From saunders at stanford.edu Tue Feb 1 00:10:50 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue Feb 1 00:12:47 2011 Subject: [BANANA] LA/Opt seminar Thursday Feb 3 (Rob Tibshirani) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Feb 3, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Prof Rob Tibshirani Depts of Health Research and Policy, and Statistics, Stanford University tibs@stanford.edu http://www-stat.stanford.edu/~tibs Strong rules for discarding predictors in lasso-type problems We propose rules for discarding predictors in lasso regression and related problems, for computational efficiency. El Ghaoui et al. (2010) propose ``SAFE'' rules that guarantee that a coefficient will be zero in the solution, based on the inner products of each predictor with the outcome. In this talk I discuss ``strong rules'' that are not foolproof but rarely fail in practice. These can be complemented with simple checks of the Karush-Kuhn-Tucker (KKT) conditions to ensure that the exact solution to the convex problem is delivered. These rules offer substantial speed and space savings in a variety of statistical convex optimization problems. This is joint work with Jacob Bien, Jerome Friedman, Trevor Hastie, Noah Simon, Jonathan Taylor and Ryan Tibshirani Forthcoming: Thu Feb 10 Zaiwen Wen IPAM, UCLA Thu Feb 17 Abdo Alfakih Math, Univ of Windsor Thu Feb 24 David Gleich Sandia, Livermore Thu Mar 10 Jim Lambers Math, Southern Mississippi From mgu at math.berkeley.edu Tue Feb 1 06:37:20 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Feb 1 07:13:05 2011 Subject: [BANANA] Reminder: LAPACK seminar on Feb. 2, 2011 In-Reply-To: <201101280009.p0S09IeT026806@panda.math.berkeley.edu> References: <201101280009.p0S09IeT026806@panda.math.berkeley.edu> Message-ID: Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 2 Speaker:Dr. David Gleich Title: Fast Katz and Commuters - Quadrature Rules and Sparse Linear Solvers for Link Prediction Heuristics Date: Feb. 9 Speaker: Jim Demmel From saunders at stanford.edu Thu Feb 3 10:29:47 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Feb 3 10:33:56 2011 Subject: [BANANA] LA/Opt seminar Thursday Feb 3 (Rob Tibshirani) In-Reply-To: <25187_1296547964_p118CfMV023759_AANLkTi=2NdTkMtKJCs0wavhNUq3DqsCANRigpo-o2qaF@mail.gmail.com> References: <25187_1296547964_p118CfMV023759_AANLkTi=2NdTkMtKJCs0wavhNUq3DqsCANRigpo-o2qaF@mail.gmail.com> Message-ID: Reminder: seminar this afternoon ?Linear Algebra and Optimization Seminar (CME 510) ?iCME, Stanford University ?http://icme.stanford.edu/seminars/seminars.php ?4:15pm Thursday Feb 3, 2011 ?Y2E2 101 ?http://campus-map.stanford.edu/index.cfm?ID=04-070 ?Prof Rob Tibshirani ?Depts of Health Research and Policy, and Statistics, Stanford University ?tibs@stanford.edu ?http://www-stat.stanford.edu/~tibs Strong rules for discarding predictors in lasso-type problems We propose rules for discarding predictors in lasso regression and related problems, for computational efficiency. ?El Ghaoui et al. (2010) propose ``SAFE'' rules that guarantee that a coefficient will be zero in the solution, based on the inner products of each predictor with the outcome. ?In this talk I discuss ``strong rules'' that are not foolproof but rarely fail in practice. These can be complemented with simple checks of the Karush-Kuhn-Tucker (KKT) conditions to ensure that the exact solution to the convex problem is delivered. ?These rules offer substantial speed and space savings in a variety of statistical convex optimization problems. This is joint work with Jacob Bien, Jerome Friedman, Trevor Hastie, Noah Simon, Jonathan Taylor and Ryan Tibshirani Forthcoming: Thu Feb 10 ?Zaiwen Wen ? ? ?IPAM, UCLA Thu Feb 17 ?Abdo Alfakih ? ?Math, Univ of Windsor Thu Feb 24 ?David Gleich ? ?Sandia, Livermore Thu Mar 10 ?Jim Lambers ? ? Math, Southern Mississippi From mgu at math.berkeley.edu Fri Feb 4 18:07:21 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Fri Feb 4 18:40:44 2011 Subject: [BANANA] LAPACK seminar on Feb. 9, 2011 In-Reply-To: References: <201101280009.p0S09IeT026806@panda.math.berkeley.edu> Message-ID: Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 9 Speaker: Jim Demmel, UC Berkeley Title: Recent Progress in Communication Avoiding Algorithms Date: Feb. 16 Speaker: Dr. Chao Yang, LBL Title: Solving Large-scale Eigenvalue Problems in Nuclei Structure Calculation From mgu at math.berkeley.edu Mon Feb 7 23:04:05 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon Feb 7 23:08:37 2011 Subject: [BANANA] Reminder: LAPACK seminar on Feb. 9, 2011 In-Reply-To: References: <201101280009.p0S09IeT026806@panda.math.berkeley.edu> Message-ID: Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 9 Speaker: Jim Demmel, UC Berkeley Title: Recent Progress in Communication Avoiding Algorithms Date: Feb. 16 Speaker: Dr. Chao Yang, LBL Title: Solving Large-scale Eigenvalue Problems in Nuclei Structure Calculation From saunders at stanford.edu Tue Feb 8 16:08:38 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue Feb 8 16:10:36 2011 Subject: [BANANA] LA/Opt seminar Thursday Feb 10 (Zaiwen Wen) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Feb 10, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Zaiwen Wen NSF Math Institutes' Postdoc, IPAM, UCLA wendouble@gmail.com http://math.sjtu.edu.cn/faculty/zw2109 First-order methods for semidefinite programming We present two first-order methods for solving Semidefinite Programming problems (SDP). The alternating direction augmented Lagrangian method minimizes the dual augmented Lagrangian function with respect to each block of dual variables separately while other blocks are fixed, and then updates the primal variables at each iteration. A feasible approach is designed for SDPs with only constraints X_ii = 1 based on a constraint-preserving updating scheme. These techniques enable a few types of SDPs to be solved efficiently. Forthcoming: Thu Feb 24 David Gleich Thu Mar 10 Jim Lambers From saunders at stanford.edu Thu Feb 10 11:22:11 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Feb 10 11:24:08 2011 Subject: [BANANA] LA/Opt seminar TODAY (Zaiwen Wen) Message-ID: Reminder: 4:15 this afternoon Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Feb 10, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Zaiwen Wen NSF Math Institutes' Postdoc, IPAM, UCLA wendouble@gmail.com http://math.sjtu.edu.cn/faculty/zw2109 First-order methods for semidefinite programming We present two first-order methods for solving Semidefinite Programming problems (SDP). The alternating direction augmented Lagrangian method minimizes the dual augmented Lagrangian function with respect to each block of dual variables separately while other blocks are fixed, and then updates the primal variables at each iteration. A feasible approach is designed for SDPs with only constraints X_ii = 1 based on a constraint-preserving updating scheme. These techniques enable a few types of SDPs to be solved efficiently. Forthcoming: Thu Feb 17 No seminar Thu Feb 24 David Gleich Thu Mar 10 Jim Lambers From mgu at math.berkeley.edu Thu Feb 10 15:13:39 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Thu Feb 10 15:47:09 2011 Subject: [BANANA] LAPACK seminar on Feb. 16, 2011 Message-ID: <201102102313.p1ANDdcf016158@phoenix.math.berkeley.edu> Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 16, 2011 Speaker: Chao Yang, Computational Research Division, LBL Title: Solving Large-scale Eigenvalue Problems in Nuclei Structure Calculation Abstract: One of the emerging computational approaches in nuclear physics is the configuration interaction (CI) method for solving the nuclear many-body problem. Like other quantum mechanics calculations, the basic computational problem is an eigenvalue problem. In many cases, one is interested in obtaining a few smallest eigenvalues and the corresponding eigenfunctions of a many-body Hamiltonian. In some applications, one may be interested in computing a relatively large number of small eigenvalues with a prescribed total angular momentum J or certain properties of a nucleus pertaining to a fixed J value. This type of calculation can be done by a simultaneous diagonalization of the Hamiltonian and the total angular momentum square operator. Although eigenvalue calculation is a well studied subject in numerical linear algebra, solving large-scale eigenvalue problems on high performance computers consisting of many thousands processing units is challenging. I will describe a number of techniques for achieving good performance in nuclear CI calculations. Date: Feb. 23, 2011 Speaker: Lior Pachter, UC Berkeley From mgu at math.berkeley.edu Tue Feb 15 08:19:00 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Feb 15 08:23:14 2011 Subject: [BANANA] Reminder: LAPACK seminar on Feb. 16, 2011 In-Reply-To: <201102102313.p1ANDdcf016158@phoenix.math.berkeley.edu> References: <201102102313.p1ANDdcf016158@phoenix.math.berkeley.edu> Message-ID: Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 16, 2011 Speaker: Chao Yang, Computational Research Division, LBL Title: Solving Large-scale Eigenvalue Problems in Nuclei Structure Calculation Date: Feb. 23, 2011 Speaker: Lior Pachter, UC Berkeley From saunders at stanford.edu Tue Feb 15 16:47:17 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue Feb 15 16:49:06 2011 Subject: [BANANA] LA/Opt seminar Thursday Feb 17 (Abdo Alfakih) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Feb 17, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Prof Abdo Y. Alfakih Dept of Mathematics and Statistics, University of Windsor, Ontario alfakih@uwindsor.ca http://apps.uwindsor.ca/uwincpb/jsp/DirectoryServicesProfile.jsp?q=alfakih (Visiting scholar, MS&E, on sabbatical leave from University of Windsor) On Global and Universal Rigidity of Bar Frameworks A bar framework G(p) in r-dimensional Euclidean space is a graph G whose vertices are points p^1, ..., p^n in R^r and whose edges are line segments connecting pairs of these points. A given framework G(p) in R^r is globally rigid if there does not exist a non-congruent framework G(q) in the same r-dimensional space with the same edge lengths as those of G(p). Furthermore, if there does not exist a non-congruent framework G(q) in any Euclidean space with the same edge lengths as those of G(p), then G(p) is said to be universally rigid. Global and Universal rigidity of bar frameworks have many important applications in molecular conformation, multidimensional scaling, and wireless sensor networks. In this talk, I'll present some recent results concerning global and universal rigidity of bar frameworks, emphasizing those results where Semidefinite Programming and Linear Algebra play a key role. Forthcoming: Thu Feb 24 David Gleich Sandia Livermore Thu Mar 03 Thu Mar 10 Jim Lambers Math, Southern Mississippi From saunders at stanford.edu Thu Feb 17 12:36:20 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Feb 17 12:38:53 2011 Subject: [BANANA] LA/Opt seminar Thursday Feb 17 (Abdo Alfakih) Message-ID: Reminder: LA/Opt seminar this afternoon: ?Linear Algebra and Optimization Seminar (CME 510) ?iCME, Stanford University ?http://icme.stanford.edu/seminars/seminars.php ?4:15pm Thursday Feb 17, 2011 ?Y2E2 101 ?http://campus-map.stanford.edu/index.cfm?ID=04-070 ?Prof Abdo Y. Alfakih ?Dept of Mathematics and Statistics, University of Windsor, Ontario ?alfakih@uwindsor.ca ?http://apps.uwindsor.ca/uwincpb/jsp/DirectoryServicesProfile.jsp?q=alfakih ?(Visiting scholar, MS&E, on sabbatical leave from University of Windsor) ?On Global and Universal Rigidity of Bar Frameworks A bar framework G(p) in r-dimensional Euclidean space is a graph G whose vertices are points p^1, ..., p^n in R^r and whose edges are line segments connecting pairs of these points. ?A given framework G(p) in R^r is globally rigid if there does not exist a non-congruent framework G(q) in the same r-dimensional space with the same edge lengths as those of G(p). ?Furthermore, if there does not exist a non-congruent framework G(q) in any Euclidean space with the same edge lengths as those of G(p), then G(p) is said to be universally rigid. Global and Universal rigidity of bar frameworks have many important applications in molecular conformation, multidimensional scaling, and wireless sensor networks. ?In this talk, I'll present some recent results concerning global and universal rigidity of bar frameworks, emphasizing those results where Semidefinite Programming and Linear Algebra play a key role. Forthcoming: Thu Feb 24 ?David Gleich ? ?Sandia Livermore Thu Mar 03 Thu Mar 10 ?Jim Lambers ? ? Math, Southern Mississippi From mgu at math.berkeley.edu Fri Feb 18 01:08:46 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Fri Feb 18 01:42:14 2011 Subject: [BANANA] LAPACK seminar on Feb. 23, 2011 Message-ID: <201102180908.p1I98kOb007119@phoenix.math.berkeley.edu> Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 23, 2011 Speaker: Prof. Lior Pachter Title: Scientific computing challenges in the analysis of high-throughput sequencing-based experiments Abstract: I will provide an introduction to recently developed 'sequence-census' experiments in molecular biology, and describe the scientific computing challenges associated with interpreting the resulting data. I'll focus on two recently developed protocols, one for measuring RNA structure and the second for measuring RNA abundance, and will discuss the large-scale convex optimization problems they motivate, and how to solve them. The work I will discuss is joint with Adam Arkin, Sharon Aviran, Julius Lucks, Adam Roberts and Cole Trapnell. From saunders at stanford.edu Tue Feb 22 09:09:47 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue Feb 22 09:11:38 2011 Subject: [BANANA] LA/Opt seminar Thurs Feb 24 (David Gleich) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Feb 24, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr David Gleich John von Neumann Research Fellow Sandia National Laboratory, Livermore dgleich@stanford.edu http://www.stanford.edu/~dgleich Spectra of Large Networks We present insights from analyzing the eigenvalues of the adjacency, normalized Laplacian, unnormalized Laplacian, and modularity matrices of a range of real-world graph and network models. This includes finding complete spectra for graphs with hundreds of thousands of nodes. In particular, we more deeply explore the origin of the null-space of the adjacency matrix, which is equivalent to the eigenvector of the normalized Laplacian with eigenvalue one. This talk presents work in progress. Forthcoming: Thu Mar 03 No seminar Thu Mar 10 Jim Lambers Math, Southern Mississippi From mgu at math.berkeley.edu Tue Feb 22 10:41:12 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Feb 22 10:45:27 2011 Subject: [BANANA] Reminder: LAPACK seminar on Feb. 23, 2011 In-Reply-To: <11921_1298022130_p1I9g6TK009503_201102180908.p1I98kOb007119@phoenix.math.berkeley.edu> References: <11921_1298022130_p1I9g6TK009503_201102180908.p1I98kOb007119@phoenix.math.berkeley.edu> Message-ID: Math 290, Section 29, CS 298, Section 6, Spring 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 11:10 - noon in Room 380 Soda Hall, Berkeley campus. The coordinators are Profs. J. Demmel (demmel@cs.berkeley.edu) and M. Gu (mgu@math.berkeley.edu). The program will be a mixture of research talks and tutorials. The tutorials will provide a partial sequel to Math 221. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Feb. 23, 2011 Speaker: Prof. Lior Pachter Title: Scientific computing challenges in the analysis of high-throughput sequencing-based experiments From saunders at stanford.edu Thu Feb 24 13:25:58 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Feb 24 13:27:56 2011 Subject: [BANANA] LA/Opt seminar Thurs Feb 24 (David Gleich) Message-ID: Reminder: LA/Opt seminar this afternoon ?Linear Algebra and Optimization Seminar (CME 510) ?iCME, Stanford University ?http://icme.stanford.edu/seminars/seminars.php ?4:15pm Thursday Feb 24, 2011 ?Y2E2 101 ?http://campus-map.stanford.edu/index.cfm?ID=04-070 ?Dr David Gleich ?John von Neumann Research Fellow ?Sandia National Laboratory, Livermore ?dgleich@stanford.edu ?http://www.stanford.edu/~dgleich ?Spectra of Large Networks We present insights from analyzing the eigenvalues of the adjacency, normalized Laplacian, unnormalized Laplacian, and modularity matrices of a range of real-world graph and network models. ?This includes finding complete spectra for graphs with hundreds of thousands of nodes. ?In particular, we more deeply explore the origin of the null-space of the adjacency matrix, which is equivalent to the eigenvector of the normalized Laplacian with eigenvalue one. This talk presents work in progress. Forthcoming: Thu Mar 03 ?No seminar Thu Mar 10 ?Jim Lambers ? ? Math, Southern Mississippi From mgu at math.berkeley.edu Fri Feb 25 01:25:18 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Fri Feb 25 01:29:35 2011 Subject: [BANANA] No seminar on March 2, 2011 In-Reply-To: <201102180908.p1I98kOb007119@phoenix.math.berkeley.edu> References: <201102180908.p1I98kOb007119@phoenix.math.berkeley.edu> Message-ID: Due to the SIAM CSE meeting in Reno next week, there will be no seminar on March 2, 2011. We resume with a talk by Prof. Per-Olaf Persson of UC Berkeley on March 9, 2011. All the best, Ming