Bad Computational Habits


  • Not reading software manuals and e-mail lists. Mark Abraham and Christopher Stilesput http://www.gromacs.org/Documentation/How-tos/Carbon_Nanotube a few very good descriptions on how to read manuals and asking questions on the Gromacs web site. I provide one of their masterpieces below. Please, read this. Although this is about modeling CNTs the most of the text contains very general recommendations that can be useful for almost any computational chemist - just substitute Gromacs by Software X and CNT by object Y (as an exercise, find other recommendations by Mark and Chris by yourself, please).

  • Using VMD for converting Gromacs trajectories to PDB or XYZ etc. This makes no sense because i) Gromacs has very efficient 'built in' tools to do that. ii) Using VMD for that one cannot be sure that there are no mistakes etc.

  • Using SCRIPT languages like Python, Perl, Ruby etc for doing sophisticated scientific calculations. These languages WERE NOT DESIGNED for that. They are good tools for extracting information, organsing work-flows, etc but NOT for scientific calculations. Please, use Matlab, R, C/C++ or Fortran for doing something more complicated than X+Y. These languages WERE DESIGNED for scientific calculations. I note that I would not allow to publish results of any sophisticated calculations done by scripting languages.