You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
By default JumpProcesses reseeds with a rand(UInt64) on every call to solve. But there is no real reason to do this for a sequence of solve calls running on the same thread, and it leads to undesirable effects like non-reproduciblity of results (because it doesn't respect the user's rng choice).
It would be breaking, but I think we should drop reseeding if no seed is passed. If this has an impact on EnsembleProblems, then they should be updated to properly handle passing new unique seeds to spawned solvers when needed instead of trying to put this decision at the level of solve.
The text was updated successfully, but these errors were encountered:
By default JumpProcesses reseeds with a
rand(UInt64)
on every call to solve. But there is no real reason to do this for a sequence of solve calls running on the same thread, and it leads to undesirable effects like non-reproduciblity of results (because it doesn't respect the user's rng choice).It would be breaking, but I think we should drop reseeding if no seed is passed. If this has an impact on EnsembleProblems, then they should be updated to properly handle passing new unique seeds to spawned solvers when needed instead of trying to put this decision at the level of
solve
.The text was updated successfully, but these errors were encountered: