Table 1. Durations and settings for optimization problems.
| problem | duration (minutes) | # mesh intervals | # DOFs | # muscles | convergence tolerance | constraint tolerance |
|---|---|---|---|---|---|---|
| MocoInverse, gait, normal (Fig 7, black) | 3.5 | 25 | 29 | 80 | 10−2 | 10−3 |
| MocoInverse, gait, minimizing knee load (Fig 7, blue) | 15 | 25 | 29 | 80 | 10−2 | 10−3 |
| MocoTrack, gait, normal (Fig 8) | 130 | 40 | 29 | 80 | 10−3 | 10−3 |
| MocoTrack, gait, weak dorsiflexors (Fig 9, orange) | 101 | 40 | 29 | 80 | 10−3 | 10−3 |
| MocoTrack, gait, weak hip abductors (Fig 9, green) | 148 | 40 | 29 | 80 | 10−3 | 10−3 |
| Predictive MocoStudy, squat-to-stand, unassisted (Fig 10, black) | 3 | 50 | 4 | 9 | 10−3 | 10−3 |
| Predictive MocoStudy, squat-to-stand, assisted (Fig 10, blue) | 3 | 50 | 4 | 9 | 10−3 | 10−3 |
Durations were obtained with a 3.6 GHz Intel Core i7 processor with 8 parallel threads. All problems were solved with the Hermite–Simpson transcription scheme; each mesh interval in this scheme contains two points on the interval boundary and a collocation point at the interval midpoint (i.e., a problem with N mesh intervals has 2N + 1 time points). DOFs: degrees of freedom; convergence tolerance: the tolerance on the KKT conditions [20]; constraint tolerance: the largest permissible value of any single constraint violation [63].