Tags: wright/dymos
Tags
Merge pull request OpenMDAO#203 from OpenMDAO/develop Develop
Phase-specific ode options (OpenMDAO#155) * ODEOptions are now optional on an ODE system. All ODE options can be set via the phase set_time_options, set_state_options, add_control, add_design_parameter, and add_input_parameter interfaces. So far only brachistochrone has been tested. More tests that include input parameters and traj parameters need to be added. * trajectories now starting to work under the new ODE options procedures. traj_parameters removed from phases -> trajectories now add the appropriate input parameters to each phase when setting up their input and design parameters. * Phase option 'ode_class' can now be set in phase.setup methods if inheriting from a Phase class. Phase time, state, control, and parameter options can now be set in phase.setup if inheriting from a Phase class. If doing this then the call to super.setup should occur after these have been set. Temporarily removed rate_param operability from simulate since rate_params have been changed to rate_targets. Proper rate_target handling will occur in SimulationPhase refactor. * fixes for polynomial controls * add_objective is now a method on PhaseBase which caches objective information until setup. During setup, the _setup_objective method determines the path to the objective output and adds it through the standard openMDAO methods. * Moved time_options checking to the Phase setup stack. Removed deprecated time options opt_initial and opt_duration. * working out issues with steady flight example * Non-opt controls automatically have continuity/rate continuity disabled. Only trajectory linkages remain to be moved over to the new system in which time/state/control/parameter options are not available until setup. * two phase cannonball working with constrained linkages * Trajectory phase linkages working for both connected and constrained linkages. * successful test of segment simulation component for use in SolveIVPPhase * SolveIVPPhase successful in integrating the simple ODE forward and backward at time. SolveIVPPhase will no longer support the times argument. All outputs are provided at all nodes of the given grid. * SolveIVPPhase now supports dense output with the output_nodes_per_seg option. If None, outputs are provided at 'all' nodes as defined in the GridData. If given as an integer (n), then each segment provides outputs at n equally distributed time points in the segment. * New SolveIVPPhase simulation is working from Trajectory.simulate(). * Tweaks to readme to reflect changes to dymos. Trajectory linkages now issue error for invalid linkage variable names. Linkage report now displays constraint linkages with `=` symbols and connection linkages with `-->` symbols. * Phase linkages now raise ValueError for invalid phase names
Merge pull request OpenMDAO#94 from kmarsteller/master re-working the caching and making of pyoptsparse with SNOPT
Brach example fix (OpenMDAO#66) * removed singularity from initial guess in brachistochrone example. added more tests * version 0.9.1 -> 0.9.2
synced to openmdao - remove explicit coloring for objective. moved mi… …n time climb differential inclusion to another repo for now. (OpenMDAO#48)