Redesign Muscle to exploit MCI and improve robustness

opensim-org / opensim-core

SimTK OpenSim C++ libraries and command-line applications, and Java/Python wrapping.

https://opensim.stanford.edu

Apache License 2.0

803 stars 324 forks source link

Redesign Muscle to exploit MCI and improve robustness #366

Open tkuchida opened 9 years ago

tkuchida commented 9 years ago

[ ] Add MuscleActivationDynamics as a subcomponent of Muscle.
[ ] Eliminate the duplicate MuscleFirstOrderActivationDynamicModel (replaced by FirstOrderMuscleActivationDynamics).
[ ] Rename ZerothOrderMuscleActivationDynamics to NoMuscleActivationDynamics.
[ ] Remove pennation angle property from Muscle.
[ ] Make MuscleFixedWidthPennationModel a property and subcomponent of Millard2012{Equilibrium|Acceleration}Muscle --or-- create pennation model classes mirroring the new activation dynamics classes and make a property of Muscle.
[ ] Avoid inconsistencies in fiber length by removing calls to MuscleFixedWidthPennationModel::clampFiberLength() from muscle models.
[ ] Add tests for each component in isolation and also ensure they function correctly once hooked into a Muscle.
[ ] …

aseth1 commented 9 years ago

I like where this is going. Why not go straight to Muscle including the slots for its submodels (e.g. its concrete MuscleActivationDynamics?). Otherwise we are going to duplicate for Millard, etc... Also a secondary point, I don't think people are going to figure out what ZerothOrderMuscleActivationDynamics means. NoActivationDynamics or something along those lines may be clearer?

tkuchida commented 9 years ago

Why not go straight to Muscle including the slots for its submodels (e.g. its concrete MuscleActivationDynamics?).

Sounds like a good plan, though that will be a big job since it will affect all muscles.

NoActivationDynamics or something along those lines may be clearer?

I propose NoMuscleActivationDynamics so we have {NULL|No|FirstOrder}MuscleActivationDynamics classes.

aseth1 commented 9 years ago

The unspecified case would default to NoMuscleActivationDynamics, so there wouldn't be a NULL MuscleActivationDynamics model case.

aseth1 commented 9 years ago

Also

I propose NoMuscleActivationDynamics

:+1:

tkuchida commented 9 years ago

The unspecified case would default to NoMuscleActivationDynamics, so there wouldn't be a NULL MuscleActivationDynamics model case.

I meant that the abstract class would be named MuscleActivationDynamics and all concrete classes would be named SOMETHINGMuscleActivationDynamics.

jenhicks commented 9 years ago

@tkuchida Do you still want to complete this (or a subset of the items in the checklist) for OpenSim 4.0?

tkuchida commented 9 years ago

@jenhicks This checklist isn't on the critical path to a 4.0 release, but it will need to be done to support upcoming work extending the muscle models (the timeline and distribution of labor isn't yet clear).

chrisdembia commented 9 years ago

Potentially related: bugs in Thelen2003Muscle: https://github.com/opensim-org/opensim-core/issues/687