Spring/Summer 2023


Group meetings are for students and faculty at CMU and UPitt.
The STAMPS webinars are open to everyone.

These meetings are Fridays 1:30-2:30 PM. Meetings will be in-person in Baker Hall (BH) 232M (second floor of the StatDS Department in Baker Hall). If that plan is to change, we will send out an alert via the mailing list.

January 20 - No Meeting #


January 27, 1:30PM EST - Public Webinar : Bobby Gramacy (Department of Statistics at Virginia Tech)


February 3 - James Carzon (CMU Stats&DS) #

Title: Statistical Constraints on Climate Model Parameters Using the A-CURE Perturbed Parameter Ensemble
Description: Atmospheric aerosols have an uncertain effect on the Earth’s climate. Constraining the aerosol-related parameters of a climate model can help reduce this uncertainty. We propose a statistical framework for constraining the parameters of an expensive simulator when a perturbed parameter ensemble is available. Using the C3.ai Suite, a cloud computing platform, we efficiently train a surrogate for the UKESM1 climate model. The discrepancy of the surrogate is estimated in a data driven way. The strict bounds method is applied for principled quantification of parametric uncertainty. We demonstrate the scalability of this framework with three-hourly aerosol optical depth model outputs over the South Atlantic and mainland African region during the first two weeks of July 2017. To our knowledge, the resulting simultaneous observation-based constraints on atmospheric aerosol input parameters are the first of this kind which employ such high time resolution data.

February 10 - Alex Malz (McWilliams Center for Cosmology) #

Title: Challenges and Opportunities in the Rubin Observatory Ecosystem
Description: During its ten year mission, the Legacy Survey of Space and Time (LSST) on the Vera C. Rubin Observatory will amass enormous catalogs of tens of billions of galaxies and hundreds of millions of time-varying transient events, two orders of magnitude greater than all observed objects to date. To accomplish this feat, it will collect coarser data that can probe the faint universe – if it were easy to see, it would already have been seen! – introducing obstacles to exploiting this data set’s potential to revolutionize our understanding of the dark universe. I will discuss a few such challenges and current work leveraging data science to enable astrophysical inference with low information density measurements at an unprecedented scale, as well as opportunities to get involved in solving open problems presently under investigation here at CMU.

February 17 - Gourav Khullar (Department of Physics and Astronomy, University of Pittsburgh) #

Title: Fast and Robust Inference of Galaxy Observations with Simulation-Based Inference (SBI)
Description: A pressing question in the field of cosmological structure formation is how the long-term assembly and evolution of baryonic matter occurs in galaxies. Galaxies take different pathways to assemble their stellar mass, signatures of which can be derived from galaxy star formation histories via stellar population synthesis (SPS) modeling. Today, we are approaching the age of trillion-galaxy surveys (DECaLS, DESI, Rubin) containing photometry and spectroscopy (describing a given galaxy’s spectral energy distribution, or SED). It is imperative that fast and efficient methods — beyond traditional Bayesian MCMC methods — are built to constrain galaxy parameters for these large samples. The combination of machine learning (ML) and simulation-based inference (SBI) is a promising path forward.

In this work, I show results from the DIGS framework – Deep Inference of Galaxy Spectra — our SBI analysis using the Python package `sbi’ to perform Sequential Neural Posterior Estimation (SNPE) and obtain multi-parameter SED fits. I demonstrate a proof-of-concept study of spectra that is a) an efficient analysis of galaxy SEDs and inference of galaxy parameters with physically interpretable uncertainties; and b) amortized calculations of posterior distributions of said galaxy parameters at the modest cost of a few galaxy fits with MCMC methods. I share my observations of the future directions of SBI usage in the field of galaxy evolution.

February 24, 1:30PM EST - Public Webinar : Aneta Siemiginowska (Harvard-Smithsonian Center for Astrophysics)


March 3 - No Meeting #


March 10 - No Meeting #


March 17 - No Meeting #


March 24 - McWilliams Center for Cosmology Wine and Cheese Talk with Luca Masserano at 3:00pm (Wean Hall 7316) #


March 31, 1:30PM EST - Public Webinar : Pietro Vischia (University of Oviedo and ICTEA)


April 7 - Sindhu Murthy (CMU Department of Physics) #

Title: Background modelling for a semi-supervised search for the four b-jet final state
Description: In search of physics beyond the Standard Model (BSM), we are developing an algorithm for a semi-supervised search for a signal in experimental data. The goal is to devise a data-dependent, model-agnostic method to find a signal that decays as A → XY → four b-jets. We use detector data from the Compact Muon Solenoid (CMS) experiment for proton-proton collisions at the Large Hadron Collider (LHC). We will discuss our method for background modelling and the challenges that come with it.

April 14 - No Meeting #


April 21, 1:30PM EST - Public Webinar : Jonathan Hobbs (Jet Propulsion Laboratory)


April 28 - No Meeting #


May 5 - No Meeting #


May 12 - No Meeting #


May 19 - Noelia Grande Gutiérrez (CMU Dept. Mechanical Engineering) #

Title: Image-based blood flow simulation and multi-scale modeling for patient-specific risk-stratification in cardiovascular medicine
Abstract: Computational methods are emerging as relevant tools for diagnosing, treating, and quantifying disease progression. This seminar will explore how we can leverage multi-physics computational simulations to provide improved patient-specific risk-stratification metrics and enable the design of personalized therapies at a reasonable computational cost. I will introduce patient-specific image-based cardiovascular modeling and how it can support the clinical decision-making process for patient diagnosis and treatment. This framework combines a deep understanding of cardiovascular physiology with advanced numerical methods and high-performance computing to obtain patient-specific hemodynamic data. I will discuss a few examples of applying these tools to compute coronary artery hemodynamics. Via computational experiments, we investigate how structural differences in the coronary vasculature affect global hemodynamics and transport to identify patients at higher risk of adverse cardiovascular events. Finally, I will discuss current strategies and identify opportunities to improve the clinical translation potential of computational simulations using reduced-order and data-driven modeling.

May 26 - Alex Shen (CMU Stats&DS) #

Title: LF2I and Atmospheric Gamma-Ray Showers: Outlier Detection under the Presence of Systematic Uncertainties

August 3 - Mike Stanley (CMU Stats&DS) #

NOTE: This talk will occur from 2:30 to 3:30pm in BH 232M.
Title: TBA