https://github.com/philzook58/trajopt
Trajectory Optimization Using Pytorch
https://github.com/philzook58/trajopt
Last synced: 7 months ago
JSON representation
Trajectory Optimization Using Pytorch
- Host: GitHub
- URL: https://github.com/philzook58/trajopt
- Owner: philzook58
- Created: 2018-07-21T20:14:04.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2019-03-03T04:23:15.000Z (over 6 years ago)
- Last Synced: 2025-01-08T15:34:15.454Z (9 months ago)
- Language: Python
- Size: 410 KB
- Stars: 2
- Watchers: 4
- Forks: 2
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# TrajOpt
Automatic Differentiation is pretty sweet. It makes it easy to make optimizers.
You can use an out of the box optimizer. Adam seemed to work pretty well.
Or you can write your own Newton step solve.r Pytorch does have the ability to get you hessians, but it is somewhat involved. The structure of trajectory optimization problems is reflected in the bandededness of the hessian. This makes it fast to invert and fast to get.
Enclosed is some experiments in using PyTorch to these ends.
See these blog posts for more.
http://www.philipzucker.com/deducing-row-sparse-matrices-matrix-vector-product-samples/
http://www.philipzucker.com/pytorch-trajectory-optimization-part-5-cleaner-code-50hz/
http://www.philipzucker.com/pytorch-trajectory-optimization-3-plugging-hessian/
http://www.philipzucker.com/pytorch-trajectory-optimization-part-2-work-progress/
http://www.philipzucker.com/pytorch-trajectory-optimization/