Closed babreu-ncsa closed 1 year ago
The script is actually here: https://github.com/FZJ-PGI-12/QAOA.jl/blob/master/notebooks/benchmark.ipynb
I will again make sure it's working out of the box now.
@babreu-ncsa The notebook has been updated here: https://github.com/FZJ-PGI-12/QAOA.jl/blob/29-add-benchmark-scripts-for-reproducibility/notebooks/benchmark.ipynb
It should now be in a state where you just have to do Run all
to reproduce the plot. Note, however, that the case with p=5
and n=20
will take something like 3 hours for PennyLane (on a M1 MacBook Pro, that is). So you might want to change the nrange
to something more economical.
Please let me know if you get results which are roughly in the same ball park. I got slightly different ones on my new machine with a different Julia version, which is to be expected though, I guess. If you give me green light, I'll do the merge.
Thanks!
I'll do the merge anyway now, should be fine.
@timbode thank you, it looks good to me!
The Benchmark section of README.md shows a performance comparison between QAOA.jl and PennyLane. Although the data is in the assets folder, having the PennyLane script would be essential for direct reproducibility. Could you please add the script that was used for benchmarking? https://github.com/openjournals/joss-reviews/issues/5364