- Martin Berzins
- Professor
- MEB 3190
- Phone: 801-585-1545
- mb 'at' cs.utah.edu
Teaching Spring 2010
A complete list of papers is at Publications
My NSF parallel computing funding is:Funding
Member of the Scientific Computing and Imaging Institute
Visiting Professor at University of Leeds
Some of my publications from 1990 related to my work at Leeds are at Papers
My EPSRC funding while I was at Leeds is at Funding
Co-Editor in Chief of Applied Numerical Mathematics APNUM
Member of edtorial board of International Journal for Numerical Methods in Fluids IJNMF
Member of edtorial board of Concurrency and Computation CandC
Background
I came to the School of Computing in 2003 from the University of Leeds in the UK where I was a Lecturer, Senior Lecturer, Reader and Professor in the School of Computing at Leeds from 1984 until 2003 and Research Dean for Engineering, from 2001-2003. My PhD was in Computer Science at Leeds in 1981. My research is in the fields of computational science and mathematical software, numerical analysis, parallel computing, problem solving environments and grid computing. Much of this work has centered around solving challenging applications problems in computational fluid dynamics, combustion, atmospheric modeling and lubrication modeling and in the writing of software for the solution of such problems on parallel computers.
These areas involve two important classes of partial differential equations - parabolic and hyperbolic systems of equations, the solutions to which depend on both space and time. The new algorithms and the associated software have resulted have then been used as part of successful interdisciplinary academic collaborations and, through the Computational PDEs Unit at Leeds (CPDE Unit), with industry, most notably Shell Research (now Shell Global Solutions). This activity has continued to grow and I still collaborate with colleagues at Leeds through my visiting professorship.
Research Interests
My research area is the study of novel computational algorithms for the numerical solution of partial differential equations (p.d.e.s). The approach I've has taken in this research has been to derive numerical methods and develop software on both serial and parallel computers for a broad, mathematically-defined problem class in which the equations are decoupled in space and time. This has made it possible for users from different physical applications areas to solve their problems by creating a mathematical model which fits inside the general problem class.
My recent past work has been part of the CSAFE DOE funded center CSAFE
The key aspects of this work in CSAFE have been connevcted to thie Uintah code and the methods it uses, e.g. to
- Help develop a better understanding of the MPM (material point method) as used in Uintah
- Help improve the ICE fluid-flow method used in Uintah
- Help devise new parallel adaptive mesh algorithms to control the computational error
- Work on approaches for helping the Uintah code to scale better
Following the end of CSAFE NSF funding has helped to get Uintah released and improved Uintah Release
Uintah now scales to 98K cores on the NSF Kraken computer.
Teaching
My firm belief is that it is the close coupling between teaching and learning and research that is one of the defining characteristics of a research-led University. Given this view it is perhaps not surprising that as a result of twenty five years of teaching Scientific Computing and Computer Science both Science and Engineering students I have evolved a teaching philosophy based on incorporating research ideas into course content.
Ph.D Students: An important part of this teaching effort has been directed towards the training of fiteen or so Ph.D students. The success of this has been reflected by the their successful research and their Ph.D. degrees. Current students are:
