使用UKIDSS深遠巡天觀測探索星系群的性質到z=2

Abstract

星系傾向在密度高的群體的環境下,因此群體在星系演化下扮演很重要的角色。 星系合併與一些環境造成的過程是造成星系演化的原因。觀測星系群在高紅移是個困難的挑戰,特別在紅色(被動)的星系系統。 其他近期的觀測方式是用可見光的波段,針對紅色(被動)的成對星系或星系群,但這個方試容易低估部分的星系演化。 我們使用最好的世界發佈資料,來自UKIDSS-UDS-DR8的近紅外波段,結合Subaru以及CFHT的可見光部分 我們採用photo-z的技術,並使用pfof的方法重建星系群到z=2. 我們以Millennuim simulation和X-ray source探測資料作為重建星系群的基準,雖然Millennuim simulation的資料分布情形並不能代表我們真實的K-band資料。且spec-z需要大量的時間,X-ray source的成員星系數量也並不齊全。導致我們使用PFOF重建出的星系群體會有"不完備性"以及"碎裂性"在第一個方法上,而第二個方法由於spec-z的數量不足造成有許多錯誤的星系群被偵測出來。Galaxies live preferentially in groups environment, therefore groups play a very important role in galaxy evolution. Merging activity is fostered in groups and someenvironmental processes could eventually take place in groups and have an impact on galaxy evolution. Observing and detecting groups at high redshift is extremely challenging, especially for the redder systems, involving passive galaxies. The current generation of optically-selected surveys are biased against the passive population in pairs and groups, leading to an underestimation of their role in the galaxy evolution. Making the best use of the worldwide release of the near-infrared data UKIDSS-UDS DR8, combined with Subaru and CFHT data in the optical waveband, we coupled photometric redshift with a probabilistic Friends-of-Friends algorithm to build a groups catalog up to redshift z = 2. As we only have a limited access to spectroscopic redshifts, our overdensity (cluster/group) finder algorithm has to rely heavily on photometric redhifts, so we are using a probablistic Friend-of-Friend method. Here we discuss the work we conducted to select groups up to z = 2, which relied heavily on how the probabilistic Friends-of-Friends algorithm is trained. We used both Mock catalogs (from the Millennium simulation) and X-ray detections (from XMM-Newton observation). We demonstrate that first the Mock catalog does not reproduce our K-band selected sample, which spread doubts on the reliability of the training and that secondly that the incomplete memberships to the X-ray groups (limited by the availability of spectroscopic data) prevent us to rely as well on X-ray training. Training our algorithm lead to a low completeness and fragmentation of our group sample in the first case, and many false detection in the second case.