adaptive time stepping

Question

Hello,
I want to solve a nonlinear PDE in fenics. After a few iterations I want to change the size of the time step: Is that possible?
Here is what I tried, but unfortunately, it doesn't work.

# Time Step
dt = 2.5e-06       
dT = Constant(dt)

# Function F for Newton Method
F = F_1 + dT*F_2   

# Derivative of F for Newton Method
J = derivative(F, u, du)

# Create Nonlinear Problem and Newton Solver
Problem = NonlinearVariationalProblem(F, u, bcs, J)
Solver  = NonlinearVariationalSolver(Problem)

t = 0.0
u.interpolate(InitialConditions())  
i = 0
# Loop
while (i < 500):
    if i == 10:
     dT = 10**4*dT
    # Iterate
    i += 1
    t += dT
    # Store old Solution
    u0.vector()[:] = u.vector()
    # Solve Nonlinear Problem
    Solver.solve()

Can anyone help me?

Thanks a lot!

Comments

Please, add some more information. What goes wrong? What is the system you want to solve? And, make your code snippet a minimal example, that can be used to reproduce the error.

Answer 1

This is a programming error resulting from insufficient understanding how python variables works. Read this for an introduction. Then you should be able understand why this

# Loop
while (i < 500):
    if i == 10:
        #dT = 1e4*dT # this creates new name (with no relation to your form F)
        dT.assign(1e4*dt) # we rather mutate the existing object
        #dT *= 1e4 # we can use this either (*= operator also mutates object (if mutable))
    # Iterate
    ...

should work (not tested).

Comments

Thank you for response and help! Thank you Jan Blechta, that solved my question. It seems to work fine!

Answer 2

It is possible to change the time step.

If you want help with your specific code, please post a simpler but complete program, and identify precisely what is not working as you would expect it. Someone trying to help should be able to paste your code into a file and run it.