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 2112.08690astro-ph.COastro-ph.IM
收藏

在CSST巡天中使用神经网络从星系通量和图像数据中提取测光红移

Extracting Photometric Redshift from Galaxy Flux and Image Data using Neural Networks in the CSST Survey

Xingchen Zhou, Yan Gong, Xian-Min Meng, Ye Cao, Xuelei Chen, Zhu Chen, Wei Du, Liping Fu, Zhijian Luo

摘要:

星系光度红移(photo- $z$)的准确性对弱引力透镜测量的分析有重要影响,特别是对未来的高精度测量。在这项工作中,我们试图提取照片$z$信息从两个星系流量和图像数据预计将获得的中国空间站望远镜(CSST)使用神经网络。我们生成模拟星系图像的基础上的观测图像从哈勃空间望远镜的高级相机巡天(HST-ACS)和COSMOS目录,考虑CSST仪器的影响。然后通过孔径测光直接从这些图像中测量星系通量数据。分别从通量和图像构建多层感知器(MLP)和卷积神经网络(CNN)来预测光$z$。我们还建议使用一个有效的混合网络,它结合了MLP和CNN,通过采用迁移学习技术来研究改进的结果与通量和图像数据。我们发现,光$z$的准确性和离群分数可以实现$σ_{\rm NMAD} = 0.023$和$η= 1.43\%$的MLP仅使用通量数据,和$σ_{\rm NMAD} = 0.025$和$η= 1.21\%$的CNN仅使用图像数据。该结果可以进一步提高作为混合传输网络的$σ_{\rm NMAD} = 0.020$和$η= 0.90\%$的高效率。这些方法导致类似的星系中值和平均红移分别为0.8和0.9,红移范围从0到4。这表明,我们的网络可以有效地和正确地提取照片- $z$信息的CSST星系流量和图像数据。

英文摘要:

The accuracy of galaxy photometric redshift (photo-$z$) can significantly affect the analysis of weak gravitational lensing measurements, especially for future high-precision surveys. In this work, we try to extract photo-$z$ information from both galaxy flux and image data expected to be obtained by China Space Station Telescope (CSST) using neural networks. We generate mock galaxy images based on the observational images from the Advanced Camera for Surveys of Hubble Space Telescope (HST-ACS) and COSMOS catalogs, considering the CSST instrumental effects. Galaxy flux data are then measured directly from these images by aperture photometry. The Multi-Layer Perceptron (MLP) and Convolutional Neural Network (CNN) are constructed to predict photo-$z$ from fluxes and images, respectively. We also propose to use an efficient hybrid network, which combines MLP and CNN, by employing transfer learning techniques to investigate the improvement of the result with both flux and image data included. We find that the photo-$z$ accuracy and outlier fraction can achieve $σ_{\rm NMAD} = 0.023$ and $η= 1.43\%$ for the MLP using flux data only, and $σ_{\rm NMAD} = 0.025$ and $η= 1.21\%$ for the CNN using image data only. The result can be further improved in high efficiency as $σ_{\rm NMAD} = 0.020$ and $η= 0.90\%$ for the hybrid transfer network. These approaches result in similar galaxy median and mean redshifts ~0.8 and 0.9, respectively, for the redshift range from 0 to 4. This indicates that our networks can effectively and properly extract photo-$z$ information from the CSST galaxy flux and image data.

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。

↑