Adding a penalty term to CMC objective function

[Daniel McFarland]

Hi,

I'm working on developing my own ActuatorForceTarget class, so that I can adjust the objective function of CMC to include a penalty term to force the JRF within the glenoid, and thereby guarantee stability during upper limb simulations. I've managed to successfully calculate the JRF within the objective function, but have run into some difficulty when trying to add in the penalty term to the performance calculation.

In the original ActuatorForceTarget, it looks like in the prepareToOptimize step, precomputed performance matrices are computed and then later used in the optimization. There's a comment in the objective function that says this "works if it's really linear". I want to make sure that adding a penalty term to this is alright as I've found a penalty term that works with this, but it had to be rather small.

The original ActuatorForceTarget also seems to be set up to be able to calculate the performance explicitly within the objective function. I've tried simulations with this method of performance calculation ,and this method is more receptive of a penalty term. These simulation, however, seems to have trouble converging even without including the penalty term. The performance criteria will hover around a value, but not converge. I've tried adjusting convergence tolerance without much luck. I'm wondering why this might be happening and if there is anyway to help it converge?

Thanks,
Daniel