Calendar

OCTOBER BREAK

Statistical BioinformaticsNO SEMINAR: FALL BREAK

Working Algebraic Geometry Seminar, Peter Scheiblechner, Purdue University, MATH 731

Algorithms for Counting Components of Algebraic Varieties
Abstract: We present a uniform method for the two problems of counting the connected and irreducible components of complex algebraic varieties. Our algorithms work in parallel polynomial time, i.e., they can be implemented by algebraic circuits of polynomial depth. The design of our algorithms relies on the concept of algebraic differential forms. A further important building block is an algorithm of Szanto computing a variant of characteristic sets.

Commutative Algebra Seminar, Professor Manoj Kummini, Purdue University, MATH 211

Homological Invariants of Bipartite Edge Ideals (II)

Exploring Statistical Sciences Research, Dr. Olga Vitek, Department of Statistics, Purdue University, REC 112

An introduction to statistical methods for molecular profiling of diseaseThe talk will introduce the general area of profiling of disease based on modern high-throughput molecular technologies, in particular using mass spectrometry-based proteomic and metabolomic experiments. Multiple steps of these experiments require statistical analysis and expertise. These include selection of biological samples, planning the process of spectral acquisition, identification and quantification of spectral features, integration of spectral data with patient's clinical characteristics and biological annotations, and planning subsequent experiments.I'll describe statistical approaches currently used with these experiments, and will give examples of research in these areas by students in my lab.

Automorphic Forms and Representation Theory Seminar, Professor Freydoon Shahidi, Purdue University, MATH 731

Some Harmonic Analysis Inspired by Functoriality

Topology Seminar, Professor Ben Walter, Purdue University, REC 117

A New Model for Harrison Homology
Abstract: A new model for Harrison homology
Abstract: Classically the Harrison complex is given by quotienting the Hochschild complex of a differential graded commutative algebra by the subcomplex of shuffles. The first Eulerian idempotent gives a homotopy splitting of the quotient map from the Hochschild complex to the Harrison complex, so Harrison homology can be used to recover part of Hochschild homology. Also classically, Harrison homology is the Andre-Quillen homology theory for commutative rings. Koszul duality theory for operads then implies that Harrison homology should have the structure of a Lie coalgebra. I will give a new model for Harrison homology using coalgebras of graphs in which the Lie coalgebra structure is clear. The model will also come with an easy way to distinguish when two elements in the Harrison complex are different which I will use to give extremely simple proofs of standard Harrison theory propositions. The model I propose is analogous to expressing a Lie algebra as a quotient of a non-associative algebra - the classical construction is analogous to expressing a Lie algebra as a subalgebra of a non-commutative algebra with the commutator bracket.

PDE Seminar, Professor Fang Li, Purdue University, REC 315

Stability from the Point of View of Diffusion, Relaxation, and Spatial Inhomogeneity

Research Colloquium, Dr. Jon Weinstock, Cisco Systems Inc., MATH 175

Less is More: Understanding and Managing Multi-Step Retry Reinitialization Protocol FlowsOver the past decade or so, residential broadband access has created massive deployments of IP based devices in the home. These devices support services such as high speed data, routing, switching, voice, video and gaming, as well as others still being planned. In certain circumstances, these devices can and will operate autonomously, as clients with various network servers. This interaction takes place, among other reasons, to allow for initial device/application configuration, control as well as for IP based service registration. Interactions with network servers include Presence, Call, TFTP, Kerberos, DNS, Provisioning, DHCP, etc. Many of these client to server interactions follow a common pattern. This pattern is one of an ordered step by step interchange of request/response pairs taking place between single clients and multiple servers. At each point in this step by step interaction, clients wait for some time limit and if a response has not been received, retry the request. After a specified number of retries, clients will reinitialize the interaction to a previous step (often starting completely over at step one). The servers’ behavior is typically not complex, as they will simply queue and provide service with some latency. I will call this step by step interaction a “Multi Step Retry Reinitialization Protocol Flow”. Another interesting aspect of these massive device deployments is the degree to which very large aggregates of these devices depend on mutual resources. Examples of these shared resources include such things as the electrical grid, shared communications nodes as well as electromagnetic spectrum. This mutual dependency on shared resources causes the IP devices themselves, under certain circumstances, to lose their operational and behavioral independence. This can happen at the worst times, such as in a disaster. A consequence of this lack of behavioral independence can be an “avalanche” of Multi- Step Retry Reinitialization Protocol Flows into the service infrastructure as initiated by the client devices. This happens as each device seeks to reestablish itself on the IP network and “reattach” to each of the services provided by that network. In this talk I will describe work that was done at Cisco in 2008 where we looked at the behavior of Multi Step Retry Reinitialization Protocol Flows under extremely high load, (arrival rates are vastly in excess of service rates), such as those created by a disaster. I will focus on a particular simple example of a Multi Step Retry Reinitialization Protocol Flow, DHCP, describe our simulation results and discuss the requirement for specialized queuing models as well as other potential future work. Refreshments will be served at 4:15 PM in HAAS 101.

Computational and Applied Mathematics Seminar, Professor Houman Owhadi, California Institute of Technology, REC 108

Uncertainty Quantification Through Concentration of Measure Inequalities
Abstract: We apply concentration-of-measure inequalities to the quantification of uncertainties in the performance of engineering systems. Specifically, we envision uncertainty quantification in the context of certification, i.e., as a tool for deciding whether a system is likely to perform safely and reliably within design specifications. We show that concentration-of-measure inequalities rigorously bound probabilities of failure and thus supply conservative certification criteria. In addition, they supply unambiguous quantitative definitions of terms such as margins, epistemic and aleatoric uncertainties, verification and validation measures, confidence factors, and others, as well as providing clear procedures for computing these quantitiesby means of concerted simulation and experimental campaigns. We also investigate numerically the tightness of concentration-of-measure inequalities with the aid of imploding and exploding ring examples. Our numerical tests establish the robustness and viability of concentration-of-measure inequalities as a basis for certification in that particular example of application.

Department of Mathematics Colloquium, Professor David Eisenbud, MSRI/UC Berkeley, BRNG 1245

Plato's Cave: What We Still Don't Know About Generic ProjectionsAbstract. Riemann Surfaces were first studied algebraically by first projecting them into the complex projective plan; later the same idea was applied to surfaces and higher dimensional varieties, projecting them to hypersurfaces. How much damage is done in this process? For example, what can the fibers of a generic linear projection look like? This is pretty easy for smooth curves and surfaces (though there are still open questions), not so easy in the higher-dimensional case. I'll explain some of what's known, including recent work of mine with Roya Beheshti, Joe Harris, and Bernd Ulrich.

Geometric Analysis Seminar, Mr. Vakhid Masagutov, Purdue University, REC 114

Homomorphisms of Infinitely Generated Analytic Sheaves

Math Biology Seminar, UNIV 217

TBA

Department of Mathematics Colloquium, Professor Troy Day, Queen's University, MA 175

TBA Refreshments will be served in the Math Library Lounge at 4:00 p.m.

Exploring Statistical Sciences Research, Dr. S. V. N. Vishwanathan, Department of Statistics, Department of Computer Science, Purdue University, REC 112

TBA

Automorphic Forms and Representation Theory Seminar, Mr. Daniel Szpruch, Tel Aviv University, MATH 731

On Some Applications of the Langland-Shahidi Method to the Metaplectic Group

Research Seminar, Dr. Bowei Xi, Department of Statistics, Purdue University, MATH 175

Classifier Evaluation and Attribute Selection against Active AdversariesMany data mining applications, ranging from spam filtering to intrusion detection, are faced with active adversaries. In all these applications, initially successful classifiers will degrade easily. This becomes a game between the adversary and the data miner: The adversary modifies its strategy to avoid being detected by the current classifier; the data miner then updates its classifier based on the new threats. We investigate the possibility of an equilibrium in this seemingly never ending game, where neither party has an incentive to change. Modifying the classifier causes too many false positives with too little increase in true positives; changes by the adversary decrease the utility of the false negative items that aren't detected. We develop a game theoretic framework where the equilibrium behavior of adversarial classification applications can be analyzed, and provide a solution for finding the equilibrium point. A classifier's equilibrium performance indicates its eventual success or failure. The data miner could then select attributes based on their equilibrium performance, and construct an effective classifier.This is joint work with Murat Kantarcioglu and Chris Clifton.Refreshments will be served at 4:15 PM in HAAS 101.

Exploring Statistical Sciences Research, Dr. Kristofer Jennings, Department of Statistics, Purdue University, REC 112

TBA

Exploring Statistical Sciences Research, Dr. Jennifer Neville, Department of Statistics, Department of Computer Science, Purdue University, REC 112

TBA

Joint Purdue University and University of Illinois Urbana- Champaign Research Colloquium, Professor Bo Li, Department of Statistics, Purdue University, AT UIUC

Testing the covariance structure of multivariate random fieldsThere is an increasing wealth of multivariate spatial and multivariate spatio-temporal data appearing. For such data an important part of model building is an assessment of the properties of the underlying covariance function describing variable, spatial and temporal correlations. In this paper we propose a methodology to evaluate the appropriateness of several types of common assumptions on multivariate covariance functions in the spatiotemporal context. The methodology is based on the asymptotic joint normality of sample space-time cross-covariance estimators. Specifically, we address the assumptions of symmetry, separability and linear models of coregionalisation. We conduct simulation experiments to evaluate the sizes and powers of our tests and illustrate our methodology on a trivariate spatio-temporal data set of pollutants (CO, NO, NO2) over California.Will leave for UIUC around noon.

Department of Mathematics Colloquium, Marie-Francoise Roy, Universite de Rennesl, MA 175

Geometric Properties of Multivariate Bernstein Basis and Certificates of Positivity
Abstract: The Bernstein basis on a simplex of polynomials of degree less than d can be easily defined. It is well known that in the univariate case, the Bernsteincoefficients of a polynomial define a control line, so called because the shape of the graph of the polynomial can be easily controlled using the Bernstein coefficients. In the multivariate case, even the definition of the control polytope cannot be given immediately from the Bernstein coefficients. In order to define the control polytope and study its distance to the graph of the polynomial, it is convenient to introduce a combinatorial construction: the standard triangulation of a simplex, and tocharacterize combinatorially the convex piecewise linear functions based on this triangulation. Two ways of making the approximation between the control polytope and the graph closer are considered: by elevation of the degree, or by subdivision. Short and adaptive certificate making visible that a polynomial is positive on a simplex follow from the subdivision method. Refreshments will be served in the Math Library Lounge at 4:00 p.m.

Exploring Statistical Sciences Research, Dr. Mark D. Ward, Department of Statistics, Purdue University, REC 112

TBA

Department of Mathematics Colloquium, Professor Steven G. Krantz, Washington University, MATH 175

TBA Refreshments will be served in the Math Library Lounge at 4:00 p.m.

Exploring Statistical Sciences Research, Dr. Stephen J. Ruberg, Eli Lilly & Company, Global Information Sciences, Adjunct Professor of Statistics, Purdue University, REC 112

Tailoring Drug Treatments to the Right PatientsThe genomics revolution is still in its infancy, and there is much to learn about how and why individual patients respond to different drug treatments in different ways. Until we know more about detailed genetic mechanisms, there are many clinical markers (disease history, standard laboratory measures, etc.) that can be used to aid in identifying who may respond or not respond to a particular treatment. The issues can have some statitical complex and different approaches may yield more interesting insights into clinical data. Several approaches will be discussed and real-life examples from the pharmaceutical industry will be used to demonstrate statistical appraoches and contributions to the problem.

Department of Mathematics Colloquium, Professor Winnie Li, Penn State University, MATH 175

TBA Refreshments will be served in the Math Library Lounge at 4:00 p.m.

Exploring Statistical Sciences Research, Dr. George P. McCabe, Associate Dean for Academic Affairs, College of Science and Professor of Statistics, Purdue University, REC 112

TBA

Department of Mathematics Colloquium, Professor Giulio Peruginelli, University of Pisa, MATH 175

TBA

Department of Mathematics Colloquium, Professor Arnaud Bodin, Universite Lille 1, MATH 175

TBA

Department of Mathematics Colloquium, Professor Ignacio Luengo, Universidad Complutense, Madrid, MATH 175

TBA