High Performance Statistical Computing (HPSC): Challenges and Best Practices

Overview

**************************************************

Organizers: Sameh Abdulah and Marc Genton 

**************************************************

Abstract

Big data in terms of volume, intensity, and complexity is one of the main challenges to statisticians. The availability of different data sources has essential implications in collecting colossal data volumes that require other ways to be managed and analyzed. Existing analytical tools cannot be applied easily to big data volumes due to memory and computation constraints. Previously, statistical applications and traditional high performance oriented computing have followed independent paths. However, important opportunities now arise that can be addressed by merging the two. As a prominent big data application, statistics is increasingly performance-bound in different fields. HPC is becoming increasingly significant in scaling existing statistical methods to larger and more complex applications and developing novel methods that are amenable to scaling within the constraints that exist in modern HPC architectures. This minisymposium aims to show the existing efforts to harness the HPC capabilities in different statistics branches to serve large-scale statistics. It also aims to cover the current challenges and opportunities towards exploiting current HPC technologies in accelerating applications related to applied statistics.

**************************************************

SIAM PP22 will run virtually with live sessions. 

**************************************************

Registration is required to gain access to the online platform.

Please use your personal login information (sent by email) to join the online virtual platform (vFairs) and attend this exciting and timely minisymposium.

All times are listed in the CST time zone (+9hrs in Arabic Standard Time).

Talks are 20min including 5min Q&A.

MS27 (PART I) - February 24th, 2022: 10:55 AM - 12:35 PM (CST)

10:55-11:15 Accelerating Space and Space-Time Statistical Modeling with Mixed-Precision Arithmetic (abstract, slides)

Sameh Abdulah, King Abdullah University of Science & Technology (KAUST), Saudi Arabia.

11:20-11:40 Scalable Statistical Learning from Dependent Data (abstract, slides)

Michael Schweinberger,  University of Missouri, Columbia, U.S.

11:45-12:05 High Performance Computing in the Context of General Software for Hierarchical Statistical Modeling (abstract, slides)

Christopher Paciorek, University of California, Berkeley, U.S.

12:10-12:30 On Conjugate Bayesian Linear Regression Frameworks for High-Dimensional Spatial Modeling (abstract, slides)

Sudipto Banerjee, University of California, Los Angeles, US.

MS40 (PART II) - February 24th, 2022: 3:20 PM - 5:00 PM (CST)

3:20-3:40 High-Performance Bayesian Computing with Applications to Spatial/Spatiotemporal Data Analysis and Genome-Wide Association Studies (abstract, slides)

Marco Ferreira, Virginia Tech, U.S.

3:45-4:05 Scaling Shrinkage Estimates of Large Covariance and Correlation Matrices (abstract, slides)

Alina Peluso, Oak Ridge National Laboratory, U.S.

4:10-4:30 Fine-Tuning the Grouping Approach to Parallelization of Statistics/Machine Learning Methods (abstract, slides)

Norman Matloff, University of California, Davis, U.S.

4:35-4:55 Supercomputing-Driven Climatic Clustering and Longitudinal Analysis with Impacts on Food, Bioenergy, and Pandemics (abstract, slides)

Daniel A. Jacobson, Oak Ridge National Laboratory, U.S.