The fraction of star-forming clumpy galaxies in diverse environments out to z 3
Abstract
Star-forming clumps have been observed in some galaxies. The identification of these clumps can unveil information about their formation and evolution. We use a convolutional neural network (CNN) to select clumpy galaxies throughout the H-band selected star-forming galaxies (SFGs) in all five CANDELS fields (GOODS-S, GOODS-N, UDS, EGS, and COSMOS) at the redshift range of 0.5< z < 3. Our CNN model is trained using ~3000 SFGs in the GOODS-S field. The clumps in the training sample are identified by the Canny edge detector algorithm and visually inspected. The CNN model reaches the accuracy of >90% in the classification of clumpy galaxies, which allows us to build a reliable sample of clumpy galaxies all over the five CANDELS fields. We measure the fraction of clumpy galaxies (fclumpy), and then investigate the evolution of fclumpy as a function of redshift, stellar mass, and galaxy environment. Such studies constrain the formation mechanism of star-forming clumps in galaxies.
- Publication:
-
American Astronomical Society Meeting Abstracts #235
- Pub Date:
- January 2020
- Bibcode:
- 2020AAS...23528310S