Dynamically allocated arrays in the fortran PDD solver

[jbuencuerpo]

In the wiki it is said that the absolute limit is 6000. If the structure has too many layers, the code fails silently. Why 6000 is the limit?

[dalonsoa]

In the Fortran code, the maximum size of the arrays is hardcoded - and set to 6000 -. Originally, when it was done 5 years ago, it was just a question of convenience and of simplifying the code, but it can be done better and passed that value dynamically.

As an intermediate solution, I would suggest you play with the mesh size and growth rate in the solver options to try to keep the total number of mesh points below 6000.

Keeping aside that this can be changed, 6000 mesh points is really a huge number for a 1D calculation. I've simulated solar cells with [up to 100 strain balanced QWs](url https://doi.org/10.1109/PVSC.2014.6925530) and didn't reach that value. I'm curious, what are you trying to model?

[jbuencuerpo]

I was playing with the tutorial for the tandem with QWs, no real structures, and the mesh size. The code fails silently using for example a jupyter notebook, and there is no Warning in the python code. I think it will be helpful to raise an Error in python.

I am not literate in Fortran, I have been able to increase the hardcode number to 8000 in the fortran code, but a dynamic upper number I think is desirable.

[dalonsoa]

I completely agree. I've changed the title of the issue and labelled it as enhancement.

[jmllorens]

I have started to implement the dynamic version of pdd. This is the branch in my fork: https://github.com/jmllorens/solcore5/tree/dynamic_dd

[jbuencuerpo]

I have created this small example for a GaAs SJ. Increasing the meshpoints crash it silently.

https://gist.github.com/jbuencuerpo/436d0791a931a0e727b1921c841a2c98