Spring 2008
In This Issue:
Materials Modeling and Simulation
The Materials Simulation Center is housed in a suite of rooms in the Physics Department at University Park. Its current director Jorge Sofo is associate professor of physics and associate professor of materials science and engineering. Associate director Ping Lin, who arrived at Penn State this past summer, received his Ph.D. from Texas A&M and did postdoctoral work at UNC Chapel Hill.
The Materials Simulation Center began in 2001 with a grant from DARPA, the Department of Defense’s futuristic research group. Penn State theoretical physicist Vin Crespi wanted to create a computer cluster where the materials community could integrate computation and simulation into their materials research. He brought Sofo to Penn State as his co-director.
Hosting a computer cluster turned out to be a time consuming struggle for the two, Sofo admits. When they discovered that a group already existed on campus that was expert at building and running “big machines,” they turned over their computer cluster to the High Performance Computing group (now Research Computing and Cyberinfrastructure), part of Penn State’s Information Technology Services. RCC was looking for faculty to use their services and help purchase more equipment; the Materials Simulation Center had plenty of faculty looking for computational expertise. The MSC became the interface between faculty and RCC’s big computers.
“I advertise the RCC to new faculty who might have money to set up a computer lab,” says Sofo. “They might be interested in having access to more computing power. And then we are in charge of the software.”
Advances in Software Design
The software programs are at the heart of computational materials design and materials simulation. Until the 1960s, the calculations required to apply the laws of physics and chemistry to understanding the properties of molecules and their interactions were too complex to solve. Then in 1964, Walter Kohn of the University of California at Santa Barbara developed a method, called density functional theory, which simplified the calculations for electronic structure. This made it feasible to begin to model a single atom. In the 1990s, the development of faster computers made it possible for the first time for scientists to study larger, more complex molecules. Today the power of supercomputers and sophisticated software programs makes it possible to model large molecules of several hundred atoms.
Some software is created by university researchers and some is commercial, Sofo says. Open Source software is free to anyone who is willing to share their own improvements on the code. “The commercial packages usually have a more sophisticated user interface, while the free code is just as powerful, but tends to require more expertise. We also create our own code, in the same way that experimentalists will build a specialized piece of equipment for their lab experiments,” Sofo explains.
Working with MSC
As Sofo spends more of his time on research and teaching, Associate Director Lin is taking on the day-to-day responsibilities of working with faculty and students in the MSC. When requests come in from experimentalists, he will help them determine what kind of software tools and computer hardware they might need to get the information they require.
“There are several ways we can work,” says Lin. “We might do some of the calculations ourselves, or we can recommend software. We might have to do some literature reviews to see if there are programs that do those sorts of calculations. Other times we train students to learn the software and do the calculations themselves. Beyond that, Jorge Sofo works with many of the experts on campus and around the world who can be called on for help.”
In a later interview Sofo adds, “Ping has started a series of short courses. Last semester it was on the use of Gaussian (one of the programs developed by John Pople, who along with Kohn shared the 1998 Nobel Prize for Chemistry). We had 16 grad students. There were 15 in the next course. The idea is to have a set of courses every semester.”
So far, these courses are free, supported by the Materials Research Institute, which oversees the Materials Simulation Center and offers its services at no charge to the Penn State research community. Every other year, Penn State and MSC host the Wien2K workshop, centered on a scientific code with more than 1000 users worldwide. Penn State grad students participate with 40 researchers from the U.S. and Europe who have come to learn how to use the code from the authors who wrote it.
The MSC maintains a library of materials simulation codes and visualization tools, software packages that turn data into images. Lists of available codes along with brief explanations are available on the MSC Web site.
The MSC is also in the process of building a materials simulation portal on their site. The portal would allow a user to log in using a Web browser from any place in the world. Through the portal they will be able to start or stop a job and see real time results of their experiment. “We’re working with the RCC group in this and hope to have beta testers by the end of the spring semester,” Sofo remarks. He believes that with some restrictions small and mid-size companies might be able to take advantage of the portal to access high speed computing at Penn State.
Asked about how he settled on computation and simulation, Lin says, “I guess when you do your studies you get the chance to do a lot of things and it just turned out I found this theoretical and computational part more interesting to me. You have the power to predict and you can use your imagination. You can use your theoretical background to do things right, and when you find that your predictions and calculations can match up with experiments, and simulations can be used to interpret many experimental observations, it can be a very powerful and useful thing.”
Contact:
Prof. Sofo and Dr. Lin can be contacted through The Materials Simulation Center, www.msc.psu.edu.
