This is very bad. Since the timestep is 0 we have basically fixed the
fields at their initial value. All the tests that I have done in June
are meaningless.
The issue was already there at that time so this is not what broke it. I
dont see where the timestep gets reset to 0...the step() function sets
it correctly!