Gustavo Chavez

Gustavo Chávez

Computer Science, Ph.D.

Postdoctoral researcher at the Lawrence Berkeley National Laboratory (LBNL) working on the Exascale Computing Project (ECP) . Previously at the Extreme Computing Research Center (ECRC) at the King Abdullah University of Science and Technology (KAUST).


Background

I develop high-performance implementations of machine learning algorithms.

I graduated summa cum laude from UNITEC as a software engineer. After a brief period in industry, I moved to Saudi Arabia to do a master’s degree in Applied Mathematics at KAUST.

After that, I completed a Ph.D. in Computer Science. My doctoral dissertation is about high-performance linear solvers based on hierarchical matrices.

My current position is postdoctoral researcher at the Lawrence Berkeley National Laboratory.

I have authored conference and journal papers, patent applications, and technical reports, some of them in collaboration with industry partners such as Saudi Aramco, Boeing, and CISCO.

My day job is to contribute to the U.S. Department of Energy-funded scalable linear solver STRUMPACK, as well as to perform research in numerical linear algebra and its intersection with machine learning.

I am passionate about technology, high-performance computing, and machine learning.


Publications

Robust and Accurate Stopping Criteria for Adaptive Randomized Sampling In Matrix-Free HSS Construction

C. Gorman, G. Chávez, P. Ghysels, T. Mary,
F.-H. Rouet, X. Sherry Li.
SIAM Journal on Scientific Computing (2019)

Parallel accelerated cyclic reduction preconditioner for three-dimensional elliptic PDEs with variable coefficients

G. Chávez, G. Turkiyyah, S. Zampini, D. Keyes.
Elsevier Journal of Computational Applied Mathematics (2017)

Accelerated cyclic reduction: a distributed-memory fast solver for structured linear systems

G. Chávez, G. Turkiyyah, S. Zampini, H. Ltaief, D. Keyes.
Elsevier Journal of Parallel Computing (2016)

Marching surfaces: Isosurface approximation using G1 multi-sided surfaces

G. Chávez, A. Rockwood.
arXiv:1502.02139 (2014)

A study of clustering techniques and hierarchical matrix formats for kernel ridge regression

E. Rebrova, G. Chávez, Y. Liu, P. Ghysels, X. Sherry Li.
IEEE Workshop on Parallel and Distributed Computing for Large-Scale Machine Learning and Big Data Analytics (2018)

Robust and scalable hierarchical matrix-based fast direct solver and preconditioner for the numerical solution of elliptic partial differential equations

G. Chávez.
Ph.D. thesis (2017)

A direct elliptic solver based on hierarchically low-rank Schur complements

G. Chávez, G. Turkiyyah, D. Keyes.
Domain Decomposition Methods in Science and Engineering XXIII (2015)

Lightweight visualization for high-quality materials on WebGL

G. Chávez, F. Ávila, A Rockwood.
Proceedings of the 18th International Conference on 3D Web Technology (2013)


Recent and Upcoming Public Speaking

February 25 th, 2019 at Spokane, Washington, USA.
SIAM Conference on Computational Science and Engineering (CSE19). Workshop on numerical linear algebra for machine learning

June 18th, 2018. Berkeley CA, USA.
13th Scheduling For Large Scale Systems Workshop

March 8th, 2018 at Tokyo, Japan.
18th SIAM Conference on Parallel Processing for Scientific Computing (PP18). Hierarchical low-rank approximation methods

Dec 7th, 2018 at Sandia National Laboratories in Livermore, USA.
Bay Area Scientific Computing Day 2018 (BASCD18). Physics-informed machine learning

May 21th, 2018 at Vancouver, Canada.
7th International Workshop on Parallel and Distributed Computing for Large Scale Machine Learning and Big Data Analytics (ParLearning18)

February 7th 2018. Knoxville TN, USA.
Exascale Computing Project 2nd Annual Meeting


Last updated on Feb 14th, 2019.