Rob Falck
Rob Falck
Thanks for the issue. Roughly half the time seems to be taken up by mpi4py under pyoptsparse. We'll have to think about what we can do to mitigate any of...
Hmm I can't open that file. I'm not sure why a redirect created tuna as a binary file.
Thanks for the suggestion. Grid refinement is due for a good refactoring and we can incorporate this.
Currently PicardShooting has the following options: - `ode_nonlinear_solver` : The nonlinear solver (BlockGS by default) that solves the dynamics in each segment. - `ms_nonlinear_solver` : The nonlinear solver (BlockGS by...
Starting to look into this. This would appear to be in conflict with this portion of the [docs](https://openmdao.org/newdocs/versions/latest/features/core_features/working_with_groups/dist_serial.html?highlight=distributed#non-distributed-output-to-distributed-input). ``` Non-distributed output to distributed input These types of connections are deprecated...
I've seen this on the windows tests a few times. According to Gemini, it might just be the fact that windows is slow in deleting things: ``` When using shutil.rmtree...
This is a good suggestion. I'll see what we can do to incorporate it
Is there anything I can do to help with this?
This is mostly complete but needs a user-facing way to check the relevance of a variable in compute_partials and other methods: ``` def is_relevant(self, vname): return self._relevance.is_relevant(vname) ```
Have you had a chance to trace down any more errors caused by this update?