Fei Xue


Current Position                                                                        

o        Research Assistant Professor

Department of Mathematics

College of Science and Technology

Temple University

Supervisor: Professor Daniel B. Szyld


Contact Information

o        544 Wachman Hall

1805 North Broad Street

Temple University

Philadelphia, PA 19122, USA

o        Phone: (215) 204-7588

Fax: (215) 204-6433

o        Email: fxue _at_ temple.edu


Research Interests

Numerical linear algebra, sparse matrix computations, scientific computing

Curriculum Vitae


Education

o        Ph.D., University of Maryland, College Park, 2009.

o        M.E., Southeast University (China), 2004.

o        B.E., Southeast University (China), 2001.

 

Research Grants

    • National Science Foundation (NSF) Research Grant DMS-1115520, Co-P.I. (P.I.: Daniel B. Szyld), “Eigenvalues problems, Krylov subspace methods, and subspace recycling”, Division of Mathematical Sciences, Program in Computational Mathematics, August 2011--July 2014.

 

Publications

    • Journal papers (submitted and published)

§       Mark Embree, Josef A. Sifuentes, Kirk M. Soodhalter, Daniel B. Szyld, and Fei Xue,

On short-term recurrence Krylov subspace methods for nearly-Hermitian matrices,

Research Report 11-10-10, Department of Mathematics, submitted (preprint).

§       Daniel B. Szyld and Fei Xue,

Local convergence analysis of several inexact Newton-type algorithms for general nonlinear eigenvalue problems,

Submitted manuscript available upon request.

§       Fei Xue and Howard C. Elman,

Fast inexact implicitly restarted Arnoldi method for generalized eigenvalue problems with spectral transformation,

Submitted in revised form to SIAM Journal on Matrix Analysis and Applications  (preprint)

§       Daniel B. Szyld and Fei Xue,

Efficient preconditioned inner solves for inexact Rayleigh quotient iteration and their connections to the single-vector Jacobi-Davidson method (pdf),

SIAM Journal on Matrix Analysis and Applications, Vol. 32 (2011), pp. 993—1018.

§       Fei Xue and Howard C. Elman,

Fast inexact subspace iteration for generalized eigenvalue problems with spectral transformation  (pdf),

Linear Algebra and Its Applications, Vol. 435, No. 3 (2011), pp. 601--622.

§       Fei Xue and Howard C. Elman,

Convergence analysis of iterative solvers in inexact Rayleigh quotient iteration (pdf),

SIAM Journal on Matrix Analysis and Applications, Vol. 31, No. 3 (2009), pp 877--899.

o        In preparation

§       Error bounds, perturbation analysis and subspace projections for invariant pairs of general nonlinear eigenvalue problems, with Daniel B. Szyld.

§       Inexact augmented Krylov subspace methods for sequences of linear systems, with Daniel B. Szyld.

    • Other publications

§       Numerical solution to eigenvalue problems with spectral transformations (Ph.D thesis) (pdf)

Thesis Advisor: Professor Howard C. Elman

Applied Mathematics, Statistics and Scientific computation (AMSC) program, University of Maryland, College Park, August 2009

§       Computing the dynamics of large multi-particle systems using Fast Multipole Method (FMM) with multi-scale time stepping (unpublished manuscript) (pdf)    

AMSC 663/664 Final report, University of Maryland, College Park, May 2006


Teaching

o        Spring 2012: Introduction to Numerical Analysis (Mathematics 5043, graduate level), scheduled

o        Fall 2011: Calculus I (Mathematics 1041, Sections 013 and 019) (webpage)

o        Spring 2011: Calculus III (Mathematics 2043, Section 001) (webpage)

o        Fall 2010: Linear Algebra (Mathematics 2101, Section 001) (webpage)

o        Fall 2009: Calculus III (Mathematics 2043, Section 001) (webpage)


Software

o        MATLAB version of the implicitly restarted Arnoldi (IRA) method (download) and the Jacobi-Davidson (JDQZ) method (download) for non-Hermitian matrices

o        The following work completed in summer 2008 has been put in MATLAB R2009a and later versions

Functionality extension of
eigs to solve generalized eigenvalue problem Av = λBv with Hermitian indefinite or non-Hermitian B;
Improvement of
gmres in memory efficiency and robustness;
Improvement of
pcg, minres, symmlq, bicg and bicgstab in CPU efficiency;
Development of new iterative solvers
tfqmr and bicgstabl