Convergence with high throughput

If we want to run different simulations with different convergence parameters in a sequential way, then we have to guess the total runtime of the slurm job. Because the runtime of simulations with different convergence parameters are hard to predic, assuming we allocate the same resources, the total runtime of a job like this is also very hard to predict. If we want to create slurm array jobs to run them in a parallel manner we face the same issue, that it will either waste a lot of resources or run out of time and terminate with an error. For this purpose automatic convergence tests are not supported in grogupy, however there is a visualization tool that can help to determine the convergence of a system with respect to some parameter. For further information see the API reference.