A5066216103- Advanced Statistical Methods and Models
- Gaussian Processes and Bayesian Inference
- Statistical Methods and Inference
- Statistical Methods and Bayesian Inference
- Statistical and numerical algorithms
- Bayesian Methods and Mixture Models
- Galaxies: Formation, Evolution, Phenomena
- Gamma-ray bursts and supernovae
- Astrophysical Phenomena and Observations
- Markov Chains and Monte Carlo Methods
- Spectroscopy and Chemometric Analyses
- Astrophysics and Cosmic Phenomena
- Medical Image Segmentation Techniques
- Neural Networks and Applications
- Statistical Distribution Estimation and Applications
- DNA and Biological Computing
- Adaptive optics and wavefront sensing
- Target Tracking and Data Fusion in Sensor Networks
- Astronomical Observations and Instrumentation
- Spectroscopy Techniques in Biomedical and Chemical Research
- Sports Analytics and Performance
- Forecasting Techniques and Applications
- Geophysics and Gravity Measurements
- Radioactivity and Radon Measurements
- Statistics Education and Methodologies
Pennsylvania State University
2019-2024
Harvard University
2015-2021
University of Notre Dame
2018
Statistical and Applied Mathematical Sciences Institute
2017-2018
Harvard University Press
2017
We present the results of first strong lens time delay challenge. The motivation, experimental design, and entry level challenge are described in a companion paper. This paper presents main challenge, TDC1, which consisted analyzing thousands simulated light curves blindly. observational properties cover range quality obtained for current targeted efforts (e.g., COSMOGRAIL) expected from future synoptic surveys LSST), include systematic errors. Seven teams participated submitting 78...
The growth of supermassive black holes is strongly linked to their galaxies. It has been shown that the population mean black-hole accretion rate ($\overline{\mathrm{BHAR}}$) primarily correlates with galaxy stellar mass ($M_\star$) and redshift for general population. This work aims provide best measurements $\overline{\mathrm{BHAR}}$ as a function $M_\star$ over ranges $10^{9.5}<M_\star<10^{12}~M_\odot$ $z<4$. We compile an unprecedentedly large sample eight thousand active galactic nuclei...
ABSTRACT In recent years, breakthroughs in methods and data have enabled gravitational time delays to emerge as a very powerful tool measure the Hubble constant H0. However, published state-of-the-art analyses require of order 1 yr expert investigator up million hours computing per system. Furthermore, precision improves, it is crucial identify mitigate systematic uncertainties. With this delay lens modelling challenge, we aim assess level accuracy techniques that are currently fast enough...
In preparation for the era of time-domain astronomy with upcoming large-scale surveys, we propose a state-space representation multivariate damped random walk process as tool to analyze irregularly-spaced multi-filter light curves heteroscedastic measurement errors. We adopt computationally efficient and scalable Kalman-filtering approach evaluate likelihood function, leading maximum $O(k^3n)$ complexity, where $k$ is number available bands $n$ unique observation times across bands. This...
Abstract Most general-purpose classification methods, such as support-vector machine (SVM) and random forest (RF), fail to account for an unusual characteristic of astronomical data: known measurement error uncertainties. In data, this information is often given in the data but discarded because popular learning classifiers cannot incorporate it. We propose a simulation-based approach that incorporates heteroscedastic into existing method better quantify uncertainty classification. The...
The well-known Bayes theorem assumes that a posterior distribution is probability distribution. However, the may no longer be if an improper prior (non-probability measure) such as unbounded uniform used. Improper priors are often used in astronomical literature to reflect lack of knowledge, but checking whether resulting sometimes neglected. It turns out 23 articles 75 (30.7%) published online two renowned astronomy journals (ApJ and MNRAS) between Jan 1, 2017 Oct 15, make use Bayesian...
Although the Metropolis algorithm is simple to implement, it often has difficulties exploring multimodal distributions. We propose repelling–attracting (RAM) that maintains simple-to-implement nature of algorithm, but more likely jump between modes. The RAM a Metropolis-Hastings with proposal consists downhill move in density aims make local modes repelling, followed by an uphill attracting. achieved via reciprocal ratio so prefers downward movement. does opposite using standard which upward...
Abstract Cosmological parameters encoding our understanding of the expansion history universe can be constrained by accurate estimation time delays arising in gravitationally lensed systems. We propose TD-CARMA, a Bayesian method to estimate cosmological modeling observed and irregularly sampled light curves as realizations continuous auto-regressive moving average (CARMA) process. Our model accounts for heteroskedastic measurement errors microlensing, an additional source independent...
In the last two decades, Bayesian inference has become commonplace in astronomy. At same time, choice of algorithms, terminology, notation, and interpretation varies from one sub-field astronomy to next, which can lead confusion both those learning familiar with statistics. Moreover, between statistics literature, too. this paper, our goal is two-fold: (1) provide a reference that consolidates clarifies terminology notation across disciplines, (2) outline practical guidance for Highlighting...
A Gaussian measurement error assumption, i.e., an assumption that the data are observed up to noise, can bias any parameter estimation in presence of outliers. heavy tailed based on Student's t distribution helps reduce bias. However, it may be less efficient estimating parameters if is uniformly applied all when most them normally observed. We propose a mixture selectively converts errors into according latent outlier indicators, leveraging best and errors; not only robust but also...
The gravitational field of a galaxy can act as lens and deflect the light emitted by more distant object such quasar. Strong lensing causes multiple images same quasar to appear in sky. Since each gravitationally lensed image traverses different path length from Earth, fluctuations source brightness are observed several at times. time delay between these be used constrain cosmological parameters inferred series data or curves image. To estimate delay, we construct model based on state-space...
We present optical multi-colour photometry of V404 Cyg during the outburst from December, 2015 to January, 2016 together with simultaneous X-ray data. This occurred less than 6 months after previous in June-July, 2015. These two outbursts were a slow rise and rapid decay-type showed large-amplitude ($\sim$2 mag) short-term ($\sim$10 min-3 hours) variations even at low luminosity (0.01-0.1$L_{\rm Edd}$). found correlated $\sim$1 hour time intervals performed Bayesian delay estimations between...
A Beta-Binomial-Logit model is a Beta-Binomial with covariate information incorporated via logistic regression. Posterior propriety of Bayesian can be data-dependent for improper hyper-prior distributions. Various researchers in the literature have unknowingly used posterior distributions or given incorrect statements about because checking challenging due to complicated functional form model. We derive necessary and sufficient conditions within class that encompass those previous studies....
Rgbp is an R package that provides estimates and verifiable confidence intervals for random effects in two-level conjugate hierarchical models overdispersed Gaussian, Poisson, Binomial data. aggregate data from k independent groups summarized by observed sufficient statistics each effect, such as sample means, possibly with covariates. uses approximate Bayesian machinery unique improper priors the hyper-parameters, which leads to good repeated sampling coverage properties effects. A special...
The production of complex astronomical data is accelerating, especially with newer telescopes producing ever more large-scale surveys. increased quantity, complexity, and variety demand a parallel increase in skill sophistication developing, deciding, deploying statistical methods. Understanding limitations appreciating nuances machine learning methods the reasoning behind them essential for improving data-analytic proficiency acumen. Aiming to facilitate such improvement astronomy, we...
We propose a variant of Hamiltonian Monte Carlo (HMC), called the Repelling-Attracting (RAHMC), for sampling from multimodal distributions. The key idea that underpins RAHMC is departure conservative dynamics systems, which form basis traditional HMC, and turning instead to dissipative conformal systems. In particular, involves two stages: mode-repelling stage encourage sampler move away regions high probability density; and, mode-attracting stage, facilitates find settle near alternative...
Abstract The acquisition of complex astronomical data is accelerating, especially with newer telescopes producing ever more large-scale surveys. increased quantity, complexity, and variety demand a parallel increase in skill sophistication developing, deciding, deploying statistical methods. Understanding limitations appreciating nuances machine learning methods the reasoning behind them essential for improving data-analytic proficiency acumen. Aiming to facilitate such improvement...
A uniform shrinkage prior (USP) distribution on the unknown variance component of a random-effects model is known to produce good frequency properties. The USP has parameter that determines shape its density function, but it been neglected whether can maintain such properties regardless choice for parameter. We investigate which produces Bayesian interval estimates random effects meet their nominal confidence levels better than several existent choices in literature. Using univariate and...
Data augmentation (DA) turns seemingly intractable computational problems into simple ones by augmenting latent missing data. In addition to simplicity, it is now well-established that DA equipped with a deterministic transformation can improve the convergence speed of iterative algorithms such as an EM algorithm or Gibbs sampler. this article, we outline framework for transformation-based DA, which call data transforming (DTA), allowing augmented be function and observed data, unknown...
We propose a Bayesian meta-analysis to infer the current expansion rate of Universe, called Hubble constant ($H_0$), via time delay cosmography. Inputs are estimates two properties for each pair gravitationally lensed images; and Fermat potential difference with their standard errors. A can be appealing in practice because obtaining estimate from even single lens system involves substantial human efforts, thus often separately obtained published. This work focuses on combining these...