Describe the bug
When running a stats model every few (approx 1/10) models gives an error like:
/home/jago/CamPyRoS/campyros/main.py:457: RuntimeWarning:overflow encountered in multiply
This line calculates the parachute force:
|
F_parachute_i = -0.5 * q * ref_area * CD * v_relative_wind_i / air_speed |
I have also seen 'invalid division' errors from this line too so suspect that the issue is occurring when air_speed is close to zero.
To Reproduce
So far I do not know what particular combination of rail angles, wind etc. result in this situation occurring but it should not be hard to investigate. Logging parachute forces and initial conditions would hopefully show a pattern.
Expected behavior
We should never have a situation where we have near 'infinite' parachute force. I suspect that the error is occurring in intermediate time steps which will be thrown away by the integrator (the integrator attempts to minimise the timesteps and in doing so calculates intermediaries which it then checks are not too different from the previous step, otherwise it throws them away). If this is the case then we don't really have a problem, otherwise we have a major one.
Setup
Stock install
Describe the bug
When running a stats model every few (approx 1/10) models gives an error like:
/home/jago/CamPyRoS/campyros/main.py:457: RuntimeWarning:overflow encountered in multiplyThis line calculates the parachute force:
CamPyRoS/campyros/main.py
Line 457 in 2ba4858
I have also seen 'invalid division' errors from this line too so suspect that the issue is occurring when
air_speedis close to zero.To Reproduce
So far I do not know what particular combination of rail angles, wind etc. result in this situation occurring but it should not be hard to investigate. Logging parachute forces and initial conditions would hopefully show a pattern.
Expected behavior
We should never have a situation where we have near 'infinite' parachute force. I suspect that the error is occurring in intermediate time steps which will be thrown away by the integrator (the integrator attempts to minimise the timesteps and in doing so calculates intermediaries which it then checks are not too different from the previous step, otherwise it throws them away). If this is the case then we don't really have a problem, otherwise we have a major one.
Setup
Stock install