Apparatus of Molecular Simulation Experiments

About forty years ago, microprocessors became common, and many persons became interested in them. At that time, I wondered which is better: to produce computers or to use computers as my study and job. I chose to use computers in my scientific studies. A little after then to the present I have specialized in molecular simulations (molecular dynamics methods and Metropolis Monte Carlo methods) in computer chemistry. I think that I was one of the first persons who thought and practiced to use personal computers for computational chemistry study (Molecular Simulation using Personal Computer − Introduction to Molecular Dynamics Experiment -(Japanese), Kaibundo, 1990.)

The computers which we use routinely are called Neumann type. By the computer, the processing of A+B = C is carried out by the following processes: 1) loading numerical data into an accumulator in a CPU from the memory address A, 2) loading numerical data into a register in the CPU from the memory address B, 3) adding the numerical value of the register to the accumulator, and 4) saving the result of the accumulator to the memory address C. Repeating similar processes for even more complex problems. This type of computer is easy to use with deterministic physical simulation, such as molecular dynamics method, and the computer programs are also easy to develop. However, since enormous calculations are required sometimes, we often feel disatiss-faction with the capabilities of those computers. Because of that, kinds of dedicated purpose computers have been made. However, they were not widely accepted. Parallel processing is relied on currently for large-scale computations.

The quantum computer became the topic recently. It is said that the solution to certain problems can be found by predominantly fast processing when we use this. I cannot understand whether this kind of computer is usable for molecular simulation calculations easily even if I read a commentary book. Is it to have rewritten a program substantially, or to be able to reorganize algorithms? Or we cannot use quantum computers for molecular simulations. We may use them for a part of Monte Carlo methods. Will the computing environment as “apparatus” of molecular simulations change substantially the near future?