Distinguishing Time Delays from Multi-Messenger Transients with Deep Learning

27 Sept 2022, 11:30
20m
Sestri Levante

Sestri Levante

Grand Hotel dei Castelli, Via Penisola Levante, 26, 16039 Sestri Levante (GE), Italy
Oral contribution Session 3

Speaker

Mikhail Denissenya (Nazarbayev University)

Description

Gravitationally lensed multi-messenger transients are promising probes for constraining cosmological parameters including the Hubble constant. We focus on developing a deep learning technique to estimate lensing time delays from various multiply imaged unresolved transients. We train convolutional neural networks and apply them to simulated supernovae lightcurves to determine whether there are one or two or four lensed images, and measure the corresponding time delays. We accurately identify the number of images and estimate the time delays exceeding ∼6 days.

Primary author

Mikhail Denissenya (Nazarbayev University)

Co-author

Eric Linder (UC Berkeley)

Presentation materials