Common Modeling Mistakes

Common Mistakes in Modeling

1. Not knowing what question(s) you are trying to answer with your calculations. You should know what you are trying to answer, since this will determine what calculations you should run.
2. Not understanding input/output files. Don't treat the calculations as a black box. Try to understand what's going on. This can really help you trouble-shoot and get good results.
3. Not being consistent in calculations (same settings, code version, computer throughout, etc.). Sometimes different versions of the code or different parameters give different outputs. You should use be consistent in your calculations for a given project.
4. Not optimizing your calculation timings (number of nodes, cores, input settings, etc.). Doing some benchmark tests could save lots of time in the long run. Pay attention to speed and the parameters you use.
5. Not starting with a good initial geometry (e.g. bond distances too long/too short or angles unreasonable). Garbage in = garbage out.
6. Not examining output/geometries/results: are they reasonable/good? Don't just trust all results or numbers you get from the simulation. Make sure the results make sense and are reasonable.
7. Not running multiple geometries/initial configurations to verify that results are the “true” minimum. Just because your calculation converged doesn't mean you have the most stable geometry or global minimum. You may be at a local minimum.
8. Convergence problems: not understanding why your job isn’t converging (electronic problems? Geometry problems?). See more here.
9. Not understanding theory behind calculations. Again, don't treat the calculations as a black box. The more you understand what's going on with the calculations, the easier it will be to interpret the results and troubleshoot any problems.
10. Not developing computer skills (linux, bash scripts, python). Doing things by hand (or cut/paste into Excel) can often get tedious and boring. Taking an hour to learn to write a script could save hours of analysis or running jobs.
11. Not understanding/managing the job queue properly. Learn to use the queue so you can get maximize your chances of jobs running.
12. Not looking for answers. Learn to be independent. This is an important skill that will help you throughout your career, and will also help your research move forward.
13. Not asking for help. Don't wait too long to ask for help. Be independent, but don't get stuck. The internet, your colleagues, your advisor are all resources for help.