Dissemination of methods and software complexity

There exist many programs for MD of biomolecules, cf. Section 2.1.1. These programs have several features that make them robust production codes. However, several of these codes are legacy codes, poorly organized, or extremely complicated because of their high degree of optimization. They are inappropriate to serve as an algorithm development platform, and thus it is difficult to test novel algorithms in them. This situation has a negative effect on the training of students and junior researchers, who have to expend a significant amount of time dealing with these difficulties.

Also, few or none of the academic programs are easy to use. This intimidates potential users, who have to learn numerous implementation details of the simulation methods to be able to prepare all the input files and analyze the results. In particular, the more sophisticated algorithms have multiple parameters that interact in subtle ways. We propose to disseminate our results in software that automatically detects appropriate algorithmic parameters, and that may be used as a stand-alone program or in collaboration with existing MD programs, cf. Section 2.3. This will facilitate incorporation of successful algorithms resulting from this research into the most popular MD programs.