From mgu at math.berkeley.edu Tue Aug 3 00:51:34 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Aug 3 01:00:26 2010 Subject: [BANANA] Special LAPACK seminar on Wednesday, August 4 Message-ID: <201008030751.o737pYF9027748@panda.math.berkeley.edu> Hi everyone, We will have a special LAPACK seminar on Wednesday, August 4, featuring a guest from Copenhagen. See below. Date: August 4 Time: 1:00-2:00PM Room: 740 Evans Hall Speaker: Prof. Rasmus Bro, University of Copenhagen Title: How linear algebra is used for detecting cancer and improving sugar production Abstract: In food, pharma and clinical analysis, increasingly complex data are measured. Extracting the information requires deep insight into the samples, the measurements and the mathematical and statistical aspects. In this talk several examples will be highlighted explaining how linear algebra is made useful in a number of diverse areas in the field of chemometrics. From mgu at math.berkeley.edu Wed Aug 4 11:03:23 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Wed Aug 4 11:12:14 2010 Subject: [BANANA] Reminder: [lapackers] Special LAPACK seminar on Wednesday, August 4 In-Reply-To: <201008030751.o737pYF9027748@panda.math.berkeley.edu> References: <201008030751.o737pYF9027748@panda.math.berkeley.edu> Message-ID: Hi everyone, Special LAPACK seminar in about an hour. Date: August 4 Time: 1:00-2:00PM Room: 740 Evans Hall Speaker: Prof. Rasmus Bro, University of Copenhagen Title: How linear algebra is used for detecting cancer and improving sugar production From saunders at stanford.edu Mon Aug 16 15:02:42 2010 From: saunders at stanford.edu (Michael A. Saunders) Date: Mon Aug 16 15:07:08 2010 Subject: [BANANA] PhD defense (Chris Maes) Message-ID: Student: Christopher Maes, iCME, Stanford University Advisor: Michael Saunders Time: Friday August 20, 2010, from 1:15pm Place: Packard 204 A Regularized Active-set Method for Sparse Convex Quadratic Programming An active-set algorithm is developed for solving convex quadratic programs (QPs). The algorithm employs primal regularization within a bound-constrained augmented Lagrangian method. This leads to a sequence of QP subproblems that are strictly convex and feasible, and whose KKT systems are guaranteed to be nonsingular for any active set. A simplified, single-phase algorithm becomes possible for each QP subproblem. There is no need to control the inertia of the KKT system defining each search direction, and a simple step-length procedure may be used without risk of cycling in the presence of degeneracy. Since all KKT systems are nonsingular, they can be factored with a variety of sparse direct linear solvers. Block-LU updates of the KKT factors allow for active-set changes. The resulting primal and dual regularization permits warm-starting from any given active-set. This is beneficial inside sequential quadratic programming (SQP) methods for nonlinear optimization, such as the SNOPT solver. The method provides a reliable approach to solving sparse generalized least-squares problems. Ordinary least-squares problems with Tikhonov regularization and bounds can be solved as a single QP subproblem. The algorithm is implemented as the QPBLUR solver (MATLAB and Fortran 95 versions) and the Fortran version has been integrated into SNOPT. The performance of QPBLUR is evaluated on a test set of large convex QPs, and on the sequences of QPs arising from SNOPT's SQP method. From mgu at math.berkeley.edu Tue Aug 31 11:00:57 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Aug 31 11:05:09 2010 Subject: [BANANA] LAPACK seminar on Sept. 8, 2010 Message-ID: <201008311800.o7VI0v6w019640@panda.math.berkeley.edu> Math 290, Section 29, CS 298, Section 6, Fall 2010 (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. Welcome back to the Fall Semester. Hope you had a wonderful summer. We kick off our seminar with a talk from our old friend, Joe Grcar. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Sept. 8 Speaker: Joseph Grcar Title The History of "Gaussian" Elimination Abstract: Gaussian elimination is universally known as "the" method for solving simultaneous linear equations. As Leonhard Euler remarked in 1771, it is "the most natural way" of proceeding. The method was invented in China about 2000 years ago, and then it was reinvented in Europe much more recently, so it is surprising that the primary European sources have not been identified until now. It is an interesting story in the history of computing and technology that Carl Friedrich Gauss came to be mistakenly identified as the inventor of Gaussian elimination even though he was not born until 1777. The European development has three phases. First came the "schoolbook" method that began with algebra lessons written by Isaac Newton; what we learn in high school or middle school is still basically Newton's creation. Second were methods that professional hand computers used to solve the normal equations of least squares problems; until comparatively recently the chief societal use for Guassian elimination was to solve normal equations for statistical estimation. Third was the adoption of matrix notation; henceforth the schoolbook lesson and the professional algorithms were understood to be trivially related in that all can be interpreted as computing triangular decompositions. Date: Sept. 15 Speaker: Benjamin Stamm Title: Reduced Order Modeling for Electromagnetic Scattering Problems