Denis, we do a mixture in ASPECT: We have a "resume computation = true" setting in the prm and ASPECT will first load all settings from the ParameterHandler and then resume from the last snapshot by deserializing necessary information (including current time, mesh, etc.). Note that we do not serialize things that come directly from the ParameterHandler (like a solver tolerance). This means you can resume a computation and change some of your settings if you want to. We also allow changing the number of processors for parallel computations.
On Wed, Dec 21, 2016 at 2:07 PM, Denis Davydov <[email protected]> wrote: > Hi all, > > This is a bit off-topic, but i would like to ask how do you prefer to > implement restart/checkpoints in your user applications? > I see two options: > > 1. Make restarting usable with the same input parameter file so that > calculations will continue exactly from the point of restart. > For example, if you wanted to do 10 adaptive refinements and one node failed > on 6th step, you would use exactly the same input parameter > file to automatically continue from the 5th step if the application see that > restart information is available. > > 2. Another approach is to consider restarts as an arbitrary input (similar > to the input mesh / initial conditions and alike). > With the same example in mind, if one uses the same input parameter file > with 10 refinement steps > but specify `Restart = true` and thereby use the restart generated from 5th > step, > then the program would actually end up doing 10 refinement steps starting > from the input data taken from the restart. > In other words, 15 steps in total. > > Maybe ASPECT (Timo, Wolfagng?), pi-DoMUS ( Luca?) or DOpElib developers > could comment on which restart strategy you have chosen > for your libraries and why? > > Regards, > Denis. > > -- > The deal.II project is located at > https://urldefense.proofpoint.com/v2/url?u=http-3A__www.dealii.org_&d=CwIBaQ&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=4k7iKXbjGC8LfYxVJJXiaYVu6FRWmEjX38S7JmlS9Vw&m=r-7wFUtobVXE2pXM7WZmXxixuk-9gpsC5egYRKzungo&s=5yqBM479thjOHwWYj9Xu5QFv1sCOhlzUu0zFAXCcWsg&e= > > For mailing list/forum options, see > https://urldefense.proofpoint.com/v2/url?u=https-3A__groups.google.com_d_forum_dealii-3Fhl-3Den&d=CwIBaQ&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=4k7iKXbjGC8LfYxVJJXiaYVu6FRWmEjX38S7JmlS9Vw&m=r-7wFUtobVXE2pXM7WZmXxixuk-9gpsC5egYRKzungo&s=br5oifg7wi1MYsz1b2YlFOu5G4tYVhsl8VEx_4qjn7w&e= > > --- > You received this message because you are subscribed to the Google Groups > "deal.II User Group" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit > https://urldefense.proofpoint.com/v2/url?u=https-3A__groups.google.com_d_optout&d=CwIBaQ&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=4k7iKXbjGC8LfYxVJJXiaYVu6FRWmEjX38S7JmlS9Vw&m=r-7wFUtobVXE2pXM7WZmXxixuk-9gpsC5egYRKzungo&s=i9do8gXgWLRPtwJOUf6hzreCFmfTHbMGftiuEfIM3z4&e= > . -- Timo Heister http://www.math.clemson.edu/~heister/ -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
