Since we cannot solve the Schroedinger equation analytically once we go beyond hydrogen-like atoms, approximations are needed. Now we consider atomic scale, Monte Carlo or Molecular Dynamics with classical potentials. This is a molecular dynamics code - we discussed Molecular Dynamics earlier , but ony from an algorithmic viewpoint, not as a code implementation. There are two questions to ask here:
The reasons for canned code are several. Own code, or code built up by successive group members is more idealistic. There are not the serious computational issues, such as large matrix inversion which limit DFT applications. In fact several group computations are using our ``own code''. However, good molecular dynamics studies need good statitical physics, and implementation of complicated temperature and pressure ensembles is HARD!!! Also to get large systems, have to do clever parallelization and this is HARD!! So a canned code can simplify these aspects. For smaller systems, with straightforward statistical mechanics, and known or complicated potential choices ``own code'' is preferable, both ideally and in practice.
There are many project examples already done in this course, and we will skim over these.
korens_at_TECHUNIX.TECHNION.AC.IL Koren Shreiber
bbliron_at_TECHUNIX.TECHNION.AC.IL Liron Ben-Bassat