Meet the Researchers: Evidence and facts

Interview with Professor Harvey Gould
Physicist Harvey Gould uses computer simulations to understand the behavior of atoms and molecules in a variety of contexts, especially those that are difficult to study using traditional experimental methods. Below is a summary of a recent interview in which he discussed his research interests, and the role that computer simulation plays in the study of physics.

How did you become interested in studying physics?

I've always wanted to understand how things work, and I did a lot of reading in this area when I was young.

One incident sticks out in my mind that relates to why I decided to go into science in general. In the 8th grade we had to do an oral book report, and I decided to do it on population control. That was a very controversial topic in those days, although I wasn't really aware of it at the time.

After giving my report, a lot of kids teased me--I was the birth control kid! When I spoke after school with the teacher, he said that he agreed with everything that I had said about the need for population control, but he was Catholic--end of discussion!

That's when I decided that I wanted to go into a subject where people discussed things based on evidence and facts. I didn't decide to study physics specifically until after my first year of college.

Most scientists tend to make up their mind to study science relatively young, which is unfortunate because once students start college, it is difficult to convince them to study science.

Can you describe briefly your major areas of interest?

I'm interested in condensed matter physics, which concerns anything you can touch--solids, liquids, gases, and polymers, for example. In particular, I'm interested in modeling and simulating these systems on the computer. I work on a variety of systems from magnetism to liquids to earthquakes.

What do you simulate? Is it interactions between molecules that make up a substance?

Yes. If I wanted to study a liquid, I'd simulate the motion of its molecules. I'm more interested in the generic properties of liquids, rather than the properties of a specific liquid. This interest lets me make a model of a liquid that's relatively simple. For example, I can assume that the molecules are spherical in shape and make other kinds of simplifications. Then, since all the molecules interact with one another, the simulation can determine the effect of the interactions in various ways. One way is by solving Newton's equations of motion for each particle, calculating the force on each particle, its acceleration, velocity, and position. The simulation can repeat these steps over and over again.

So you try to simulate the behavior of liquids in certain contexts?

Yes. I'm particularly interested in the behavior of liquids at temperatures below the freezing point. For example, if the temperature of water drops below freezing, it will become ice. But if you look at water outside on a cold winter day when the air is below freezing, the water won't be frozen; instead, it will be in a supercooled state. In general, when the temperature gets low enough, a liquid can, under certain conditions, become a glass. To form a glass, the liquid needs to be cooled very quickly so that it doesn't have time to become a solid. You couldn't do that with water, but you could with honey if you put it in the refrigerator for a while. That's because honey is a very viscous liquid.

So is glass something between a liquid and a solid?

The kind of glass we usually think of, like window glass, is a special kind of glass. I use computer simulation to study much simpler glasses, because I'm interested in the general properties of glasses. The structure and characteristics of glasses are not well understood. So we're looking at a simpler, related problem: what's going on in a liquid right above the temperature at which it becomes a glass--the glass transition. People are still not sure what the glass transition is or even if it can be well defined.

Can you comment on theory, experimentation, and computer simulation as three complementary approaches to studying physics?

Physics is an empirical science--it's based on reality. Our job as physicists is to describe nature. So experiment is always the final arbiter. We devise theories and run simulations to make sense out of what we observe. Even though there are these three ways of studying physics, experimentation is the most important, and I say that as someone who is a theoretical physicist. What we actually observe takes priority over what we see with a computer simulation or what a theory may suggest. A successful theory has to fit the experimental results and predict other results.

That understood, there are ways that simulations can be particularly helpful. Suppose I take a beaker of supercooled liquid, or of a supersaturated solution of sugar in water. If I tap the edge of the beaker, I create a little disturbance in the liquid so that the density of the liquid is a bit higher in some regions. That will start a process called nucleation: in some region of the liquid, a little solid nucleus can appear. If it gets to be a sufficient size--the critical nucleus--it will keep growing and the system will solidify. The critical nucleus is so small--only about 20-50 molecules--that it can't be observed in an experiment. And nucleation also happens relatively quickly. So this process is a great candidate for simulation. I can now observe something that with an experiment I could observe only very indirectly. (Although, just within the past year, some scientists have been able, using a microscope, to observe nucleation in a specific material that has big molecules.)

Is the scope of possible simulations limited by currently available computer power?

It depends on the complexity of the system, but yes, there are many limitations. For example, if I wanted to simulate a liquid, the fastest computers available would let me simulate a total of one millionth of a second. But as computers become more powerful, and as we develop better models and more efficient algorithms, we'll be able to simulate more complex systems.

Introduction to Computer Simulation Methods: Applications to Physical Systems. What kind of feedback do you get from students about their experience with computer simulation?

Some students respond that simulation is a much better way of doing things. I immediately correct them, because simulation is not a better way, it's another way. The course I teach at Clark, based on that textbook, encourages students to do simulation projects that are very much like doing real research. Most students like that, but not all, because the questions are very open-ended--the answers aren't in the back of the book.

Simulation is like doing an experiment, but on a computer. And like an experiment, you don't have to know all the details and understand everything in order to get started with simulation. Your understanding increases with the results you get. That's one reason I started doing simulations. When I first came to Clark I only did theory using pencil and paper. About 10 years later I started doing simulations, in part so I could involve students more easily, both graduate and undergraduate.

Simulation allows undergrads to think about systems they wouldn't be able to imagine with their current level of experience. In this class, students can also simulate systems that are not physics-related. Some students choose to model social systems, biological systems, problems in chemistry, etc.

The challenge to the student is to take a physical system and write a program that simulates it. Getting an algorithm (a series of instructions for the computer) to work and getting answers is not a trivial task. What I've found is that those students who excel are also the students who excel at research in general. It takes creativity and a willingness to go forward even when you don't know what the answer will be. But generally, even first-year students who have that attitude can keep pace with upper-level students.

What are the opportunities for undergraduates to participate in physics research?

In physics we mostly conduct research in groups, and students work in a group, under someone's direction. It's a very efficient approach. People think of scientists as working alone and isolated in a lab, but that's not how it is in the real world. We use an apprenticeship model, and we actually have a course called research apprenticeship.

Our department prides itself on involving students in research as soon as they wish to. There are plenty of opportunities. If we could convince all first-year students to be involved in research, that would be great!

I tend not to have as many students working with me as the experimental groups do. When a student indicates an interest in working with me, I ask if he or she has already worked in an experimental group, because it's most important for students to get as much hands-on experience in the lab as possible. But if someone persists, of course I'm happy to take that person on. And anyone from a first-year student on can do meaningful work, if they're motivated. I've had a number of students who have published papers with me, and I'd love to have more.