search

Rapid-Design-of-Systems-Laboratory / beluga

General purpose indirect trajectory optimization
Other
26 stars 6 forks source link

Bisection strategy does not function #134

Closed SeanMatthewNolan closed 5 years ago

SeanMatthewNolan commented 5 years ago

When the shooting method fails (at least with the error exceeds max_error), a bisection does not occur. Instead the solver just exits.

SeanMatthewNolan commented 5 years ago

The bisection strategy works sometimes. I don't know why.

msparapa commented 5 years ago

If the first iteration in a continuation doesn't converge, bisection won't retry if it fails. If it's the second or third or later on, then it should retry with a smaller step. I think this is correct behavior. I don't recall who programmed this.

SeanMatthewNolan commented 5 years ago

I know that. It was quitting on the second step though. I've also had it quit after the second step too.

SeanMatthewNolan commented 5 years ago

In the first step it exits because there is nothing to bisect. Thomas programmed it.

msparapa commented 5 years ago

Huh weird. We probably just need to go through the entire module in detail sometime.

SeanMatthewNolan commented 5 years ago

Agreed, I think it has to do with exception handling.

SeanMatthewNolan commented 5 years ago

I haven't run into this issue in a while. I think it was a matter of running into the wrong type of exception. I think we should close insofar as I don't think I can recreate the problem anymore. If it does occur again, a more specific issue should be made.