From mgu at math.berkeley.edu Mon May 2 18:49:13 2011 From: mgu at math.berkeley.edu (Ming Gu) Date: Mon May 2 18:50:10 2011 Subject: [BANANA] LAPACK seminar on May 4, 2011 In-Reply-To: <22152_1304030813_p3SMkn61010918_201104282245.p3SMjJqZ008577@panda.math.berkeley.edu> References: <22152_1304030813_p3SMkn61010918_201104282245.p3SMjJqZ008577@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: May 4, 2011 Speaker: Takeshi Iwashita, Academic Center for Computing and Media Studies, Kyoto University, Japan Title: Hybrid parallel ordering method for a parallelized multiplicative Schwarz smoother in a multigrid solver for time-harmonic electromagnetic field problems This is the last talk for this semester. Have a great summer. From saunders at stanford.edu Tue May 3 09:31:58 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue May 3 09:33:46 2011 Subject: [BANANA] LA/Opt SCREAM seminar Thursday May 5 (Nicole Taheri) Message-ID: SIAM Stanford Student Chapter SCREAM seminar, and Linear Algebra and Optimization Seminar (CME 510) http://www.stanford.edu/group/siam/events.html http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday May 5, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Refreshments served! Nicole Taheri ICME PhD student ntaheri@stanford.edu A Dynamic Mechanism to Provide Grid Services to a Fleet of Plug-in Electric Vehicles Plug-in Electric Vehicles (PEVs) will play an important role in the future electricity grid. They will run on energy provided by a battery, which will also have the capability of transmitting electricity to the grid. This feature makes the grid structure more complicated, but if managed properly, it can improve the electricity supply-demand stability. We consider the role of an *aggregator*, who determines the charging and discharging schedule for groups of PEVs, or fleets, by communicating each fleet's demand to utility providers. The focus of our work is to construct a dynamic and robust mechanism that demonstrates the aggregator's decision-making process. Linear Programming is used to solve the aggregated problem by determining a charging schedule that maximizes profit while ensuring that the overall fleet has enough battery storage to meet its transport load. We prove that if there exists an aggregated solution of a charging schedule that satisfies the fleet as a whole, then there is always a feasible schedule to the disaggregated problem so that each individual vehicle is satisfied. And finally, we create and implement a dynamic decision-making process that determines individual PEV charging schedules instantly as they connect to the grid, using threshold pricing dependent on dual variables from the aggregated problem. Forthcoming: Thu May 12 Andrew Spann How a Pixar rendering technique turned out to be applicable to biological membrane flow problems Thu May 19 Aravindakshan Babu Statistical topology: Zigzags and Reeb graphs Thu May 26 Mike Lesnick Developments in the theory of multidimensional persistence From saunders at stanford.edu Thu May 5 10:03:06 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu May 5 10:09:13 2011 Subject: [BANANA] LA/Opt SCREAM seminar TODAY (Nicole Taheri) Message-ID: Reminder: SIAM Stanford Student Chapter SCREAM seminar, and Linear Algebra and Optimization Seminar (CME 510) http://www.stanford.edu/group/siam/events.html http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday May 5, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Refreshments served beforehand! Nicole Taheri ICME PhD student ntaheri@stanford.edu A Dynamic Mechanism to Provide Grid Services to a Fleet of Plug-in Electric Vehicles Plug-in Electric Vehicles (PEVs) will play an important role in the future electricity grid. They will run on energy provided by a battery, which will also have the capability of transmitting electricity to the grid. This feature makes the grid structure more complicated, but if managed properly, it can improve the electricity supply-demand stability. We consider the role of an *aggregator*, who determines the charging and discharging schedule for groups of PEVs, or fleets, by communicating each fleet's demand to utility providers. The focus of our work is to construct a dynamic and robust mechanism that demonstrates the aggregator's decision-making process. Linear Programming is used to solve the aggregated problem by determining a charging schedule that maximizes profit while ensuring that the overall fleet has enough battery storage to meet its transport load. We prove that if there exists an aggregated solution of a charging schedule that satisfies the fleet as a whole, then there is always a feasible schedule to the disaggregated problem so that each individual vehicle is satisfied. And finally, we create and implement a dynamic decision-making process that determines individual PEV charging schedules instantly as they connect to the grid, using threshold pricing dependent on dual variables from the aggregated problem. Forthcoming: Thu May 12 Andrew Spann How a Pixar rendering technique turned out to be applicable to biological membrane flow problems Thu May 19 Aravindakshan Babu Statistical topology: Zigzags and Reeb graphs Thu May 26 Mike Lesnick Developments in the theory of multidimensional persistence From saunders at stanford.edu Tue May 10 09:55:58 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue May 10 09:57:52 2011 Subject: [BANANA] LA/Opt SCREAM seminar Thursday May 12 (Andrew Spann) Message-ID: SIAM Stanford Student Chapter SCREAM seminar, and Linear Algebra and Optimization Seminar (CME 510) http://www.stanford.edu/group/siam/events.html http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday May 12, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Refreshments served beforehand! Andrew Spann ICME PhD student http://antares.stanford.edu/index.php/AndrewSpann/HomePage Subdivision Schemes in Boundary Integral Method Simluations of Vesicles: How a Pixar Rendering Technique Turned Out to be Applicable to Biological Membrane Flow Problems Vesicles are a category of membrane-bound objects that are useful in modeling blood flow related problems. Unlike problems involving liquid drops, vesicles have a bending energy that is a nonlinear function of curvature. Thus we must devise a highly accurate way to compute the curvature of a discrete cloud of points. We accomplish this by taking the technique of subdivision, originally used as a means to transform 3D polygon meshes into smooth surfaces. In the first half of the talk we will introduce subdivision surfaces, and then we will explore results and movies from vesicle simulations. Forthcoming: Thu May 19 Aravindakshan Babu Statistical topology: Zigzags and Reeb graphs Thu May 26 Mike Lesnick Developments in the theory of multidimensional persistence From saunders at stanford.edu Thu May 12 10:03:30 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu May 12 10:05:22 2011 Subject: [BANANA] LA/Opt SCREAM seminar TODAY (Andrew Spann) Message-ID: Reminder: seminar this afternoon. Refreshments served beforehand! ?SIAM Stanford Student Chapter SCREAM seminar, and ?Linear Algebra and Optimization Seminar (CME 510) ?http://www.stanford.edu/group/siam/events.html ?http://icme.stanford.edu/seminars/seminars.php ?4:15pm Thursday May 12, 2011 ?Y2E2 101 ?http://campus-map.stanford.edu/index.cfm?ID=04-070 ?Andrew Spann ?ICME PhD student ?http://antares.stanford.edu/index.php/AndrewSpann/HomePage ?Subdivision Schemes in Boundary Integral Method Simluations of Vesicles: ?How a Pixar Rendering Technique Turned Out to be Applicable to Biological ?Membrane Flow Problems Vesicles are a category of membrane-bound objects that are useful in modeling blood flow related problems. ?Unlike problems involving liquid drops, vesicles have a bending energy that is a nonlinear function of curvature. ?Thus we must devise a highly accurate way to compute the curvature of a discrete cloud of points. ?We accomplish this by taking the technique of subdivision, originally used as a means to transform 3D polygon meshes into smooth surfaces. ?In the first half of the talk we will introduce subdivision surfaces, and then we will explore results and movies from vesicle simulations. Forthcoming: Thu May 19 ?Aravindakshan Babu ?Statistical topology: ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? Zigzags and Reeb graphs Thu May 26 ?Mike Lesnick ? ? ? ?Developments in the theory of ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?multidimensional persistence From saunders at stanford.edu Mon May 16 19:32:04 2011 From: saunders at stanford.edu (Michael Saunders) Date: Mon May 16 19:33:59 2011 Subject: [BANANA] LA/Opt SCREAM seminar Thursday May 19 (Aravindakshan Babu) Message-ID: SIAM Stanford Student Chapter SCREAM seminar, and Linear Algebra and Optimization Seminar (CME 510) http://www.stanford.edu/group/siam/events.html http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday May 19, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Refreshments served beforehand! Aravindakshan Babu ICME PhD student http://icme.stanford.edu/people/students/newProfile.php?personnel_id=127 Statistical topology: Zigzags and Reeb graphs Reeb graph reconstruction is an important problem in topological data analysis. We study the error in reconstructing the Reeb graph of a point cloud sampled from a manifold, using the MAPPER algorithm. Using the language of zigzag persistence, we are able to succinctly capture the difference between the reconstructed and actual Reeb graphs. A worst-case bound on the error is derived in terms of the sampling density. We also demonstrate practical applications such as dataset classification. Forthcoming: Thu May 26 Mike Lesnick Developments in the theory of multidimensional persistence From saunders at stanford.edu Thu May 19 11:05:25 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu May 19 11:07:20 2011 Subject: [BANANA] LA/Opt SCREAM seminar TODAY (Aravindakshan Babu) Message-ID: Reminder: seminar this afternoon ?SIAM Stanford Student Chapter SCREAM seminar, and ?Linear Algebra and Optimization Seminar (CME 510) ?http://www.stanford.edu/group/siam/events.html ?http://icme.stanford.edu/seminars/seminars.php ?4:15pm Thursday May 19, 2011 ?Y2E2 101 ?http://campus-map.stanford.edu/index.cfm?ID=04-070 ?Refreshments served beforehand! ?Aravindakshan Babu ?ICME PhD student ?http://icme.stanford.edu/people/students/newProfile.php?personnel_id=127 ?Statistical topology: Zigzags and Reeb graphs Reeb graph reconstruction is an important problem in topological data analysis. ?We study the error in reconstructing the Reeb graph of a point cloud sampled from a manifold, using the MAPPER algorithm. Using the language of zigzag persistence, we are able to succinctly capture the difference between the reconstructed and actual Reeb graphs. ?A worst-case bound on the error is derived in terms of the sampling density. ?We also demonstrate practical applications such as dataset classification. Forthcoming: Thu May 26 ?Mike Lesnick ? ?Developments in the theory of multidimensional persistence From saunders at stanford.edu Tue May 24 09:36:07 2011 From: saunders at stanford.edu (Michael Saunders) Date: Tue May 24 09:38:01 2011 Subject: [BANANA] LA/Opt SCREAM seminar Thursday May 26 (Mike Lesnick) Message-ID: SIAM Stanford Student Chapter SCREAM seminar, and Linear Algebra and Optimization Seminar (CME 510) http://www.stanford.edu/group/siam/events.html http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday May 26, 2011 Y2E2 101 http://campus-map.stanford.edu/index.cfm?ID=04-070 Refreshments served beforehand! Mike Lesnick ICME PhD student mlesnick@stanford.edu The Optimality of the Interleaving Distance on Multidimensional Persistence Modules Persistent homology, a multiscale extension of the classical homology functor of algebraic topology, has been the subject of a great deal of interest in the last decade. It is used to construct algebraic signatures of a very wide class of geometric objects, including point cloud data sets and functions on arbitrary topological spaces. These signatures have some very nice theoretical and computational properties, and have been applied to a wide variety of problems in computational geometry and exploratory data analysis. They are central objects of study in the nascent field of topological statistics. One of the nice properties of these signatures is that there exists a well behaved and readily computable metric on them called the bottleneck distance. The bottleneck distance plays an important role in theoretical results and applications of persistence. This talk will discuss recent theoretical work on the problem of adapting the bottleneck distance to a generalized form of persistent homology called multidimensional persistent homology. From saunders at stanford.edu Thu May 26 10:09:04 2011 From: saunders at stanford.edu (Michael Saunders) Date: Thu May 26 10:10:58 2011 Subject: [BANANA] LA/Opt SCREAM seminar TODAY (Mike Lesnick) Message-ID: Reminder: Last seminar for Spring 2011 Refreshments served beforehand! SIAM Stanford Student Chapter SCREAM seminar, and Linear Algebra and Optimization Seminar (CME 510) http://www.stanford.edu/group/siam/events.html http://icme.stanford.edu/seminars/seminars.php 4:15pm Thursday May 26, 2011 Mike Lesnick ICME PhD student mlesnick@stanford.edu The Optimality of the Interleaving Distance on Multidimensional Persistence Modules Persistent homology, a multiscale extension of the classical homology functor of algebraic topology, has been the subject of a great deal of interest in the last decade. It is used to construct algebraic signatures of a very wide class of geometric objects, including point cloud data sets and functions on arbitrary topological spaces. These signatures have some very nice theoretical and computational properties, and have been applied to a wide variety of problems in computational geometry and exploratory data analysis. They are central objects of study in the nascent field of topological statistics. One of the nice properties of these signatures is that there exists a well behaved and readily computable metric on them called the bottleneck distance. The bottleneck distance plays an important role in theoretical results and applications of persistence. This talk will discuss recent theoretical work on the problem of adapting the bottleneck distance to a generalized form of persistent homology called multidimensional persistent homology. Whereas ordinary persistent homology produces algebraic invariants of topological spaces filtered by a single real parameter, multidimensional persistent homology produces algebraic invariants of topological spaces filtered by several real parameters. This generalization allows for the construction of signatures of metric spaces that encode much more geometric information than those that can be constructed using ordinary persistence alone. Our central result is a proof that the interleaving distance, a generalization of the bottleneck distance to the setting of multidimensional persistent homology, is optimal in a certain sense. This result is new even for ordinary persistent homology. We also show that the computation of the interleaving distance between two signatures of total size m reduces to determining whether solutions exist to O(log m) multivariate systems of quadratic equations, each with O(m^2) variables and O(m^2) equations.