arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2310.07590astro-ph.COastro-ph.IM
收藏

用神经网络恢复CMB偏振图:在真实模拟中的性能

Recovering CMB polarization maps with neural networks: Performance in realistic simulations

Instituto de Astrofísica de Canarias(加那利群岛天体物理学研究所);Universidad de La Laguna, Departamento de Astrofísica(拉帕尔马大学天体物理系);Departamento de Física, Universidad de Oviedo(奥维耶多大学物理系);Instituto Universitario de Ciencias y Tecnologías Espaciales de Asturias (ICTEA)(阿斯图里亚斯太空科学与技术大学研究所);Dipartimento di Fisica e Astronomia, Universitá degli Studi di Catania(卡塔尼亚大学物理与天文学系);INFN – Sezione di Catania(意大利国家核物理研究院卡塔尼亚分部);SISSA(国际高等研究学院);INFN-Sezione di Trieste(意大利国家核物理研究院的里雅斯特分部);Departamento de Informática, Universidad de Oviedo(奥维耶多大学计算机系);Escuela de Ingeniería de Minas, Energía y Materiales, Universidad de Oviedo(奥维耶多大学矿业、能源与材料工程学院)

J. M. Casas, L. Bonavera, J. González-Nuevo, G. Puglisi, C. Baccigalupi, S. R. Cabo, M. M. Cueli, D. Crespo, C. González-Gutiérrez, F. J. de Cos

AI总结:

本文提出一种基于神经网络的成分分离方法重建CMB偏振图,在真实Planck模拟与观测中验证了其有效性,表明该方法具有潜力但尚需改进以实现高精度宇宙学估计。

摘要:

恢复偏振宇宙微波背景(CMB)对于阐明极早期宇宙的指数膨胀(即宇宙暴胀)至关重要。实现这一目标不仅需要提高仪器灵敏度,还需要开发稳健且多样化的数据分析技术。在这项工作中,我们探索了一种基于神经网络的新型成分分离方法,以重建Stokes $Q$和$U$偏振图。为了验证该方法,我们首先在真实的\textit{Planck}天空模拟上测试该网络,发现$E$模式的平均绝对误差为$0.1 \pm 0.3~μK^{2}$,$B$模式为$-0.1 \pm 0.3~μK^{2}$。然后,我们将训练好的网络应用于公开的\textit{Planck}观测数据,其结果与使用Commander方法获得的结果一致。基于这些发现,我们得出结论:基于神经网络的方法在偏振CMB实验中作为成分分离技术显示出潜力。然而,在这些方法能够提供可靠的高精度宇宙学估计之前,仍需进行实质性的改进和更全面的分析。

英文摘要:

Recovering the polarized cosmic microwave background (CMB) is essential for shedding light on the exponential expansion of the very early Universe, known as cosmic inflation. Achieving this goal requires not only improved instrumental sensitivity but also the development of robust and diverse data analysis techniques. In this work, we explore a novel component separation approach based on neural networks to reconstruct the Stokes $Q$ and $U$ polarization maps. To validate the method, we first test the network on realistic \textit{Planck} sky simulations, finding a mean absolute error of $0.1 \pm 0.3~μK^{2}$ for the $E$-mode and $-0.1 \pm 0.3~μK^{2}$ for the $B$-mode. We then apply the trained network to public \textit{Planck} observations, with results consistent with those obtained using the Commander method. Based on these findings, we conclude that neural network-based methods show potential as component separation techniques in polarization CMB experiments. However, substantial improvements and more comprehensive analyses are necessary before these methods can provide reliable high-precision cosmological estimates.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑