Edgar Solomonik

Assistant Professor
Department of Computer Science
University of Illinois at Urbana-Champaign


solomon2@illinois.edu
4229 Thomas M. Siebel Center for Computer Science
201 North Goodwin Avenue
Urbana, IL 61801-2302

Announcements

(Jan 2017) TOPC paper appears with a technique that yields numerous novel bounds on communication and synchronization complexity of numerical algorithms.

(Jan 2017) New paper led by Master's student Tobias Wicky on parallel TRSM algorithms accepted to IPDPS.

(Dec 2016) Finished building and teaching an advanced course of 29 lectures (36 hours) on communication complexity of numerical and combinatorial algorithms!

Teaching


(Fall 2017) CS 554: Parallel Numerical Algorithms




(Spring 2017) CS 357: Numerical Methods




(Fall 2016) CS 598: Communication Cost Analysis of Algorithms

Research Interests

parallel algorithms
numerical linear algebra
communication complexity
algebraic programming abstractions
computational quantum chemistry
high performance computing
graph theory

Software

Cyclops Tensor Framework
algebraic parallel programming

CANDMC
distributed-memory matrix factorization algorithms

Fellowships and Awards

ETH Postdoctoral Fellowship 2014-2016
DOE Computational Science Graduate Fellowship 2010-2014
David J. Sakrison Memorial Prize 2014
NERSC Award for Innovative Use of High Performance Computing 2013
ACM-IEEE George Michael HPC Fellowship 2013

Previous Affiliations

Postdoctoral Researcher, ETH Zurich, 2014-2016
PhD, University of California Berkeley, 2010-2014
BS, University of Illinois at Urbana-Champaign, 2008-2010
More information is available in my CV

Publications

Maciej Besta, Michal Podstawski, Linus Groner, Edgar Solomonik, and Torsten Hoefler To push or to pull: on reducing communication and synchronization in graph computations 26th ACM Symposium on High Performance Parallel and Distributed Computing, Washington DC, June 2017 (to appear).
report Edgar Solomonik, Grey Ballard, James Demmel, and Torsten Hoefler A communication-avoiding parallel algorithm for the symmetric eigenvalue problem 29th ACM Symposium on Parallelism in Algorithms and Architectures (SPAA), Washington DC, July 2017 (to appear), arXiv:1604.03703 [cs.DC], April 2016. bibtex
report Tobias Wicky, Edgar Solomonik, and Torsten Hoefler Communication-avoiding parallel algorithms for solving triangular systems of linear equations IEEE International Parallel and Distributed Processing Symposium (IPDPS), Orlando, Florida, June 2017 (to appear), arXiv:1612.01855 [cs.DC], December 2016. bibtex
Maciej Besta, Florian Marending, Edgar Solomonik and Torsten Hoefler. SlimSell: A vectorizable graph representation for breadth-first search IEEE International Parallel and Distributed Processing Symposium (IPDPS), Orlando, Florida, June 2017 (to appear).
paper Edgar Solomonik, Erin Carson, Nicholas Knight, and James Demmel Tradeoffs between synchronization, communication, and computation in parallel linear algebra computations ACM Transactions on Parallel Computing, January 2017. bibtex
report Edgar Solomonik, Maciej Besta, Flavio Vella, and Torsten Hoefler Betweenness centrality is more parallelizable than dense matrix multiplication arXiv:1609.07008 [cs.MS], September 2016. bibtex
report Edgar Solomonik and Torsten Hoefler Sparse tensor algebra as a parallel programming model arXiv:1512.00066 [cs.MS], November 2015. bibtex
report Edgar Solomonik, James Demmel, and Torsten Hoefler Communication lower bounds for tensor contraction algorithms Technical report, ETH Zurich, January 2015. bibtex
report Edgar Solomonik and James Demmel Contracting symmetric tensors using fewer multiplications Technical report, ETH Zurich, January 2015. bibtex
thesis Edgar Solomonik Provably efficient algorithms for numerical tensor algebra Dissertation, University of California, Berkeley, August 2014. bibtex
paper Edgar Solomonik, Devin Matthews, Jeff R. Hammond, John F. Stanton, and James Demmel A massively parallel tensor contraction framework for coupled-cluster computations Journal of Parallel and Distributed Computing, June 2014. report bibtex
paper Edgar Solomonik, Erin Carson, Nicholas Knight, and James Demmel Tradeoffs between synchronization, communication, and work in parallel linear algebra computations 26th ACM Symposium on Parallelism in Algorithms and Architectures (SPAA), Prague, Czech Republic, June 2014. report bibtex Superseded by extended journal paper
paper Grey Ballard, James Demmel, Laura Grigori, Mathias Jacquelin, Hong Diep Nguyen, and Edgar Solomonik Reconstructing Householder vectors from Tall-Skinny QR IEEE International Parallel and Distributed Processing Symposium (IPDPS), Phoenix, AZ, May 2014. report bibtex
paper Edgar Solomonik, Devin Matthews, Jeff Hammond, and James Demmel Cyclops Tensor Framework: reducing communication and eliminating load imbalance in massively parallel contractions IEEE International Parallel and Distributed Processing Symposium (IPDPS), Boston, MA, May 2013. report bibtex
paper Edgar Solomonik, Aydin Buluc, and James Demmel Minimizing communication in all-pairs shortest-paths IEEE International Parallel and Distributed Processing Symposium (IPDPS), Boston, MA, May 2013. report bibtex
paper Michael Driscoll, Evangelos Georganas, Penporn Koanantakool, Edgar Solomonik, and Katherine Yelick A communication-optimal n-body algorithm for direct interactions IEEE International Parallel and Distributed Processing Symposium (IPDPS), Boston, MA, May 2013. bibtex
paper Evangelos Georganas, Jorge Gonzalez-Dominguez, Edgar Solomonik, Yili Zheng, Juan Tourino and Katherine Yelick Communication avoiding and overlapping for numerical linear algebra ACM/IEEE Supercomputing Conference, Salt Lake City, Utah, November 2012. report bibtex
paper Edgar Solomonik and James Demmel Matrix multiplication on multidimensional torus networks Springer Lecture Notes in Computer Science (LNCS), Proceedings of VECPAR, Kobe, Japan, July 2012. report bibtex
report Edgar Solomonik, Jeff Hammond, and James Demmel A preliminary analysis of Cyclops Tensor Framework EECS Department, University of California, Berkeley, March 2012. bibtex
paper Edgar Solomonik, Abhinav Bhatele, and James Demmel Improving communication performance in dense linear algebra via topology aware collectives ACM/IEEE Supercomputing Conference 2011, Seattle, WA, November 2011. report bibtex
paper Edgar Solomonik and James Demmel Communication-optimal parallel 2.5D matrix multiplication and LU factorization algorithms Lecture Notes in Computer Science, Euro-Par, Bordeaux, France, August 2011. "Distinguished Paper" report errata bibtex
Laxmikant Kale and Edgar Solomonik Parallel sorting Encyclopedia of Parallel Computing, Springer Verlag, David Padua, Ed., 2011.
paper Edgar Solomonik and Laxmikant V. Kale Highly scalable parallel sorting IEEE International Parallel and Distributed Processing Symposium (IPDPS), Atlanta, GA, April 2010. bibtex
paper Vivek Kale and Edgar Solomonik Parallel sorting pattern Workshop on Parallel Programming Patterns (ParaPLOP), March 2010. bibtex
paper Abhinav Bhatele, Lukasz Wesolowski, Eric Bohm, Edgar Solomonik, and Laxmikant V. Kale Understanding application performance via micro-benchmarks on three large supercomputers: Intrepid, Ranger and Jaguar International Journal of High Performance Computing Applications (IJHPCA), November 2010. bibtex

Talks

slides May 2017; MolSSI Workshop on Core Software Blocks in Quantum Chemistry: Tensors and Integrals; Monterey Bay; CA, USA; An Overview of Cyclops Tensor Framework

slides April 2017; Scientific and Statistcal Computing Seminar; University of Chicago; Chicago IL, USA; Strassen-like algorithms for symmetric tensor contractions

slides March 2017; Theory Seminar; University of Illinois at Urbana-Champaign; Urbana IL, USA; Communication cost models and a few lower and upper bounds

slides February 2017; SIAM CSE; Atlanta GA, USA; A distributed memory library for sparse tensor functions and contractions

slides July 2016; ISTCP; Grand Forks ND, USA; Developing scalable and portable electronic structure methods with Cyclops Tensor Framework

slides April 2016; Charm++ Workshop; University of Illinois at Urbana-Champaign; Urbana IL, USA; A parallel library for multidimensional array computations with runtime tuning

slides April 2016; SIAM PP; Paris, France; Low rank bilinear algorithms for symmetric tensor contractions

slides March 2016; University of California, Davis; Davis CA, USA; Provably efficient algorithms for multilinear algebra

slides March 2016; University of Toronto; Toronto ON, Canada; Algorithms as multilinear tensor equations

slides March 2016; University of Illinois at Urbana-Champaign; Urbana IL, USA; Algorithms as multilinear tensor equations

slides March 2016; Stony Brook University; Stony Brook NY, USA; Algorithms as multilinear tensor equations

slides February 2016; California Institute of Technology; Pasadena CA, USA; Algorithms as multilinear tensor equations

slides February 2016; University of Colorado, Boulder; Boulder CO, USA; Algorithms as multilinear tensor equations

slides February 2016; Georgia Institute of Technology; Atlanta GA, USA; Algorithms as multilinear tensor equations

slides February 2016; Cornell University; Ithaca NY, USA; Provably efficient algorithms as tensor equations

slides January 2016; TU München; Munich, Germany; Algorithms as multilinear tensor equations

slides December 2015; Guest Lecture; Design of Parallel and High-Performance Computing; ETH Zurich, Switzerland; Distributed-memory models and algorithms

slides November 2015; MATHICSE Institute Seminar; EPFL; Lausanne, Switzerland; Provably efficient algorithms for tensor computations

slides October 2015; SIAM LA; Atlanta GA, USA; Minimizing communication in tensor contraction algorithms

slides October 2015; DMML; Berkeley CA, USA; Communication lower bounds for numerical tensor algebra

slides June 2015; QESC; Kobe, Japan; Efficient tensor contraction algorithms for coupled cluster

slides February 2015; Max-Planck-Institute for Chemical Energy Conversion; Mülheim, Germany; Efficient algorithms for tensor contractions in coupled-cluster

slides December 2014; Guest Lecture; Design of Parallel and High-Performance Computing; ETH Zurich, Switzerland; Distributed-memory models and algorithms

slides August 2014; Dissertation talk; Berkeley CA, USA; Provably efficient algorithms for numerical tensor algebra

slides July 2014; DOE CSGF Program Review; Washington DC, USA; Minimizing communication in numerical linear algebra

slides June 2014; SPAA; Prague, Czech Republic; Tradeoffs between synchronization, communication, and computation in parallel linear algebra computations

slides June 2014; ETH Zurich, Switzerland; Algorithms for contraction of tensors over a commutative ring

slides June 2014; Householder Symposium XIX; Spa, Belgium; Contracting symmetric tensors via fully-symmetric intermediates

slides Feb 2014; LBNL, Berkeley, CA; Cyclops Tensor Framework

slides Feb 2014; SIAM PP; Portland, OR; Tradeoffs between synchronization, communication, and work in parallel linear algebra computations

slides Feb 2014; Berkeley, CA; Tradeoffs between synchronization, communication, and work in parallel schedules

slides Feb 2014; Berkeley, CA; Fast tensor contraction algorithms for Coupled Cluster

slides Dec 2013; Laguna Beach, CA; A distributed-memory framework for tensor contractions

slides Dec 2013; Berkeley, USA; A framework for mapping and redistribution of multidimensional distributed arrays

slides Oct 2013; Berkeley, USA; Communication-avoiding algorithms for dense matrix factorizations

slides Oct 2013; Berkeley, CA; Efficient algorithms for symmetric tensor contractions

slides June 2013; ETH Zurich; Switzerland; Communication-avoiding parallel algorithms for dense linear algebra

slides June 2013; Swiss National Supercomputing Center; Lugano, Switzerland; Communication-avoiding parallel numerical algorithms for dense matrices and tensors

slides May 2013; IPDPS; Boston, USA; Cyclops Tensor Framework: reducing communication and eliminating load imbalance in massively parallel contractions

slides May 2013; IPDPS; Boston, USA; Minimizing communication in all-pairs shortest paths

slides March 2013; SIAM CSE; Boston, USA; Scalable numerical algorithms for electronic structure calculations

slides February 2013; Berkeley, USA; Communication-avoiding parallel algorithms for dense linear algebra and tensor computations

slides January 2013; University of Southern California; LA, USA; A parallel tensor framework for Coupled Cluster

slides Sep. 2012; seminar; Lawrence Livermore National Laboratory; Livermore, CA; Scalable numerical algorithms for electronic structure calculations

slides July 2012; University of Tokyo; Tokyo, Japan; 2.5D algorithms for distributed-memory computing

slides July 2012; VECPAR; Kobe, Japan; Matrix multiplication on multidimensional torus networks

slides June 2012; SIAM ALA; Valencia, Spain; 2.5D Algorithms for dense linear algebra

slides Feb. 2012; SIAM PP; Savannah, GA; Topology-aware parallel algorithms for symmetric tensor contractions

slides Nov. 2011; ACM/IEEE Supercomputing; Seattle, WA; Improving communication performance in dense linear algebra via topology-aware collectives

slides Sep. 2011; CS 294 lecture; Berkeley, CA; 2.5D algorithms: from hardware to theory and back

slides Sep. 2011; Bordeaux, France; Communication-optimal parallel 2.5D matrix multiplication and LU factorization algorithms

slides Aug. 2011; seminar; Argonne National Laboratory; Argonne, IL; Reducing communication in dense matrix/tensor computations

slides Apr. 2010; IPDPS; Atlanta, GA; Highly scalable parallel sorting

Posters

poster Jul. 2013; CSGF conference; Arlington, VA; Cyclops Tensor Framework.

poster Jul. 2011; CSGF conference; Arlington, VA; 2.5D algorithms for dense linear algebra

poster Nov. 2009; ACM/IEEE Supercomputing; Portland, OR Performance comparison of Intrepid, Jaguar and Ranger using scientific applications