OpenMDAO / dymos / 1085

91%

push

travis-pro

Vectorized scaling (#491)

* moved solve_segments validation to _configure_solve_segments in PseudospectralBase from StateIndependentsComp.  Valid values for solve_segments are now True, False, None, "forward", and "backward".

* view_connections_bug

* reworking things to allow solve_segments="backward"

* checkpoint

* StateIndependentsComp update

* starting to fix some testing errors now that things are fundamentally working

* now correctly negating vector fix_initial/fix_final specification when determining indices of state_inputs that are design variables.

* continuing to get tests to pass.  need to provide separate ivc and state indeps comp next.

* down to two failing tests when connect=True is used

* pseudospectral base now picks the correct indices for the state design variables when connected_initial is True

* cleaning up solve_segments examples to use forward or backward instead of True or False.

* Merged out Dymos 0.17.0

* fixed some errant SNOPT usages

* cleanup of deprecation warning when solve_segments=True

* solve_segments=False was incorrectly triggering a deprecation warning

* changed print_level for IPOPT since logs are getting too long for CI

* Fixed issue with vectorized scaling

* fixed issue with trying to divide by array valued variable

* removed snopt from committed tests

* Added vectorized scaling brachistochrone problem unit tests

Co-authored-by: Rob Falck <rfalck@nasa.gov>
Co-authored-by: Kaushik Ponnapalli <kaushik.s.ponnapalli@nasa.gov>

13 of 13 new or added lines in 3 files covered. (100.0%)

24637 of 25629 relevant lines covered (96.13%)

2.88 hits per line