Erik Deumens

(picture)

News in spring of 2012

  • Service UF Research Computing and SSERCA activties
  • Research Design new software for describing molecular and materials processes using domain specific programming languages that allow rapid creation of software that scales to tens of thousands processors.
  • Teaching Overview lecture on the state of Quantum Chemistry can be viewed any time.

Scientist

Service

  • I am a physicist, mathematician, computational scientist and computer engineer.
  • Details of my activities and list of publications can be found in my CV.
  • In April 2011, the Provost and VP for Research and the VP and CIO committed to support Research Computing at UF for the next 5 years, and I was asked to act as full-time director of this effort. Research Computing acquired new staff members and more resources, including funds for a matching program. The activities are described in the minutes of the Research Computing Advisory Committee and its subcommittees. Updates and news can be found at the new UFIT News site.
  • Together with Frank Harris, I organized an exploratory workshop with an NSF grant into the organization of a Scientific Software Innovation Institute for the Atomistic Modeling and simulation community. The full list of workshops and their reports can be found at http://www.acsu.buffalo.edu/~abani/softwr2 The report for the one we organized is at S2I2Report.pdf.
  • Since May 1, 2010, I am director of the UF Research Computing.
  • I also act as consultant on computer and computational projects, including building power and cooling in server rooms, at UF and for other universities and organizations.

Teaching

Research

  • At the 2009 MERCURY conference at Hamilton College in Clinton, NY, I gave a talk on The state of Computational Quantum Chemistry with two movies made by Olivier Quinet of a proton scattering on the water dimer: impact and breakup. The lecture was recorded in MP4 format and can be viewed e.g. with QuickTime. The first 8 minutes are the introduction to the conference. The head visible at the bottom of the screen is that of Prof. Roald Hoffman.
  • With Beverly Sanders of the UF CISE department and Victor Lotrich, Mark Ponton, Ajith Perera, and Rod Bartlet we are developing a domain-specific-language (DSL) based approach to massively parallel programming of complex software systems for molecular and materials processes. Funding comes from DOE and NSF, as well as DOD Army and Air Force.
  • In the February 2009, Victor Lotrich and I taught an introductory workshop for programming in SIAL (super instruction assembly language), which is the productive language to write software designed with the Super Instruction Architecture (SIA). I taught a 5 lecture course on SIA and SIAL at the Summer school on HPC in Chemistry at the University of Tennessee, Knoxville, Aug 4-7, 2009: Architecture and language definitions, Workings and performance, Algorithms
  • Since July 2004, I have been working on a book on Quantum Mechanics that presents a cohenerent view of quantum mechanics and quantum field theory, as well as statistcal quantum mechanics.
  • In the Fall 2004 I gave a series of lectures High Performance Computing Topics. It discusses all issues involved in programming for scientific computing: including architecture of modern CPU's and parallel computers, object oriented design, correct programming (Fortran 95 is used as example language), debugging and performance analysis, message passing programming, and thread programing.
  • This material has been updated since the previous courses taught in Summer of 2001 and Spring of 2000 on parallel and advanced programming.
  • Since 1994, I use the development of a high performance, portable, parallel software library for quantum chemical integrals, called QTIP as testing ground for research and teaching of high quality software engineering.
  • Since 2003 I have been working as software architect and project manager with Rod Bartlett and ACES Q.C.in the design of the super instruction architecture (SIA) for scalable parallel software development. This environment has been used to write ACES III, a parallel implementation of Coupled Cluster methods. During the Summer of 2009, we managed to run a CCSD(T) calculation in the molecule RDX with ACES III on 60,000 cores of the Cray XT5 Jaguarpf of the DOE facility at Oak Ridge, Tennessee. During the Spring of 2010, we got up to 86,000 cores with a new SIAL program for building the Fock matrix.
  • Since 1986, Yngve Öhrn, and several postdocs and graduate students and I have developed the theory of the Electron Nuclear Dynamics (END). I am the principle author of the software package ENDyne implementing this theory. The effort from 2005 until now has been to extend END to multi-configuration wave function. We call this dynamic MC wave function method vector Hartree-Fock (VHF). The code was completed and debugged just before Christmas 2008. We are now starting to run the first VHF calculations on atoms and some simple molecules to test the theory and the algorithms.
  • Since 2007, I am working with Beverly Sanders to investigate applications of the Super Instruction Architecture to other domains than computational chemistry.
  • During 2004 and 2005, John Klauder and I worked a project to investigate self interacting field theories using lattice Monte-Carlo computations.

Email address: deumens at qtp.ufl.edu
Office: 2334 New Physics Building
P.O. 118435 University of Florida Gainesville, Fl 32611-8435
Phone: (352)392-6980

Last modified 31 May 2013