CompositeMeasure integration not working properly in some cases

[JordiManyer]

PR for issue #8

[JordiManyer]

This now works properly. We now get the same number of iterations with single processor runs and multi-processor runs.

[amartinhuertas]

Fantastic! How much the number of iterations degrade with the number of levels for the optimal setting of the parameters? (Whatever this means)

[JordiManyer]

Not much. We go from 17/18 iterations with 2 levels to 20 with 4/5 levels (in both single and multi processor runs).

[amartinhuertas]

Can you let me know the setup of the algorithm to achieve that? I mean smoothers, number of smoothing steps per iteration etc.

[JordiManyer]

It's this script: https://github.com/JordiManyer/GridapSolversBenchmark.jl/blob/e22c31d638f30a3a3577da5e22b22b8fb24b3a29/experiments/GMGSingleProc/run_experiment.jl

I have it run with 1 and 4 procs and get the same results (and I did not before).