Meet the Editors: Bruce Boghosian

Meet Bruce M. Boghosian, PhD
Professor, Department of Mathematics
Bromfield-Pearson Hall, Tufts University

  • In what “slice” of CiSE do you work?

I began my career in computational plasma physics at Lawrence Livermore National Laboratory. A move to industry at Thinking Machines Corporation broadened my interest in HPC, particularly as applied to fluid turbulence and quantum Monte Carlo. From there I moved to academia as a member of the faculty at Boston University’s Center for Computational Science, where I continued work on the dynamics of complex fluids. Finally, in 2000 I joined the faculty of Mathematics at Tufts University where I continue research on computational and theoretical fluid dynamics. Most recently, I have become interested in applications of kinetic theory to economics.

  • What sorts of changes have you seen in the field over the years?

In the 1970s and early 1980s, HPC was driven by scientific applications. Supercomputer manufacturers used to visit computational science working groups, for example at LLNL, to learn more about our needs and to help us use their hardware more effectively. Beginning in the late 1980s, HPC was driven by commercial applications, especially video games. While this may seem like a step backwards from a scientific standpoint, the seemingly inexorable march of Moore’s Law continues to make computational science a very exciting area of research. My cell phone is now more powerful than some of the computers that I worked on at LLNL in my early career. GPUs designed for video games are excellent SIMD parallel computers that excel at highly parallel applications, such as QCD and lattice Boltzmann models of fluids. Programming environments such as PETSc and Trilinos have made parallel scientific computing simple and transparent. The ability to mine large scientific databases will be an important development in the coming years. It is still an exciting space in which to work and I feel privileged to be part of it.

  • What is the most exciting aspect about your work for the near future? The far future?

As noted above, I have become very interested in applications of kinetic theory to economics. I hope to focus on that in the coming years.

  • If you were to explain Computing in Science and Engineering (either the magazine or the field(s) it represents) to a five-year-old, what would you say?

I would say: Scientists can describe the way that many things in nature behave. They know how fluids move, why the sky is blue and sunsets are red, and why you look a bit like your parents. The descriptions that scientists use, however, often involve lots of mathematical calculations. Most of the time, these calculations are just the kinds of things you are learning in school now—adding and multiplying numbers—but they have to do it huge numbers of times to describe real systems. So they use very powerful computers to do this.

  • Big Data… What’s more exciting or important (or is there anything more important)?

Big data is very exciting indeed. I just finished reading “Data Science for Business” by Provost and Fawcett, and enjoyed it very much. It made me realize that a good fraction of what is now called “data science” is material already well known to computational scientists. High-energy physicists were among the world’s first large-scale data miners, after all, and most of the methods used by data miners are either straight from statistics or numerical analysis textbooks, or they involve learning algorithms, such as neural networks and genetic algorithms. So computational scientists reading new books on data science will inevitably feel a bit like Molière’s Monsieur Jourdain, who was delighted to find that he had been speaking in prose all his life.

  • What is one thing that would fundamentally change the average person’s reality if he or she worked with you day to day and saw what you saw?

At the moment, I am taking a leave of absence from Tufts University to serve as the president of the American University of Armenia—an American-accredited affiliate of the University of California, located in Yerevan, Armenia. In fact, we have just initiated a new bachelor’s degree in computational science there, with an innovative curriculum that combines applied mathematics, computer science and numerical analysis.

American schools and hospitals in the developing world spread hope for economic growth and political reform far more effectively than any other US foreign policy endeavor, and I have directly seen the transformative effect of an AUA education on the lives of our graduates. So I think that most Americans would be amazed to learn that US funding for all American schools and hospitals abroad totals only $23.5 million per year — less than 1% of the cost of a single B2 bomber.

  • What is the most important application of HPC/computational science/data visualization in your opinion? (Protein simulation, climate modeling, etc.) Why?

We need to use HPC to develop and implement sustainable practices. We need sustainable engineering, sustainable economic policy, and sustainable business practices. Instead of the incessant drumbeat for growth, we should be focusing on closed-loop systems that do minimal damage to the environment, the climate and the economy. Hopefully, scientific modeling will allow us to develop alternative energy sources, economic modeling will guide us to better public policy, and big data will allow us to evaluate business practices not simply by the profit generated at the end of each quarter but also by their externalities and social costs.

  • Conversely, what is the scariest?

At this moment, I think that the scariest application of HPC is the use of data mining to create a pervasive surveillance state in which nothing is private. Of course, we know about that only because of a single whistleblower. I am confident that there are even scarier things going on of which we are unaware.

  • Why do you do what you do?

Because it’s fun and it’s useful.

  • Anything else you want to add?

To those contemplating a career in this field: Learn as much of the mathematics and the science as you can before you jump on the computer. The best mathematical modelers are people who knew real, complex and functional analysis well before they tried to learn numerical analysis. Likewise, if your goal is to model sunspots, you should spend quite a bit of time learning about them, and all the open problems associated with them, before you try to simulate them on a computer. Computer simulation, after all, ought not to be an end in itself; just as is the case with theory and experiment; it ought to be the means to an end.

Tagged with: , ,
Posted in Meet the Editorial Board, Uncategorized