|
Activity Number:
|
567
- New Approaches for Sparse Gaussian Processes
|
|
Type:
|
Invited
|
|
Date/Time:
|
Thursday, August 6, 2020 : 3:00 PM to 4:50 PM
|
|
Sponsor:
|
Uncertainty Quantification in Complex Systems Interest Group
|
|
Abstract #308026
|
|
|
Title:
|
Distributed Spatiotemporal Inference Embedded in High-Fidelity Physics Simulators Using Sparse Gaussian Processes
|
|
Author(s):
|
Michael Grosskopf* and Earl Lawrence and Nathan Urban and Mary Dorn and Ayan Biswas
|
|
Companies:
|
Los Alamos National Laboratory and Los Alamos National Laboratory and Los Alamos National Laboratory and Los Alamos National Laboratory and Los Alamos National Laboratory
|
|
Keywords:
|
Gaussian Processes;
In Situ Inference;
Exascale Computing
|
|
Abstract:
|
As supercomputers expand to larger scales, the volume of data produced by simulation of physical systems may soon exceed reasonable bottlenecks for I/O communication and storage. In this work we present results using sparse Gaussian processes to model spatio-temporal variations in patterns of extreme events in a physics simulator toward the end goal of embedding the inference inside a high-fidelity physics model to perform on-line inference as the simulation is carried out.
|
Authors who are presenting talks have a * after their name.
