To improve performance and utilize embarrassingly parallel nature of Monte Carlo simulations we should perform all tasks we can on compute nodes, rather than on server.
This task would require:
major changes in aamks.py;
passing iteration id (sim_id) to manager.OnInit and manager.OnEnd and other methods;
change of logger to iteration-located;
creating mirrors of aamks.sqlite in iteration directory;
To improve performance and utilize embarrassingly parallel nature of Monte Carlo simulations we should perform all tasks we can on compute nodes, rather than on server.
This task would require:
aamks.py
;sim_id
) tomanager.OnInit
andmanager.OnEnd
and other methods;aamks.sqlite
in iteration directory;This task is a natural part of #305.