Erin Carson

I am a PhD candidate in Computer Science at U.C. Berkeley, advised by James Demmel and Armando Fox.

I am affiliated with the Berkeley Benchmarking and Optimization Group (BeBOp) within the ASPIRE Lab

I am also a student in the Designated Emphasis in Computational Science and Engineering Program.

My research sits at the intersection of high performance computing, parallel algorithms, scientific computing, and numerical linear algebra.



Office

586E Soda Hall
Computer Science Division
University of California at Berkeley
Berkeley, CA 94720-1776
Email: ecc2z@cs.berkeley.edu

Publications

My Google Scholar profile

Journal Papers

  • G. Ballard, E. Carson, J. Demmel, M. Hoemmen, N. Knight, and O.Schwartz, Communication lower bounds and optimal algorithms for numerical linear algebra, Acta Numerica, 23 (2014), pp. 1-155.
  • N. Knight, E. Carson and J. Demmel. Exploiting data sparsity in parallel matrix powers computations, in Parallel Processing and Applied Mathematics, R. Wyrzykowski, J. Dongarra, K. Karczewski, and J. Waniewski, eds., Lecture Notes in Computer Science, Springer Berlin Heidelberg, 2014, pp.15-25.
  • E. Carson and J. Demmel. A residual replacement strategy for improving the maximum attainable accuracy of s-step Krylov subspace methods. SIAM J. Matrix Anal. Appl. 35(1), 2014.
  • E. Carson, N. Knight, and J. Demmel. Avoiding communication in two-sided Krylov subspace methods. SIAM J. Sci. Comp. 35 (5), 2013.

Conference Papers

  • E. Solomonik, E. Carson, N. Knight, and J. Demmel. Tradeoffs between synchronization, communication, and work in parallel linear algebra computations. Proceedings of the 26th ACM Symposium on Parallelism in Algorithms and Architectures (SPAA), June 2014 (to appear).
  • S. Williams, E. Carson, M. Lijewski, N. Knight, A. Almgren, B. Van Straalen, and J. Demmel. s-step Krylov Subspace Methods as Bottom Solvers for Geometric Multigrid. Proceedings of the 28th IEEE International Parallel and Distributed Processing Symposium (to appear).

Technical Reports

  • E. Carson and J. Demmel. Error analysis of the s-step Lanczos method in finite precision. Technical Report UCB/EECS-2014-55, EECS Dept., U.C. Berkeley, May 2014. [pdf]
  • E. Carson and J. Demmel. Analysis of the finite precision s-step biconjugate gradient method. Technical Report UCB/EECS-2014-18, EECS Dept., U.C. Berkeley, Mar 2014. [pdf]
  • E. Solomonik, E. Carson, N. Knight, and J. Demmel. Tradeoffs between synchronization, communication, and work in parallel linear algebra computations. Technical Report UCB/EECS-2014-8, EECS Dept., U.C. Berkeley, Jan 2014. [pdf]
  • E. Carson, N. Knight, and J. Demmel. Avoiding communication in two-sided Krylov subspace methods. Technical Report UCB/EECS-2011-93, EECS Dept., U.C. Berkeley, Aug 2011. [pdf]
  • E. Carson and J. Demmel. A residual replacement strategy for improving the maximum attainable accuracy of s-step Krylov subspace methods. Technical Report UCB/EECS-2012-197, EECS Dept., U.C. Berkeley, Sept. 2012. [pdf]

Talks and Extended Abstracts

  • S. Williams, E. Carson, N. Knight, M. Lijewski, A. Almgren, B. van Straalen and J. Demmel. Avoiding synchronization in geometric multigrid. SIAM Parallel Processing for Scientific Computing, Feb 2014. [abstract]
  • E. Carson. Communication-Avoiding Krylov Subspace Methods in Finite Precision, Bay Area Scientific Computing Day, Dec 2013. [abstract]
  • E. Carson and J. Demmel. Efficient deflation for communication avoiding Krylov methods (extended abstract). In Proc. Numerical Analysis and Scientific Computation with Applications, Jun 2013.
  • N. Knight, E. Carson, and J. Demmel. Avoiding communication with hierarchical matrices (abstract). In Proc. SIAM Conference on Applied Linear Algebra, Jun 2012.
  • E. Carson, N. Knight, and J. Demmel. Improving the stability of communication-avoiding Krylov subspace methods (abstract). In Proc. SIAM Conference on Applied Linear Algebra, Jun 2012.
  • E. Carson, N. Knight, and J. Demmel. Exploiting low-rank structure in computing matrix powers with applications to preconditioning (abstract). In Proc. SIAM Conference on Parallel Processing for Scientific Computing, Feb 2012. [ pdf | pptx ]
  • E. Carson and J. Demmel. A residual replacement strategy for improving the maximum attainable accuracy of communication-avoiding Krylov subspace methods (extended abstract). In Proc. 9th International Workshop on Accurate Solution of Eigenvalue Problems, pages 1921, Jun 2012.
  • E. Carson, N. Knight, and J. Demmel. Hypergraph partitioning for computing matrix powers (extended abstract). In Proc. Fifth SIAM Workshop on Comb. Sci. Comput., pages 3133, May 2011. [pdf]

Past Projects


Teaching

U.C. Berkeley

  • Math 54: Linear Algebra and Differential Equations, Spring 2011. Instructor: Constantin Teleman. Topics: Basic linear algebra; matrix arithmetic and determinants. Vector spaces; inner product as spaces. Eigenvalues and eigenvectors; linear transformations. Homogeneous ordinary differential equations; first-order differential equations with constant coefficients. Fourier series and partial differential equations.

University of Virginia


Activities

Since 2009, I've served as a Feature Editor for ACM's XRDS Magazine.