Developing optimal strategies for observing the electro-magnetic counterparts of gravitational-wave events

这个项目的目标是发展和应用贝叶斯方法来优化对紧致双并合引力波事件的电磁对应物的后续观测问题。这项工作的主要背景和重点将是对双中子星合并后的千新星的光学和近红外观测，但预计贝叶斯工具将有更广泛的应用。该项目将寻求制定策略，以推断千新星观测的最佳数量、epoch、曝光时间和颜色带，从而最大限度地提高检测千新星和/或消除其他瞬态事件的假阳性检测的可能性。所开发的方法将识别并充分适应地面引力波干涉仪网络有限的天空定位能力，因此也将寻求研究策略，以最佳地“平铺”观测引力波候选者的天空定位区域——建立并扩展到最近在格拉斯哥开发的更现实的案例方法。这方面的研究可能包括调查引力波源、千新星及其宿主星系特性之间的可能关系，发展格拉斯哥大学早期关于“多信使概率函数”的工作。开发的方法将涉及计算千新星检测与其他瞬态事件检测的贝叶斯后验比值比。随着项目的发展，将探索不仅仅是探测千新星，而且是表征和估计其源参数的策略。机器学习方法可能适合解决这些问题，并且将向学生提供这些方法应用方面的培训。还将安排与LIGO天文台的长期合作，以研究与干涉仪相关的校准误差的特征，以及这些误差如何影响探测器网络的天空定位能力。

The objective of this project will be to develop and apply Bayesian methodology to the problem of optimising follow up observations of the electro-magnetic counterparts of compact binary coalescence gravitational-wave events. The main context and focus for this work will be optical and near infra-red observations of kilonovae, following binary neutron star coalescences, but it is anticipated that the Bayesian tools developed will have broader applications. The project will seek to develop strategies for inferring the optimal number, epoch, exposure time and colour bands of kilonovae observations that will maximise the probability of detecting the kilonova and/or eliminating false positive detections from other transient events. The methods developed will recognise and fully accommodate the limited sky localisation capabilities of ground-based networks of gravitational-wave interferometers, and will also therefore seek to investigate strategies for optimally 'tiling' the sky localisation regions in which the gravitational-wave candidates are observed - building on and extending to more realistic cases methodology recently developed at Glasgow. This strand of the research may involve investigation of possible relationships between gravitational-wave sources, kilonovae and their host galaxy properties, developing earlier work at Glasgow on the "multi-messenger probability function". The methods developed will involve computing Bayesian posterior odds ratios for kilonova detections versus detections of other transient events. As the project develops, strategies for not simply detecting kilonovae but also characterising and estimating their source parameters will be explored. Machine learning methods may be appropriate to address such problems, and training in the application of these methods will be provided to the student. An extended long-term attachment to the LIGO Observatories will also be arranged, to work on characterisation of the calibration errors associated with the interferometers - and how these affect the sky localisation capabilities of the detector network.