Error, recommended step is NaN

[JJPBos]

Hi all!

I am trying to run a propagation with a variable step size integrator, with a minimum time step. However, I run into the error:

Warning in step-size control, minimum step size 8.000000 is higher than required time step 3.065272, minimum step will be used.
Error, recommended step is NaN
Error, propagation terminated at t=792685727.055000, returning propagation data up to current time

which is immediately after the first time step. I'm not sure what can cause this error, since it works fine if I set the minimum step size to np.finfo(float).eps.
I've set up the integrator settings as the following: (please let me know if I'm doing anything wrong)

min_step = 8
tolerance = 1e-10
    step_size_validation = propagation_setup.integrator.step_size_validation(
              min_step,
              np.inf,
              propagation_setup.integrator.MinimumIntegrationTimeStepHandling.set_to_minimum_step_single_warning)
             
    step_size_control = propagation_setup.integrator.step_size_control_elementwise_scalar_tolerance(
            tolerance,
            tolerance)

   coefficient_set = getattr(propagation_setup.integrator.CoefficientSets,rkf_78

    integrator_settings = propagation_setup.integrator.runge_kutta_variable_step(
            min_step,
            coefficient_set,
            step_size_control,
            step_size_validation

Thanks in advance!
kind regards,
Justin Bos

[DominicDirkx]

To figure out what's the cause of this, could you send a link to the script you're running? If you use a fixed rather than a variable time step integrator, does the propagation proceed normally?

[JJPBos]

So my code with regards to the whole propagation is split up in many parts and would be very confusing to go through to find the error. But I managed to set up a small script using the tudat examples that reproduces the same error. If I use lines 75-80 (now commented out) instead of lines 85-105, it works correctly with a fixed integrator. If I set the min_step_size to a lower value such that it does not reach this, it also works correctly. I am afraid it might be due to my old version of tudat compiled on my system, but it will be good to check this.
test_variable_step.txt

[DominicDirkx]

I ran your script and I get the same problem. I think it's an issue with the way the minimum time is handled. When I set a more reasonable value for the minimum time step (0.01) it works fine. Could you open an issue on tudat with this script and a link to this discussion explaining the problem?

[JJPBos]

Allright, I had hoped it was a small typo on my side somewhere, because that would be faster to solve, but good to know. I will do that 👍 thank you for the replies.