方面的广泛适用性。该仿真器作为 Aletheia 仿真器套件的一部分公开提供。
英文摘要
We present an emulator of the halo mass function (HMF) based on Gaussian-process regression, that exploits the evolution mapping framework. This framework splits cosmological parameters into shape parameters ($\mathbfΘ_\mathrm{s}$), which set the shape of the linear power spectrum, and evolution parameters ($\mathbfΘ_\mathrm{e}$), which only rescale its amplitude, with their net effect entirely captured by the clustering amplitude, $σ_{12}$. The emulation acts in two steps: first, an emulator predicts the overall shape of the HMF given $\mathbfΘ_\mathrm{s}$ and $σ_{12}$; then, a second emulator applies a small correction accounting for the residual dependence on the integrated growth history of density fluctuations. The emulator is trained on a suite of 100 simulations of flat $Λ$CDM cosmologies but, owing to evolution mapping, can be applied to predict the abundance of haloes in cosmologies with any extended space of $\mathbfΘ_\mathrm{e}$, such as curvature, quintessence and dynamical dark energy. We test the emulator against independent $Λ$CDM runs and on an additional validation suite of dynamical dark energy simulations, achieving per cent-level accuracy over a wide range of masses and redshifts, with the performance only slightly degrading at early times, where the training data points are fewer. Compared to other state-of-the-art HMF emulators, its accuracy is higher in nearly all tested cases. Our results demonstrate the broad applicability of the evolution mapping framework to build very accurate emulators of statistics of the density field. The emulator is publicly available as part of the Aletheia emulator suite.
CommentsSubmitted to MNRAS
