BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Dr Jeffrey D. Scargle (NASA Ames Research Center\, US) DTSTART:20211012T160000Z DTEND:20211012T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/1 DESCRIPTION:Title: Adventures in Astronomical Time Series Analysis\nby Dr Jeffr ey D. Scargle (NASA Ames Research Center\, US) as part of IAU-IAA Astrosta ts & Astroinfo seminar(archived version by January 2023)\n\n\nAbstract\nWe lcome to a tour of the volatile\, highly active Universe — in stark cont rast to earlier serene '"clockwork’’ visions. Innovative data analysis techniques have illuminated explosive physical processes animating these systems. Examples include a Fourier transform suited to the irregular samp ling characteristic of much astronomical data\, but time domain techniques will be emphasized for these applications: gamma-ray activity in the Crab Nebula\, gamma-ray bursts\, active galactic nuclei\, and gravitational wa ves. I hope this talk will change some of the ways you carry out statistic al data analysi\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/1/ END:VEVENT BEGIN:VEVENT SUMMARY:Prof Ilya Mandel (Monash University\, Australia) DTSTART:20211109T080000Z DTEND:20211109T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/2 DESCRIPTION:Title: Astrostatistics in Gravitational-wave Astronomy\nby Prof Ily a Mandel (Monash University\, Australia) as part of IAU-IAA Astrostats & A stroinfo seminar(archived version by January 2023)\n\n\nAbstract\nModern a stronomical data sets often raise challenges associated with selection bia ses\, accounting for confusion between backgrounds and foregrounds\, and p erforming inference on big data with complex\, multi-parameter models. I w ill discuss some of the techniques that we used to attack these problems\, illustrating them with results from gravitational-wave observations of me rging black holes … and a bit further afield.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/2/ END:VEVENT BEGIN:VEVENT SUMMARY:Dr Torsten Enßlin (Max-Planck-Institute for Astrophysics\, German y) DTSTART:20211214T160000Z DTEND:20211214T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/3 DESCRIPTION:Title: Information field theory\, from astronomical imaging to artifici al intelligence\nby Dr Torsten Enßlin (Max-Planck-Institute for Astro physics\, Germany) as part of IAU-IAA Astrostats & Astroinfo seminar(archi ved version by January 2023)\n\n\nAbstract\nTurning the raw data of an ins trument into high-fidelity pictures of the Universe is a central theme in astronomy. Information field theory (IFT) describes probabilistic image re construction from incomplete and noisy data exploiting all available infor mation. Astronomical applications of IFT are galactic tomography\, gamma- and radio- astronomical imaging\, and the analysis of cosmic microwave bac kground data. This talk introduces into the basic ideas of IFT\, highlight s its astronomical applications\, and explains its relation with contempor ary artificial intelligence.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/3/ END:VEVENT BEGIN:VEVENT SUMMARY:Makoto Uemura (Hiroshima University\, Japan) DTSTART:20220111T080000Z DTEND:20220111T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/4 DESCRIPTION:Title: Follow-up observations of galactic transients with astroinformat ics\nby Makoto Uemura (Hiroshima University\, Japan) as part of IAU-IA A Astrostats & Astroinfo seminar(archived version by January 2023)\n\n\nAb stract\nMethods such as Bayesian inference and machine learning have recen tly become readily available\, and are used not only on state-of-the-art d ata\, but also in various aspects of astronomical research. Our group has a 1.5-m optical telescope in Hiroshima\, Japan\, which is used for time-d omain astronomy. I will talk about the applications of astroinformatics to ols for the follow-up observations of galactic transients. The topics incl ude the discriminative model of transients\, decision making based on the information theory\, and reconstruction of the geometrical structure of th e accretion disk.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/4/ END:VEVENT BEGIN:VEVENT SUMMARY:Dan Foreman-Mackey (Flatiron Institute) DTSTART:20220208T160000Z DTEND:20220208T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/5 DESCRIPTION:Title: Methods for scalable probabilistic inference\nby Dan Foreman -Mackey (Flatiron Institute) as part of IAU-IAA Astrostats & Astroinfo sem inar(archived version by January 2023)\n\n\nAbstract\nMost data analysis p ipelines in astrophysics now have some steps that require detailed probabi listic modeling. As datasets get larger and our research questions get mor e ambitious\, we are often pushing the limits of what our statistical fram eworks are capable of. In this talk\, I will discuss recent (and not so re cent) developments in the field probabilistic programming that enable rigo rous Bayesian inference with large datasets\, and high-dimensional or comp utationally expensive models. In particular\, I will highlight some scalab le methods for time series analysis using Gaussian Processes\, and some of the open source tools and computational techniques that have the potentia l to be broadly useful for accelerating inference in astrophysics.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/5/ END:VEVENT BEGIN:VEVENT SUMMARY:Renate Meyer (University of Auckland\, New Zealand) DTSTART:20220308T080000Z DTEND:20220308T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/6 DESCRIPTION:Title: Bayesian Nonparametric Spectral Analysis for Gravitational Wave Astronomy\nby Renate Meyer (University of Auckland\, New Zealand) as p art of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 2023)\n\n\nAbstract\nThe new era of gravitational wave astronomy truly beg an on September 14\, 2015 with the sensational first direct observation of gravitational waves\, when LIGO recorded the signature of the merger of t wo black holes. In the subsequent three observing runs of the LIGO/Virgo n etwork\, gravitational waves from 90 compact binary mergers have been ann ounced. Moreover\, the future space-based observatory LISA will open the l ow-frequency window on gravitational waves and will be sensitive to a vast range of sources including the white dwarf binaries in our Milky Way and mergers of supermassive black holes at the centre of galaxies. Beyond sign al detection\, a major challenge has been the development of statistical m ethodology for estimating the physical waveform parameters and quantifying their uncertainties. Bayesian methods and MCMC have played a key role in this new era of astrophysics. I will review the statistical methods that e nabled the estimation of the waveform parameters. This challenge has also been a key driver for new theoretical and methodological advancements in s tatistics. The call for a more robust instrumental noise characterization aiming at a simultaneous estimation of noise characteristics and gravitati onal wave parameters has triggered ongoing research into Bayesian nonparam etric analysis of time series. Starting with nonparametric Bayesian approa ches to spectral density estimation of univariate Gaussian stationary time series\, I will review novel extensions to multivariate\, non-Gaussian\, and locally stationary time series.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/6/ END:VEVENT BEGIN:VEVENT SUMMARY:Roberto Trotta (SISSA\, Italy) DTSTART:20220614T160000Z DTEND:20220614T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/7 DESCRIPTION:Title: A general-purpose method for supervised learning under covariate shift with applications to observational cosmology\nby Roberto Trotta (SISSA\, Italy) as part of IAU-IAA Astrostats & Astroinfo seminar(archive d version by January 2023)\n\n\nAbstract\nSupervised machine learning will be central in the analysis of upcoming large-scale sky surveys. However\, selection bias for astronomical objects yields labelled training data tha t are not representative of the unlabelled target data distribution. This affects the predictive performance with unreliable target predictions and poor generalization. I will present StratLearn\, a novel and statistically principled method to improve supervised learning under such covariate shi ft conditions\, based on propensity score stratification. In StratLearn\, learners are trained on subgroups ("strata") of the data conditional on th e propensity scores\, leading to improved covariate balance and much-reduc ed bias in the model fit. This general-purpose method has promising applic ations in observational cosmology\, improving upon existing conditional de nsity estimation of galaxy redshift from Sloan Data Sky Survey (SDSS) data \; in the classification of Supernovae (SNe) type Ia from photometric data \, it obtains the best reported AUC on the SNe photometric classification challenge. If time allows\, I'll discuss the embedding of such a classific ation into a full analysis of SNe data to estimate cosmological parameters .\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/7/ END:VEVENT BEGIN:VEVENT SUMMARY:Josh Speagle (Toronto University\, Canada) DTSTART:20220412T160000Z DTEND:20220412T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/8 DESCRIPTION:Title: Statistical Challenges in Stellar Parameter Estimation from Theo ry and Data\nby Josh Speagle (Toronto University\, Canada) as part of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 2023)\n \n\nAbstract\nUnderstanding how the Milky Way fits into the broader galaxy population requires studying the properties of other galaxies as well as our own. While it is possible to observe the structure of other galaxies d irectly\, understanding the structure of our own Galaxy from within requir es inferring the 3-D positions\, velocities\, and other properties of bill ions of stars. In this talk\, I will discuss some of the statistical chall enges in inferring stellar parameters from modern photometric surveys such as Gaia and SDSS\, focusing in particular on issues with existing theoret ical stellar models\, the complex nature of parameter uncertainties\, and scalability to large datasets. I will then describe some ongoing work tryi ng to solve these problems using a combination of physics-inspired but dat a-driven calibrations along with a host of inference approaches including gradient-based optimization\, grid-based searches\, importance sampling\, and nested sampling.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/8/ END:VEVENT BEGIN:VEVENT SUMMARY:Takahiko Matsubara (KEK\, Japan) DTSTART:20220510T080000Z DTEND:20220510T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/9 DESCRIPTION:Title: Weakly non-Gaussian formulas of cosmological random fields\n by Takahiko Matsubara (KEK\, Japan) as part of IAU-IAA Astrostats & Astroi nfo seminar(archived version by January 2023)\n\n\nAbstract\nIn cosmology\ , various kinds of random fields play important roles\, including 3D distr ibutions of galaxies and other astronomical objects\, 2D distributions of cosmic microwave background radiations and weak lensing fields\, etc. The features of non-Gaussianity in these fields contain a lot of cosmological information. In this talk\, I will present a method to analytically descri be the effects of weak non-Gaussianity in field statistics\, such as the p eak abundance\, peak correlations\, Minkowski functionals\, etc.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/9/ END:VEVENT BEGIN:VEVENT SUMMARY:Eric Thrane (Monash University\, Australia) DTSTART:20220809T080000Z DTEND:20220809T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/10 DESCRIPTION:Title: The population properties of merging compact binaries from grav itational waves.\nby Eric Thrane (Monash University\, Australia) as pa rt of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 2 023)\n\n\nAbstract\nWith the publication of the third gravitational-wave t ransient catalog (GWTC-3)\, the LIGO and Virgo Collaborations have confide ntly identified 90 signals from merging compact binaries. By analysing the morphology of each gravitational waveform\, we are able to work out the m asses and spins of the black holes and neutron stars that source these sig nals. By studying the distributions of black-hole mass\, spin\, and distan ce\, we are painting a picture of the population properties of compact mer gers\, providing clues about the fate of massive stars and telling us how and where binary black holes are assembled. In this talk\, I describe how we use Bayesian hierarchical modelling to study merging black holes. I emp hasise the importance of model checking to avoid faulty conclusions from m odel misspecification.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/10/ END:VEVENT BEGIN:VEVENT SUMMARY:Shiro Ikeda (Institute of Statistical Mathematics\, Japan) DTSTART:20220913T080000Z DTEND:20220913T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/11 DESCRIPTION:Title: Data Science and Imaging the Black Hole Shadow\nby Shiro Ik eda (Institute of Statistical Mathematics\, Japan) as part of IAU-IAA Astr ostats & Astroinfo seminar(archived version by January 2023)\n\n\nAbstract \nIn April 2019\, the EHTC (Event Horizon Telescope collaboration) release d the first image of the M87 black hole shadow and this May\, the black ho le shadow image of our Milky Way galaxy was released. The EHTC has more th an 300 members from different backgrounds and countries. I have been invol ved in this project as a data scientist for more than 8 years and collabor ated with EHTC members to develop a new imaging method. The EHT is a huge VLBI (very long baseline interferometer)\, which is different from optical telescopes in that a lot of computation is required to obtain a single im age. Black hole imaging is also very interesting from the data scientific viewpoint. In this talk\, I will explain how the new imaging technique has been developed and the final images were created through our discussions. \n LOCATION:https://stable.researchseminars.org/talk/Astrostats/11/ END:VEVENT BEGIN:VEVENT SUMMARY:Jason McEwen (University College London\, UK) DTSTART:20221011T160000Z DTEND:20221011T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/12 DESCRIPTION:Title: Bayesian model selection for likelihood-based and simulation-ba sed inference.\nby Jason McEwen (University College London\, UK) as pa rt of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 2 023)\n\n\nAbstract\nIn the study of cosmology\, where we seek to uncover a n understanding of the fundamental physical processes underlying the origi n\, content\, and evolution of our Universe\, we are not blessed with the ability to perform experiments - rather\, we have only one Universe to obs erve. In this scenario\, while we are of course interested in estimating the parameters of models describing the physical processes observed\, we a re often most interested in selecting the best underlying model\, which ha s given rise to the prevalence of Bayesian model selection in cosmology an d astrophysics. While I will motivate recent developments in Bayesian mod el selection from problems in cosmology and astrophysics\, I will mostly f ocus on new methodological advances. I will discuss new approaches that l everage ideas across statistics\, optimization and machine learning to bri ng to bear the respective strengths of these paradigms to the highly compu tationally challenging problem of Bayesian model selection. In particular\ , I will review the learnt harmonic mean estimator for both likelihood-bas ed and simulation-based inference and the proximal nested sampling framewo rk for high-dimensional model selection.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/12/ END:VEVENT BEGIN:VEVENT SUMMARY:A/Prof. Aaron Robotham (University of Western Australia) DTSTART:20221108T080000Z DTEND:20221108T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/13 DESCRIPTION:Title: Exploring the Limits of the Bayesian Universe: How to Tackle Br eadth and Depth\nby A/Prof. Aaron Robotham (University of Western Aust ralia) as part of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 2023)\n\n\nAbstract\nIn the last 10 years it is notable that st udents are much more enthused about projects involving “machine learning ”\, but it is important we do not lose perspective on the scientific ins ights still offered by a comprehensive and pragmatic application of Bayesi an principles. Here I will discuss the work my group has undertaken over t he last 7 years to build up a fully generative model of galaxies that has culminated in the Bayesian modelling software ProFuse (Robotham+ 2022). Th e positive is that encoding our knowledge and ignorance in a Bayesian mann er has opened up new insights to physical processes that form galaxies\, t he negative is that this approach has a high barrier of entry which can be a poor fit to a modern ~3 year PhD\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/13/ END:VEVENT BEGIN:VEVENT SUMMARY:Dr Jessi Cisewski Kehe (University of Wisconsin) DTSTART:20221213T160000Z DTEND:20221213T170000Z DTSTAMP:20260404T110821Z UID:Astrostats/14 DESCRIPTION:Title: Getting something out of nothing: topological data analysis fo r cosmology\nby Dr Jessi Cisewski Kehe (University of Wisconsin) as pa rt of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 2 023)\n\n\nAbstract\nThe transference from data to information is a key com ponent of many areas of research in astronomy and cosmology. This process can be challenging when data exhibit complicated spatial structures\, suc h as the large-scale structure (LSS) of the Universe. Methods that target shape-related features may be helpful for summarizing qualitative propert ies that are not retrieved with standard techniques. Topological data ana lysis (TDA) provides a framework for quantifying shape-related properties of data. Persistent homology is a popular TDA tool that offers a procedur e to represent\, visualize\, and interpret complex data by extracting topo logical features which may be used to infer properties of the underlying s tructures. Persistent homology is used to find different dimensional hole s in a dataset across different scales\, where zero-dimensional holes are clusters\, one-dimensional holes are closed loops\, two-dimensional holes are voids\, and so on. The information is summarized in a persistence dia gram\, which may be used for further analysis such as visualization\, infe rence\, or classification. I will give an overview of persistent homology and discuss its use in some cosmology applications\, such as discriminati ng LSS under varying cosmological assumptions.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/14/ END:VEVENT BEGIN:VEVENT SUMMARY:A/Prof. Yuan-Sen Ting (Australian National University) DTSTART:20230110T080000Z DTEND:20230110T090000Z DTSTAMP:20260404T110821Z UID:Astrostats/15 DESCRIPTION:Title: Galaxy Merger Reconstruction with Generative Graph Neural Netwo rks\nby A/Prof. Yuan-Sen Ting (Australian National University) as part of IAU-IAA Astrostats & Astroinfo seminar(archived version by January 202 3)\n\n\nAbstract\nA key yet unresolved question in modern-day astronomy is how galaxies formed and evolved. The quest to understand how galaxies evo lve has led many semi-analytic models to infer the galaxy properties from their merger history. However\, most classical approaches rely on studying the global connection between dark matter haloes and galaxies\, often red ucing the study to crude summary statistics. The recent advancement in gra ph neural networks might open up many new possibilities\; graphs are a nat ural descriptor of galaxy progenitor systems – any progenitor system at a high redshift can be regarded as a graph\, with individual progenitors a s nodes on the graph. In this presentation\, I will discuss the power of g enerative graph neural networks to connect high-redshift progenitor system s with local observables. We showed that based on equivariant graph normal izing flow\, our model could robustly recover the progenitor systems\, inc luding their masses\, merging redshifts and pairwise distances at redshift z = 2 conditioned on their z = 0 properties. In addition\, the probabilis tic nature of our model enables other downstream tasks\, including detecti ng anomalies in galaxy configuration and identifying subtle correlations o f the progenitor features.\n LOCATION:https://stable.researchseminars.org/talk/Astrostats/15/ END:VEVENT END:VCALENDAR