Changing parameter in case of divergence and trying to solve again

Question

Hi!

I would like to try to solve my problem and, if it diverges, change some parameter (for instance, to reduce the timestep) and retry the solution.

Which is the best way of doing this?

Thank you!

Answer 1

You need to save the current solution, so you can recall in case of divergence

Ubck = Function(U)
while True:
     try:
         solver.solve()
         break
     except:
         U.assign(Ubck)
         dt.assign(0.5*float(dt))

This assumes that you have defined a solver object, which writes to U and throws an error in case of divergence.

This is a way. Maybe not the best way. At least you should also consider increasing the timestep.