Dear All, I am trying to do a large computation using current worm code. What one can infer from recent exchange of emails of myself and M. Troyer is that I encounter some stability issues. I have N inputs (one input data set each) configurations for which I would like to do MC calculation and in case of a failed I would like to kill the task automatically and restart it with a different seed.
I would like to do the followng: redirect worm output to a file. If file is not written let's say in several (e.g. 15) minutes the task is considered frozen and automatically killed and requeued with a new seed.
I would like to ask if in the worm code the output
Checking if Simulation 1 is finished: not yet, next check in 60 seconds ( 0% done). really implies there will be output within (approximately) next 60 seconds? Is an appearance of the "avoided problem" an important issue here?
Regards, Mateusz Łącki