Lunch talk on Dec. 21, 2021

Enhance the detection efficiency and optimize follow-up strategy for cosmic explosions using GPU-powered and data-driven methods

Speaker:Lei Hu (PMO)

Venue:SWIFAR Building 2111

Time:12:30 PM, Tuesday, Dec. 21, 2021

Abstract:

The advent of wide-field surveys, as typified by ZTF, has enabled a wealth of discoveries of cosmic explosions that will continue to shed light on the physical nature of our dynamic Universe. The modern survey facilities are moving the time-domain field forward by allowing us to undertake transient surveys that are wider, deeper, and faster. On the flip side, the data deluge manifested itself in the large streaming image data and the vast number of transient discoveries is causing unprecedented challenges in the new era. In this talk, I will introduce a new GPU-powered image subtract algorithm SFFT that can significantly boost the efficiency of transient detections. The method has been applied in various wide-field transient surveys, such as the DESIRT supernova survey based on CTIO-4m DECam. It is also the detection engine of the real-time transient pipeline of WFST under development. I will also advertise a data-driven method based on LSTM neural networks to analyze spectral time-series of SNeIa. Given the deficiency of spectroscopic resources for future time-domain surveys, the spectral follow-ups of transients will need to be built with the knowledge of the existing dataset. This trend motivated us to develop this data-driven method for spectral inference, and we expect it can help to optimize the planning of spectroscopic follow-up strategy of future supernova surveys.

Report PPT:SWIFAR_Lei Hu.pptx