Jason K. Moore
Jason K. Moore
Given the new documentation organization (Tutorials, How-To, Explanations, API), this seems like it fits with the explanations more than the API.
The loop in `ufuncify_matrix` is specified from the shape of the matrix: https://github.com/csu-hmc/opty/blob/master/opty/utils.py#L469 It may be possible to simply change the number of nodes on an existing `Problem` to try...
@Peter230655 the paper is this: Ackermann, M., & van den Bogert, A. J. (2010). Optimality principles for model-based prediction of human gait. Journal of Biomechanics, 43(6), 1055–1060. https://doi.org/10.1016/j.jbiomech.2009.12.012 I'll email...
This runs! See #29
Closed after 10 years!
Looks like cholesky is faster when using scipy's solvers: ```python In [1]: import numpy as np In [2]: from scipy.linalg import solve In [3]: from pydy.models import n_link_pendulum_on_cart In [4]:...
A pure sympy example needs to be created and then an issue on sympy opened about cholesky having more flops.
I added https://github.com/pydy/pydy/blob/master/bin/compare_linear_systems_solvers.py which can be useful for timing the overhead of the different linear system solvers. It outputs (for 4 pendulums): ``` SymPy LU decomposition symbolic solve time: 0.03291200300009223...
Looks useful: https://sqlpey.com/python/top-methods-to-distribute-cython-extensions-utilizing-lapack-with-python/
And: https://proceedings.scipy.org/articles/Majora-7b98e3ed-008.pdf