The Department of Mathematics sponsors a regular colloquium that hosts several talks by
external speakers every year and a general faculty seminar for talks by faculty and
students. Seminars are listed below after the colloquia.
On Thursday, November 8, Dr. Joshua Hudson, Applied Physics Lab at Johns Hopkins University, will present a colloquium lecture on Using data assimilation to better approximate flows and as a means to measure physical parameters. The lecture will take place at 5:30 pm, in Room 320 of 7800 York Road.
Abstract: Often in today's world, for a given task, one finds a proliferation of data available, but the data available may not directly fit the task. Rather than finding new ways of measuring exactly what is needed, the innovative solution involves finding better ways to make use of the data already available. One motivating example of this is in weather prediction, where atmospheric measurements are sparsely available over the earth, but recorded extremely frequently in time. To accurately predict the course of a hurricane, a finer measurement resolution may be required, but the cost of building more weather stations to increase the resolution would be prohibitive. Fortunately, the extra resolution in time can be used (in conjunction with a physical model) to compensate.
In 2014, Azouani, Olson, and Titi successfully applied one such data assimilation technique (known as nudging) to the Navier-Stokes equation, for the case that measurement data is collected on the velocity of a moving fluid. Their method is general in its application, and has since been applied to several other dissipative systems. We will consider their technique applied to the Magnetohydrodynamic (MHD) equations, which govern plasmas and other electrically conductive fluids when coupled with an external magnetic field. We present some rigorous results, and demonstrate numerically the effectiveness of the nudging algorithm for the MHD equations. In addition, we show some recent work where we show for the Navier-Stokes equations that the nudging technique can be used to compensate when there is error in the estimate of the viscosity of the fluid, and can also be used as an indirect way of measuring the viscosity.
| Date | Topic | Speaker |
|---|---|---|
| Oct 18, 2018 | Generic maps over divisible ordered Abelian groups | Dr. Alfred Dolich, Kingsborough Community College (CUNY) |
| Oct 1, 2018 | The optimal strategy for hedge fund investing | Dr. John Chadam, University of Pittsburgh |
| Apr 13, 2018 | The mathematics of the finite element and its construction | Dr. Shangyou Zhang, University of Delaware |
| Apr 8, 2018 | Topological data analysis | Dr. Hal Schenck, Iowa State University |
| Apr 6, 2018 | Computational forensics for airplane crashes | Dr. Goong Chen, Texas A&M University |
| Feb 2, 2018 |
Liquidity premium of corporate bonds based on Merton's model | Dr. Xiaoping Min, Jiangxi University of Finance and Economics |
| Dec 8, 2017 | Site-specific recombination on circular DNA molecules and band surgery along the trefoil knot | Dr. Allison H. Moore, UC-Davis |
The Mathematics Seminar is the venue where Towson faculty and students report on their research activities. The Graduate Seminar hosts expository talks by faculty that introduce graduate students in the departmental APIM Master's program to topics that present opportunities for graduate research projects.
The next meeting of the Graduate Seminar will take place on Thursday, September 20, 2018. Dr. Mike O’Leary, Professor of Mathematics & Computer Science and Chair of the Department of Mathematics, will open the seminar series for this semester with a talk entitled Models for Offender Target Location Selection with Explicit Dependency Structures. The talk will take place in YR 320 from at 5:30 pm. All graduate students are strongly encouraged to attend.
Abstract: The geographic profiling problem is the one of estimating the home base of a serial criminal from the known crime site locations. One approach to the problem is to construct a mathematical model for offender behavior, and then estimate the home base by performing Bayesian analysis. There is evidence that shows that the distance between the offender’s home base and the crime sites can be well modeled by a Rayleigh distribution, but that the underlying two-dimensional distribution is not bivariate normal. In 2011-2012 I worked with a graduate student to develop models for offender behavior with explicit dependency structures. These models were tested for effectiveness against historical data from Baltimore county.