I'm using comsol semiconductor module to solve the drift diffusion equations for solar cells. I have noticed that converged solution is different when I'm using comsol multiphysics environment and using matlab livelink to solve the same model. What are possible reasons for this? Can a difference in initial solution cause this? Also, I would like to verify whether the converged solution is the correct one or not. What can I do for this?
Thanks.
In general, you should refine your spatial resolution to check if your mathematical model and the applied numerical methods converge towards a unique solution. After several steps of grid refinement you should get a solution that is independent of your mesh size. If this works in your case, you can assume that your solution is correct.
Unfortunately, I have no further experience with COMSOL and the livelink to MATLAB. So I can't tell you something more specific.
I think proposing sensible and reasonably accurate argument for observed discrepancy is something critical. It will be better to examine the literature for numerical/experimental data processing, analysis, comparison etc. DOI 10.1007/s00231-009-0478-9 may help.
- Badges
- Science topic