From mgu at math.berkeley.edu Mon Oct 4 21:00:57 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon Oct 4 21:05:13 2010 Subject: Reminder: [BANANA] LAPACK seminar on Oct. 6, 2010 In-Reply-To: <19609_1285828824_o8U6eFZZ023053_201009300635.o8U6ZV6Y029010@panda.math.berkeley.edu> References: <19609_1285828824_o8U6eFZZ023053_201009300635.o8U6ZV6Y029010@panda.math.berkeley.edu> Message-ID: 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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 6, 2010 Speaker: Prof. Jim Demmel, UC Berkeley Title: Recent Progress in Communication Avoiding Algorithm for Linear Algebra Date: Oct. 13, 2010 Speaker: Erich Strohmaier, Lawrence Berkeley Lab From saunders at stanford.edu Tue Oct 5 17:39:19 2010 From: saunders at stanford.edu (Michael A. Saunders) Date: Tue Oct 5 17:39:33 2010 Subject: [BANANA] OR Seminar Wed October 6 -- Erling D. Andersen Message-ID: ---------- Forwarded message ---------- Date: Tue, 5 Oct 2010 12:05:27 -0700 From: Pranav Dandekar To: or-seminars@lists.stanford.edu Subject: Reminder: OR Seminars -- Wednesday (October 6) -- Erling D. Andersen Please join us tomorrow for the next OR seminar talk by Dr. Erling Andersen: *Ten years of experience with conic quadratic optimization* Erling D. Andersen MOSEK APS Denmark Wednesday, October 6, 2010 4:30 - 5:30 pm Y2E2 101 (*please note the new seminar location*) Abstract: For about 10 years the software package MOSEK has been capable of solving large-scale sparse conic quadratic optimization (CQO) problems. Based on the experience gained with CQO during those 10 years we will present a few important applications of conic quadratic optimization, discuss properties of the CQO problem and give an overview of how MOSEK solves a CQO problem. We will also present numerical results demonstrating the performance of MOSEK on CQO problems and discuss future developments. --- The OR seminar series gratefully acknowledges support from Yahoo! Inc and the Dantzig family. From mgu at math.berkeley.edu Mon Oct 11 11:18:31 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon Oct 11 11:51:51 2010 Subject: [BANANA] LAPACK seminar on Oct. 13, 2010 Message-ID: <201010111818.o9BIIVW0011062@phoenix.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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 13, 2010 Speaker: Erich Strohmaier, LBNL Title:Using Numerical Algorithms for Computer Science Research Abstract: A variety of numerical algorithms have been used during the last decades for research in different aspects of computer science. Many have been implemented as benchmarks for measuring various aspects of computer architectures. In recent years research in parallel programming languages and multi-core processor architectures have renewed the interest in using broader set of numerical methods for evaluation. This talk will review some of the previous usages of numerical kernels as benchmarks in computer science and present work on assembling a Testbed of numerical Kernels for research in different aspects of computer science. Date: Oct. 20, 2010 Speaker: Ming Gu Title: Reduced Rank Regression via Convex Optimization From saunders at stanford.edu Mon Oct 11 12:06:09 2010 From: saunders at stanford.edu (Michael A. Saunders) Date: Mon Oct 11 12:05:48 2010 Subject: [BANANA] LA/Opt seminar Thursday (David Gleich) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminar.php?seminar_id=2¤t=true 4:15pm Thursday Oct 14, 2010 Y2E2 101 (NOTE NEW ROOM) http://campus-map.stanford.edu/index.cfm?ID=04-070 See small square part of Y2E2 across the road from Cypress Hall. Parking Structure 2 has free parking after 4pm. Dr David Gleich John von Neumann Research Fellow Sandia National Laboratory, Livermore dgleich@stanford.edu http://www.stanford.edu/~dgleich Fast Katz and Commuters - Quadrature Rules and Sparse Linear Solvers for Link Prediction Heuristics Motivated by social network data mining problems such as link prediction and collaborative filtering, significant research effort has been devoted to computing topological measures including the Katz score and the commute time. Existing approaches approximate all pairwise relationships simultaneously. We are interested in computing the score for a single pair of nodes; the top-k nodes with the best scores from a given source node. For the pairwise problem, we introduce an iterative algorithm that computes upper and lower bounds for the measures we seek. This algorithm exploits a relationship between the Lanczos process and a quadrature rule. For the top-k problem, we propose an algorithm that only accesses a small portion of the graph, similar to algorithms used in personalized PageRank computing. To test scalability and accuracy, we experiment with three real-world networks and find that our algorithms run in milliseconds to seconds without any preprocessing. ---------------------------------------------------------------------- Forthcoming LA/Opt seminars: Thu 21 Oct Alkis Vazacopoulos, FICO Thu 28 Oct Kevin Carlberg, Stanford Thu 04 Nov Lindsay Erickson, Sandia Thu 18 Nov Todd Plantenga, Sandia From mgu at math.berkeley.edu Tue Oct 12 16:49:35 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Oct 12 16:53:47 2010 Subject: Reminder: [BANANA] LAPACK seminar on Oct. 13, 2010 In-Reply-To: <19498_1286823112_o9BIploh028378_201010111818.o9BIIVW0011062@phoenix.math.berkeley.edu> References: <19498_1286823112_o9BIploh028378_201010111818.o9BIIVW0011062@phoenix.math.berkeley.edu> Message-ID: 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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 13, 2010 Speaker: Erich Strohmaier, LBNL Title:Using Numerical Algorithms for Computer Science Research Date: Oct. 20, 2010 Speaker: Ming Gu Title: Reduced Rank Regression via Convex Optimization From saunders at stanford.edu Thu Oct 14 10:34:46 2010 From: saunders at stanford.edu (Michael A. Saunders) Date: Thu Oct 14 10:34:10 2010 Subject: [BANANA] LA/Opt seminar TODAY (David Gleich) Message-ID: Reminder: seminar this afternoon Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminar.php?seminar_id=2¤t=true 4:15pm Thursday Oct 14, 2010 Y2E2 101 (NOTE NEW ROOM) 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 Fast Katz and Commuters - Quadrature Rules and Sparse Linear Solvers for Link Prediction Heuristics Motivated by social network data mining problems such as link prediction and collaborative filtering, significant research effort has been devoted to computing topological measures including the Katz score and the commute time. Existing approaches approximate all pairwise relationships simultaneously. We are interested in computing the score for a single pair of nodes; the top-k nodes with the best scores from a given source node. For the pairwise problem, we introduce an iterative algorithm that computes upper and lower bounds for the measures we seek. This algorithm exploits a relationship between the Lanczos process and a quadrature rule. For the top-k problem, we propose an algorithm that only accesses a small portion of the graph, similar to algorithms used in personalized PageRank computing. To test scalability and accuracy, we experiment with three real-world networks and find that our algorithms run in milliseconds to seconds without any preprocessing. ---------------------------------------------------------------------- Forthcoming LA/Opt seminars: Thu 21 Oct Alkis Vazacopoulos, FICO Thu 28 Oct Kevin Carlberg, Stanford Thu 04 Nov Lindsay Erickson, Sandia Thu 11 Nov David Fong, iCME Thu 18 Nov Todd Plantenga, Sandia From mgu at math.berkeley.edu Thu Oct 14 23:49:17 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Thu Oct 14 23:53:36 2010 Subject: [BANANA] LAPACK seminar on Oct. 20, 2010 Message-ID: <201010150649.o9F6nHcf026654@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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 20, 2010 Speaker: Ming Gu, UC Berkeley Title: Reduced Rank Regression via Convex Optimization Abstact: Reduced rank regression is a well-known technique for dimension reduction and coefficient estimation for multivariate linear regression. In this talk, we discuss the formulation of various reduced rank regression models, using convex optimization techniques; we also develop a general solution technique, based on spectral projection, to efficiently solve such problems. We present numerical experiments that demonstrate the effectiveness of our method. Date: Oct. 27, 2010 Speaker: Erin Carson, UC Berkeley From mgu at math.berkeley.edu Mon Oct 18 13:54:03 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon Oct 18 14:27:42 2010 Subject: [BANANA] Reminder: LAPACK seminar on Oct. 20, 2010 In-Reply-To: <201010150649.o9F6nHcf026654@panda.math.berkeley.edu> References: <201010150649.o9F6nHcf026654@panda.math.berkeley.edu> Message-ID: 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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 20, 2010 Speaker: Ming Gu, UC Berkeley Title: Reduced Rank Regression via Convex Optimization Date: Oct. 27, 2010 Speaker: Erin Carson, UC Berkeley Title: Recent Work in Communication-Avoiding Krylov Subspace Methods for solving Linear Systems From saunders at stanford.edu Mon Oct 18 15:57:12 2010 From: saunders at stanford.edu (Michael A. Saunders) Date: Mon Oct 18 15:56:17 2010 Subject: [BANANA] LA/Opt seminar Thursday (Alkis Vazacopoulos) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminar.php?seminar_id=2¤t=true 4:15pm Thursday Oct 21, 2010 Y2E2 101 (NOTE NEW ROOM) http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Alkis Vazacopoulos Vice-President Research, FICO AlkisVazacopoulos@fico.com http://www.linkedin.com/in/vazacopoulos USING MIXED INTEGER PROGRAMMING TO SOLVE SEQUENCING, SCHEDULING, AND PACKING PROBLEMS Recent advancements in Mixed Integer Programming solvers give us the ability to solve larger and more complex sequencing, scheduling, and packing problems. We will demonstrate this fact by showing examples from tournament scheduling, space retail optimization, production scheduling, and sequencing in energy applications. Thu 28 Oct Kevin Carlberg, Aero-Astro, Stanford Thu 04 Nov Lindsay Erickson, Sandia Thu 11 Nov David Fong, iCME Thu 18 Nov Todd Plantenga, Sandia From saunders at stanford.edu Thu Oct 21 10:28:47 2010 From: saunders at stanford.edu (Michael Saunders) Date: Thu Oct 21 10:30:38 2010 Subject: [BANANA] LA/Opt seminar TODAY (Alkis Vazacopoulos) Message-ID: Reminder: seminar this afternoon: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminar.php?seminar_id=2¤t=true 4:15pm Thursday Oct 21, 2010 Y2E2 101 (NOTE NEW ROOM) http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Alkis Vazacopoulos Vice-President Research, FICO AlkisVazacopoulos@fico.com http://www.linkedin.com/in/vazacopoulos USING MIXED INTEGER PROGRAMMING TO SOLVE SEQUENCING, SCHEDULING, AND PACKING PROBLEMS Recent advancements in Mixed Integer Programming solvers give us the ability to solve larger and more complex sequencing, scheduling, and packing problems. We will demonstrate this fact by showing examples from tournament scheduling, space retail optimization, production scheduling, and sequencing in energy applications. Thu 28 Oct Kevin Carlberg, Aero-Astro, Stanford Thu 04 Nov Lindsay Erickson, Sandia Thu 11 Nov David Fong, iCME Thu 18 Nov Todd Plantenga, Sandia -------------- next part -------------- An HTML attachment was scrubbed... URL: http://csmr.ca.sandia.gov/pipermail/banana/attachments/20101021/c69c44df/attachment.html From mgu at math.berkeley.edu Fri Oct 22 10:57:34 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Fri Oct 22 11:31:03 2010 Subject: [BANANA] LAPACK seminar on Oct. 27, 2010 Message-ID: <201010221757.o9MHvYJt027318@phoenix.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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 27, 2010 Speaker: Erin Carson, UC Berkeley Title: Recent Work in Communication-Avoiding Krylov Subspace Methods for Solving Linear Systems Abstract: The performance of standard Krylov Subspace Methods (KSMs) for solving linear systems is bound by communication, with one or more SpMV operations required per iteration. In this talk, we motivate the need to avoid communication, and demonstrate how this can be achieved in KSMs using communication-avoiding kernels. We discuss the derivation, implementation, and convergence properties of our communication-avoiding versions of CG, GMRES, and BiCG. Preliminary results for preconditioning and the implementation of CGS and BiCG-Stab are presented. Date: Nov. 3, 2010 Speaker: Alberto Grunbaum, UC Berkeley Title: Quantum walks and linear algebra From mgu at math.berkeley.edu Mon Oct 25 15:10:50 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon Oct 25 15:44:24 2010 Subject: [BANANA] Reminder: LAPACK seminar on Oct. 27, 2010 In-Reply-To: <18850_1287772261_o9MIUvB8026241_201010221757.o9MHvYJt027318@phoenix.math.berkeley.edu> References: <18850_1287772261_o9MIUvB8026241_201010221757.o9MHvYJt027318@phoenix.math.berkeley.edu> Message-ID: 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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Oct. 27, 2010 Speaker: Erin Carson, UC Berkeley Title: Recent Work in Communication-Avoiding Krylov Subspace Methods for Solving Linear Systems Date: Nov. 3, 2010 Speaker: Alberto Grunbaum, UC Berkeley Title: Quantum walks and linear algebra From saunders at stanford.edu Tue Oct 26 12:22:42 2010 From: saunders at stanford.edu (Michael Saunders) Date: Tue Oct 26 12:29:22 2010 Subject: [BANANA] LA/Opt seminar Thursday (Kevin Carlberg) Message-ID: Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminar.php?seminar_id=2¤t=true 4:15pm Thursday Oct 28, 2010 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Kevin Carlberg PhD Candidate, Aeronautics & Astronautics, Stanford University http://stanford.edu/~carlberg Model reduction-based iterative methods for real-time simulation and repeated analyses of mathematical models Despite the development of efficient solution algorithms, parallel computer architectures, and fast processors, the computational cost of analyzing large-scale, high-fidelity mathematical models remains a significant barrier in many engineering applications. One example is real-time simulation, which arises in such areas as nonlinear model predictive control, fast turnaround design, and "in field" analysis. For general nonlinear problems, standard methods (e.g. mesh coarsening, POD-Galerkin model reduction) often fail to meet solution time requirements without introducing unacceptable errors. Another example is the repeated analyses context, which occurs in nonlinear analysis, design optimization, and parameter space sampling, for example. These problems typically require solving a sequence of "nearby" linear systems of equations to specified tolerances; while existing augmented Krylov-subspace iterative methods are able to deliver the required accuracy, they are only efficient when the number of linear systems remains modest or the matrices have spectra of special form. This talk describes two iterative methods that aim to meet the demands of real-time simulation and repeated analyses problems. The first method is a system-approximated Gauss-Newton method, which enables near real-time nonlinear analysis. To achieve high levels of accuracy in the "online" solution, the method searches for solutions in a low-dimensional proper orthogonal decomposition (POD) subspace that is computed a priori during an "offline" data collection stage. The method employs the Gauss-Newton method to solve the resulting nonlinear least-squares problem. To decrease the cost of computing online solutions, the method approximates the Jacobian and residual of the nonlinear system using a "Gappy POD" approach, which requires computing only a few rows of these quantities online. Large-scale fluid and structural dynamics problems highlight the ability of the proposed method to deliver accurate solutions in real time. The second method is a proper orthogonal decomposition (POD)-augmented conjugate-gradient algorithm, which is applicable to repeated analyses characterized by nearby, symmetric positive definite matrices. This method generates a POD basis on-the-fly and uses it to accelerate convergence of the typical preconditioned conjugate-gradient algorithm to any specified tolerance. The resulting method can be combined with other augmented Krylov-subspace iterative methods if desired. The analysis of a parameterized V-22 tiltrotor wing panel demonstrates the efficiency of the algorithm. Forthcoming: Thu 04 Nov Lindsay Erickson Sandia Thu 11 Nov David Fong iCME Thu 18 Nov Todd Plantenga Sandia -------------- next part -------------- An HTML attachment was scrubbed... URL: http://csmr.ca.sandia.gov/pipermail/banana/attachments/20101026/1a676203/attachment-0001.html From mgu at math.berkeley.edu Thu Oct 28 06:12:00 2010 From: mgu at math.berkeley.edu (Ming Gu) Date: Thu Oct 28 06:16:10 2010 Subject: [BANANA] LAPACK seminar on Nov. 3, 2010 Message-ID: <201010281312.o9SDC0mD019014@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. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Nov. 3, 2010 Speaker: F. Alberto Grunbaum, UC Berkeley Title: Quantum walks and linear algebra Abstract: I will describe from scratch what people call a Quantum walk, and indicate how the spectral theory of the appropriate unitary operator can be helpful in answering some questions about recurrence and localization for these walks. I will try to show how these connections give rise to (possibly) new questions in linear algebra. I welcome a lot of feedback from the audience. Date: Nov. 10, 2010 Speaker: Bin Yu, UC Berkeley Title: Sparse Modeling: a Statistical View From saunders at stanford.edu Thu Oct 28 10:36:09 2010 From: saunders at stanford.edu (Michael Saunders) Date: Thu Oct 28 10:38:00 2010 Subject: [BANANA] Fwd: [icme-linear-algebra-opt LA/Opt seminar TODAY (Kevin Carlberg) Message-ID: Reminder: seminar this afternoon Linear Algebra and Optimization Seminar (CME 510) iCME, Stanford University http://icme.stanford.edu/seminars/seminar.php?seminar_id=2¤t=true 4:15pm Thursday Oct 28, 2010 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Kevin Carlberg PhD Candidate, Aeronautics & Astronautics, Stanford University http://stanford.edu/~carlberg Model reduction-based iterative methods for real-time simulation and repeated analyses of mathematical models Despite the development of efficient solution algorithms, parallel computer architectures, and fast processors, the computational cost of analyzing large-scale, high-fidelity mathematical models remains a significant barrier in many engineering applications. One example is real-time simulation, which arises in such areas as nonlinear model predictive control, fast turnaround design, and "in field" analysis. For general nonlinear problems, standard methods (e.g. mesh coarsening, POD-Galerkin model reduction) often fail to meet solution time requirements without introducing unacceptable errors. Another example is the repeated analyses context, which occurs in nonlinear analysis, design optimization, and parameter space sampling, for example. These problems typically require solving a sequence of "nearby" linear systems of equations to specified tolerances; while existing augmented Krylov-subspace iterative methods are able to deliver the required accuracy, they are only efficient when the number of linear systems remains modest or the matrices have spectra of special form. This talk describes two iterative methods that aim to meet the demands of real-time simulation and repeated analyses problems. The first method is a system-approximated Gauss-Newton method, which enables near real-time nonlinear analysis. To achieve high levels of accuracy in the "online" solution, the method searches for solutions in a low-dimensional proper orthogonal decomposition (POD) subspace that is computed a priori during an "offline" data collection stage. The method employs the Gauss-Newton method to solve the resulting nonlinear least-squares problem. To decrease the cost of computing online solutions, the method approximates the Jacobian and residual of the nonlinear system using a "Gappy POD" approach, which requires computing only a few rows of these quantities online. Large-scale fluid and structural dynamics problems highlight the ability of the proposed method to deliver accurate solutions in real time. The second method is a proper orthogonal decomposition (POD)-augmented conjugate-gradient algorithm, which is applicable to repeated analyses characterized by nearby, symmetric positive definite matrices. This method generates a POD basis on-the-fly and uses it to accelerate convergence of the typical preconditioned conjugate-gradient algorithm to any specified tolerance. The resulting method can be combined with other augmented Krylov-subspace iterative methods if desired. The analysis of a parameterized V-22 tiltrotor wing panel demonstrates the efficiency of the algorithm. Forthcoming: Thu 04 Nov Lindsay Erickson Sandia Thu 11 Nov David Fong iCME _______________________________________________ icme-linear-algebra-optimization mailing list icme-linear-algebra-optimization@lists.stanford.edu https://mailman.stanford.edu/mailman/listinfo/icme-linear-algebra-optimization -------------- next part -------------- An HTML attachment was scrubbed... URL: http://csmr.ca.sandia.gov/pipermail/banana/attachments/20101028/d0cd24ca/attachment.html