Reduced runtime for PintRun

[JensHahne]

Hi @tlunet,

just a quick update: We talked a while back about two ways to speed up the simulations:

1. use larger (in terms of N and K) PintRuns instead of recalculating other settings.
2. using a lookup table for some "generic" versions.

I implemented both, with the functions it can be much faster now. In the end it's not very complicated, you can just reduce the number of block rules in a PintRun call, and thus create reduced PintRuns from another PintRun. The lookup works the same way, I computed some default settings for PintRuns (N=100, K=10) that can be loaded when the same update rule is used for new experiments. The costs etc are independent of this, so we just use the block rules from existing runs instead of recomputing them.

It will automatically check if there is a corresponding lookup file. There is an example in the script 05_speedupPFASST.py how to easily use a larger run without lookup to not recompute things

If you have any questions or problems, just let me know