科研之友
微信
新浪微博
Facebook
分享链接
摘要
With an eye towards the computational requirements of future large-scale surveys such as Euclid and Large Synoptic Survey Telescope (LSST) that will require photometric redshifts (photo-z's) for greater than or similar to 10(9) objects, we investigate a variety of ways that 'fuzzy archetypes' can be used to improve photometric redshifts and explore their respective statistical interpretations. We characterize their relative performance using an idealized LSST ugrizY and Euclid YJH mock catalogue of 10 000 objects spanning z = 0-6 at Y = 24 mag. We find most schemes are able to robustly identify redshift probability distribution functions that are multimodal and/or poorly constrained. Once these objects are flagged and removed, the results are generally in good agreement with the strict accuracy requirements necessary to meet Euclid weak lensing goals for most redshifts between 0.8 less than or similar to z less than or similar to 2. These results demonstrate the statistical robustness and flexibility that can be gained by combining template-fitting and machine-learning methods and provide useful insights into how astronomers can further exploit the colour-redshift relation.
- 出版日期2017-7
