Is a refactor of the SVF computation module welcome?

[kwtk86]

Hi, previously I worked on using UMEP to calculate SVF and found it very slow in city-scale calculation. I'm considering rewriting the SVF computation module in svfworker.py with C++ for performance reasons, while keeping it accessible from the existing Python library. AI coding tools will assist the process (I have relevant development experience and will carefully review all generated code before submission).

Before investing time in this, I'd like to check: Is this direction something the maintainers would accept?

Happy to discuss further if there's interest.

Thank you.

[nilswallenberg]

Hi,

You are of course free to develop this. There is some related work on this. Below are two other examples where the SVF calculations are executed on GPU.

https://doi.org/10.21105/joss.09535

https://doi.org/10.1007/s00704-021-03692-z

Best regards,
Nils

[kwtk86]

Hi,

You are of course free to develop this. There is some related work on this. Below are two other examples where the SVF calculations are executed on GPU.

https://doi.org/10.21105/joss.09535

https://doi.org/10.1007/s00704-021-03692-z

Best regards, Nils

Thanks, I will take optimizing the SVF calculation into consideration