What is Molecular Simulation?

Molecular simulation describes a collection of methods for doing computer "experiments" on model molecules.

What is a "model molecule"?

Molecules are modeled by postulating (quantitatively) how they interact with one another. This entails specifying the force that they feel due to each other's presence, for all possible ways of arranging them in space. It is known how to compute these forces exactly, but doing these calculations takes too long to be useful. Instead researchers approximate the forces using relatively simple, rapidly computed functions. Thus we construct "model molecules".

What is a computer experiment, and what does it measure?

When many real molecules are placed together to form the materials that surround us, they exhibit interesting collective behavior which can be quantified and measured. Some of the more familiar behaviors are described by properties such as the boiling point, the heat capacity, the pressure, and the viscosity, to name a few. Values for all of these thermodynamic properties follow from the molecular interactions, and all the equations needed to evaluate them from a given molecular model are known. The problem is that these equations are too difficult to solve in a simple and accurate form. Instead computers can be applied as part of a numerical solution. These calculations have the character of an experiment performed on the system of model molecules: the calculation does not produce an exact result (as one is accustomed from computers when applied to simple calculations), but instead the outcome is subject to some degree of "experimental error" which can be estimated and controlled.

Why do this?

To better understand Nature.

Engineers need to predict how materials will behave; this ability is integral to the design process (which in turn is integral to efficient, safe, and environmentally friendly manufacturing). To perform these predictions engineers rely on relatively simple models and equations that provide a result very quickly. Speed is important because these predictions are made many times during the design process. However is very hard to come up with equations that are both fast and effective.

Advances in this art are now made because we understand better than ever before how the behavior of materials builds from phenomena occurring at the molecular level. This understanding has arrived through decades of effort by skilled researchers from a vast array of fields. The foundation was built by inspired theorists and careful experimentalists, and now patient simulationists contribute too. Simulation bridges the gap between theory and experiment in ways that were previously impossible. As a result models useful to engineers can be constructed in a much more methodical way (although it is still an art).

Molecular simulation is nearing a stage where it can serve as a direct supplement to, and sometimes a surrogate for, real experiments. Molecular simulation has limitations, and it will never replace experiment; indeed it would be reckless to advocate such a step. Nevertheless as simulation algorithms continue to improve and advances in computing hardware continue to astonish us, simulation will take on more of the duties of experiment, in the same sense that engineering models do, but more reliably. Additionally with simulation comes an infinitely powerful microscope, one which permits us to see, measure and manipulate the smallest detail in ways that will be forever beyond the abilities of experiment. Advances in leading-edge technologies such as materials or bioengineering are made at this molecular scale. Simulation's ability to probe events at this level positions it as major contributor to the growth of these and other important industries.

Back to the home page.

David Kofke <kofke@eng.buffalo.edu>