From saunders at stanford.edu Tue Nov 1 09:58:29 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue Nov 1 09:59:23 2011 Subject: [BANANA] LA/Opt seminar Thursday Nov 3 (Jon Gretarsson) Message-ID: Linear Algebra and Optimization Seminar (CME 510) http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Nov 3, 2011 Y2E2 111 http://campus-map.stanford.edu/index.cfm?ID=04-070 Jon Gretarsson ICME PhD student jon.gretarsson@gmail.com http://www.stanford.edu/people/jontg Fluid-Structure Interactions in Compressible Flow We propose a novel implicit method to capture fluid-structure interactions between Lagrangian solids and compressible flows. This permits the use of large time-steps in simulation without the numerical instabilities inherent in a fully explicit methodology. We split the compressible flow fluxes into a bulk advection term and an implicit pressure term, as in the incompressible flow formulation, obtaining an implicit system of equations for pressure and reducing the CFL time-step restriction from |u|+/-c to |u| (where c is the sound speed of the flow). The implicit pressure term is then two-way coupled with solid damping forces, capturing the momentum transfer across the interface conservatively. The coupling is similar to previous work for incompressible flows, but is quite general and works for arbitrary equations of state. The momentum transfer that arises from the implicit pressure flux is fully conservative, and naturally handles nonlinear phenomena such as shocks, contacts, and rarefactions. The resulting system is symmetric, and can be made positive definite under certain reasonable assumptions. We conclude by exploring some applications and extensions into computer graphics. If there's time, I'll also talk about some soon-to-be-published results where I explore a fully conservative cut-cell method that avoids introducing any new time-step restrictions. Forthcoming: Thu Nov 10 Paul Constantine, Stanford Thu Nov 17 Richard Li-Yang Chen, Sandia Livermore Thu Nov 24 Thanksgiving Thu Dec 01 Oren Livne, Univ of Chicago From saunders at stanford.edu Thu Nov 3 10:25:03 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Nov 3 10:25:56 2011 Subject: [BANANA] LA/Opt seminar TODAY (Jon Gretarsson) Message-ID: Reminder: seminar this afternoon: Linear Algebra and Optimization Seminar (CME 510) http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Nov 3, 2011 Y2E2 111 http://campus-map.stanford.edu/index.cfm?ID=04-070 Jon Gretarsson ICME PhD student jon.gretarsson@gmail.com http://www.stanford.edu/people/jontg Fluid-Structure Interactions in Compressible Flow We propose a novel implicit method to capture fluid-structure interactions between Lagrangian solids and compressible flows. This permits the use of large time-steps in simulation without the numerical instabilities inherent in a fully explicit methodology. We split the compressible flow fluxes into a bulk advection term and an implicit pressure term, as in the incompressible flow formulation, obtaining an implicit system of equations for pressure and reducing the CFL time-step restriction from |u|+/-c to |u| (where c is the sound speed of the flow). The implicit pressure term is then two-way coupled with solid damping forces, capturing the momentum transfer across the interface conservatively. The coupling is similar to previous work for incompressible flows, but is quite general and works for arbitrary equations of state. The momentum transfer that arises from the implicit pressure flux is fully conservative, and naturally handles nonlinear phenomena such as shocks, contacts, and rarefactions. The resulting system is symmetric, and can be made positive definite under certain reasonable assumptions. We conclude by exploring some applications and extensions into computer graphics. If there's time, I'll also talk about some soon-to-be-published results where I explore a fully conservative cut-cell method that avoids introducing any new time-step restrictions. Forthcoming: Thu Nov 10 Paul Constantine, Stanford Thu Nov 17 Richard Li-Yang Chen, Sandia Livermore Thu Nov 24 Thanksgiving Thu Dec 01 Oren Livne, Univ of Chicago From mgu at math.berkeley.edu Thu Nov 3 22:58:13 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Thu Nov 3 22:59:49 2011 Subject: [BANANA] LAPACK seminar on Nov. 9, 2011 In-Reply-To: <201110220317.p9M3HnBK000627@phoenix.math.berkeley.edu> References: <201110220317.p9M3HnBK000627@phoenix.math.berkeley.edu> Message-ID: Math 290, Section 25, CS 298, Section 6 Fall 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 12:10 - 1:00PM 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. Note that Prof. Kahan has prepared some reading material for his tutorial on Nov. 9 (see below.) Please read it before his lectures if possible. This might help prepare you to ask him questions during the lectures. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Nov. 9, 2011 Speaker: Prof. W. Kahan, UC Berkeley Title: A Tutorial Overview of Vector and Matrix Norms Abstract: Intended for new graduate students whose experience as undergraduates may have prepared them inadequately to apply norms to numerical error-analyses and to proofs of convergence, this tutorial surveys norms for finite-dimensional real spaces in a way that may ease a transition to the infinite-dimensional spaces of Functional Analysis. The tutorial?s notation is mostly standard but interpreted in ways not always taught to undergraduates, so attendees may prepare for it by reading just a few of my lecture notes for Math H110 posted at eecs.berkeley.edu/~wkahan/MathH110/2dspaces.pdf and ... /pts.pdf in that order and afterwards .../geo.pdf skimmed lightly. This tutorial omits proofs; almost all can be found in .../NORMlite.pdf and .../GIlite.pdf, and a few other places cited. This is the last of a series of lectures on vector and matrix norms. This tutorial?s text is to be posted at eecs.berkeley.edu/~wkahan/MathH110/NormOvrv.pdf Date: Nov. 16, 2011 Speaker: Andy Packard, UC Berkleey From saunders at stanford.edu Mon Nov 7 14:07:20 2011 From: saunders at stanford.edu (Michael Saunders) Date: Mon Nov 7 14:08:13 2011 Subject: [BANANA] LA/Opt seminar Thursday Nov 10 (Paul Constantine) Message-ID: Linear Algebra and Optimization Seminar (CME 510) http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Nov 10, 2011 Y2E2 111 http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Paul Constantine ICME graduate 2009 Postdoctoral Fellow, Mechanical Engineering, Stanford Univ paulcon@stanford.edu http://www.stanford.edu/~paulcon/ Experiences with Model Reduction and Interpolation I have spent the last two years as the von Neumann Fellow at Sandia National Labs in Albuquerque studying methods for constructing reduced order models of parameterized partial differential equation models. A reduced order model provides a practical surrogate for an expensive PDE simulation that must be executed many times at different points in the parameter space, e.g., for uncertainty, sensitivity, or optimization studies. The surrogate is constructed from the outputs of a few simulation runs, and it can be evaluated independently of the high performance code. I will present an overview of this work and discuss experiences with these methods including some algorithmic considerations and plenty of directions for future research. Forthcoming: Thu Nov 17 Richard Li-Yang Chen, Sandia Livermore Thu Nov 24 Thanksgiving Thu Dec 01 Oren Livne, Univ of Chicago From mgu at math.berkeley.edu Tue Nov 8 06:51:11 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Nov 8 06:53:01 2011 Subject: [BANANA] Reminder: LAPACK seminar on Nov. 9, 2011 In-Reply-To: References: <201110220317.p9M3HnBK000627@phoenix.math.berkeley.edu> Message-ID: Math 290, Section 25, CS 298, Section 6 Fall 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 12:10 - 1:00PM 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. Note that Prof. Kahan has prepared some reading material for his tutorial on Nov. 9 (see below.) Please read it before his lectures if possible. This might help prepare you to ask him questions during the lectures. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Nov. 9, 2011 Speaker: Prof. W. Kahan, UC Berkeley Title: A Tutorial Overview of Vector and Matrix Norms This tutorial?s text is posted at eecs.berkeley.edu/~wkahan/MathH110/NormOvrv.pdf Date: Nov. 16, 2011 Speaker: Andy Packard, UC Berkleey From saunders at stanford.edu Thu Nov 10 11:56:54 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Nov 10 11:57:49 2011 Subject: [BANANA] LA/Opt seminar TODAY Nov 10 (Paul Constantine) Message-ID: Reminder: LA/Opt seminar this afternoon ?Linear Algebra and Optimization Seminar (CME 510) ?http://icme.stanford.edu/seminars/seminars.php ?4:15pm Thursday Nov 10, 2011 ?Y2E2 111 ?http://campus-map.stanford.edu/index.cfm?ID=04-070 ?Dr Paul Constantine ?ICME graduate 2009 ?Postdoctoral Fellow, Mechanical Engineering, Stanford Univ ?paulcon@stanford.edu ?http://www.stanford.edu/~paulcon/ ?Experiences with Model Reduction and Interpolation I have spent the last two years as the von Neumann Fellow at Sandia National Labs in Albuquerque studying methods for constructing reduced order models of parameterized partial differential equation models. ?A reduced order model provides a practical surrogate for an expensive PDE simulation that must be executed many times at different points in the parameter space, e.g., for uncertainty, sensitivity, or optimization studies. ?The surrogate is constructed from the outputs of a few simulation runs, and it can be evaluated independently of the high performance code. I will present an overview of this work and discuss experiences with these methods including some algorithmic considerations and plenty of directions for future research. Forthcoming: Thu Nov 17 ?Richard Li-Yang Chen, Sandia Livermore Thu Nov 24 ?Thanksgiving Thu Dec 01 ?Oren Livne, Univ of Chicago From mgu at math.berkeley.edu Mon Nov 14 07:39:43 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon Nov 14 07:41:18 2011 Subject: [BANANA] LAPACK seminar on Nov. 16, 2011 Message-ID: <201111141539.pAEFdhfi015967@panda.math.berkeley.edu> Math 290, Section 25, CS 298, Section 6 Fall 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 12:10 - 1:00PM 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. There will be no seminar on Nov. 23 due to the Thanksgiving holiday. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Nov. 16, 2011 Speaker: Andy Packard, UCB Title: Establishing dynamical system properties with semidefinite and sum-of-squares programming Abstract: Assessing the region-of-attraction, reachability, and input/output gain of dynamical systems are different instances of quantitative certification metrics. Some of the earliest applications of SDP trace back to assessing related certification questions for uncertain linear systems, where optimizations with indefinite quadratic objectives and indefinite quadratic constraints arise quite naturally. Less scalable generalizations to nonlinear dynamical systems can lead to SOS programming. In this talk, we will introduce and summarize some results in these directions. From saunders at stanford.edu Mon Nov 14 13:01:39 2011 From: saunders at stanford.edu (Michael Saunders) Date: Mon Nov 14 13:02:40 2011 Subject: [BANANA] LA/Opt seminar Thursday Nov 17 (Richard Li-Yang Chen) Message-ID: Linear Algebra and Optimization Seminar (CME 510) http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Nov 17, 2011 Y2E2 111 http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Richard Li-Yang Chen Senior Member of Technical Staff Quantitative Modeling and Analysis Department Sandia National Laboratories, Livermore, CA rlchen@sandia.gov Contingency-constrained Optimization for Electric Power Systems Disruptions in electric power systems can have serious economic and even catastrophic consequences. As such, power system operators must emphasize both security and cost effectiveness in their operations and planning decisions. In this talk, we present a contingency-constrained unit-commitment (CCUC) model. The objective is to determine a minimal cost turn-on and turn-off schedule for a set of electrical power generating units to meet uncertain load demand while satisfying a set of operational and security constraints. We describe a cutting plane algorithm (CPA) for solving the CCUC. The CPA relies on bilevel optimization to identify critical system vulnerabilities and generate associated cutting planes. The effectiveness of the CPA depends largely on our ability to efficiently solve the bilevel program. We present efficient reformulations of the bilevel program and combine network-based separation models to more efficiently identify system vulnerabilities. Finally, we demonstrate the effectiveness of our approach on standard IEEE test systems. Forthcoming: Thu Nov 24 Thanksgiving Thu Dec 01 Oren Livne, Univ of Chicago From mgu at math.berkeley.edu Tue Nov 15 09:57:53 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Tue Nov 15 09:59:33 2011 Subject: [BANANA] Reminder: LAPACK seminar on Nov. 16, 2011 In-Reply-To: <201111141539.pAEFdhfi015967@panda.math.berkeley.edu> References: <201111141539.pAEFdhfi015967@panda.math.berkeley.edu> Message-ID: Math 290, Section 25, CS 298, Section 6 Fall 2011 (Matrix Computations and Scientific Computing) We meet WEDNESDAYS 12:10 - 1:00PM 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. There will be no seminar on Nov. 23 due to the Thanksgiving holiday. For the schedule and other details about the seminar, please see math.berkeley.edu/~mgu/LAPACKSeminar.htm Date: Nov. 16, 2011 Speaker: Andy Packard, UCB Title: Establishing dynamical system properties with semidefinite and sum-of-squares programming From saunders at stanford.edu Thu Nov 17 10:06:13 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu Nov 17 10:07:06 2011 Subject: [BANANA] LA/Opt seminar TODAY (Richard Chen) Message-ID: Reminder: seminar this afternoon Linear Algebra and Optimization Seminar (CME 510) http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday Nov 17, 2011 Y2E2 111 http://campus-map.stanford.edu/index.cfm?ID=04-070 Dr Richard Li-Yang Chen Senior Member of Technical Staff Quantitative Modeling and Analysis Department Sandia National Laboratories, Livermore, CA rlchen@sandia.gov Contingency-constrained Optimization for Electric Power Systems Disruptions in electric power systems can have serious economic and even catastrophic consequences. As such, power system operators must emphasize both security and cost effectiveness in their operations and planning decisions. In this talk, we present a contingency-constrained unit-commitment (CCUC) model. The objective is to determine a minimal cost turn-on and turn-off schedule for a set of electrical power generating units to meet uncertain load demand while satisfying a set of operational and security constraints. We describe a cutting plane algorithm (CPA) for solving the CCUC. The CPA relies on bilevel optimization to identify critical system vulnerabilities and generate associated cutting planes. The effectiveness of the CPA depends largely on our ability to efficiently solve the bilevel program. We present efficient reformulations of the bilevel program and combine network-based separation models to more efficiently identify system vulnerabilities. Finally, we demonstrate the effectiveness of our approach on standard IEEE test systems. Forthcoming: Thu Nov 24 Thanksgiving Thu Dec 01 Oren Livne, Univ of Chicago Thu Dec 08 Chen Greif, UBC