Next-Generation Data-Driven Probabilistic Modelling of Type Ia Supernova SEDs in the Optical to Near-Infrared for Robust Cosmological Inference

Projektinformationen

BayeSN

ID Finanzhilfevereinbarung: 101002652

DOI

10.3030/101002652

EK-Unterschriftsdatum 20 Januar 2021

Startdatum 1 Oktober 2021

Enddatum 30 September 2027

Finanziert unter

• EXCELLENT SCIENCE - European Research Council (ERC)

Gesamtkosten

€ 2 446 736,00

EU-Beitrag

€ 2 446 736,00

Koordiniert durch

THE CHANCELLOR MASTERS AND SCHOLARS OF THE UNIVERSITY OF CAMBRIDGE
United Kingdom

Periodic Reporting for period 2 - BayeSN (Next-Generation Data-Driven Probabilistic Modelling of Type Ia Supernova SEDs in the Optical to Near-Infrared for Robust Cosmological Inference)

Berichtszeitraum: 2023-04-01 bis 2024-09-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

Type Ia supernovae (SNe Ia) are bright stellar explosions that are used to measure distances to galaxies and thus to trace the cosmic expansion history, constrain cosmological parameters, and help us understand the nature of the Universe. Supernova distances are estimated from their variations in brightness over time and wavelength (light curves or spectral energy distributions - SEDs). However, interstellar dust in the galaxy of the supernova also affects its brightness and can cause a systematic error in the distance estimates if not properly accounted for. While conventional supernova analyses focus on optical light, a wider wavelength range including near-infrared (NIR) light, should help constrain the effect of dust and improve the precision and accuracy of supernova distances. The objective of this project is to develop BayeSN, the next-generation data-driven probabilistic model for SN Ia spectral energy distributions, jointly leveraging optical and NIR data to determine the optimal distance estimates for current and future supernova cosmology surveys, such as the Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) to ensure principled, accurate, and optimal cosmological analyses.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

We have developed the initial version of the BayeSN model, trained on existing optical and NIR data. We have leveraged the latest numerical computing and machine learning frameworks to accelerate the speed of this model by a factor of 100x, enabling us to scale our statistical analyses to 100x larger supernova datasets. We have applied BayeSN to analyse optical and NIR datasets, from surveys of nearby and distant supernovae using ground-based observatories and the Hubble Space Telescope, to determine the improvement in supernova distances and to address contemporary questions about the effects of dust in the supernova galaxies and how they may depend on the properties of the galaxy (e.g. its mass in stars). We are participating in ongoing time-domain surveys to obtain new supernova datasets in the optical and NIR, and develop and apply new machine learning techniques to analyse new, large datasets of stellar explosions. We are integrating BayeSN with supernova survey simulation software for deployment in the next-generation LSST survey which is due to begin in late 2025.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

We have developed the state-of-the-art and only statistical framework for SN Ia SEDs in the optical and NIR that models physically-distinct effects of dust vs intrinsic supernova variations and enables the probabilistic inference of dust and distance for individual supernovae. We have used our model to put novel constraints on the nature of dust in the galaxies hosting nearby and distant supernovae. We have applied techniques we have developed for BayeSN to engineer new methods to analyse gravitationally lensed supernovae, which have been applied to novel space-based observations of these rare events. We have incorporated our BayeSN model with new machine learning-enabled simulation-based inference techniques that will enable us to incorporate more realistic systematic effects in analyses of future large datasets.

In the remaining grant period, we will continue to further develop BayeSN using new, larger supernova datasets from current surveys, and apply it to derive cosmological constraints and astrophysical insights into the connections between supernovae and their galaxies. We will continue to integrate BayeSN with survey simulation frameworks to deploy it for analysis of future LSST data.