Skip to content

Latest commit

 

History

History
34 lines (22 loc) · 2.12 KB

README.md

File metadata and controls

34 lines (22 loc) · 2.12 KB

This repository provides the script to reproduce the examples from the Bioptim paper (https://www.biorxiv.org/content/10.1101/2021.02.27.432868v1)

Dependencies

The following instruction were tested on Linux Ubuntu 20.04. While bioptim is now Windows compatible, it was not at the time of the paper. The bioptim API slightly changed since the release of the paper. It is therefore expected to fail if one wants to run the example with the latest version of bioptim.

In order to make these examples work you must have the following libraries=versions installed: ipopt=3.14.4 casadi=3.5.5 eigen=3.4.0 biorbd=1.8.1 bioviz=2.1.3 bioptim=2.1.0 cmake>=3.19 # Only required to build ACADOS

The easiest way to install the right versions in one go is to run the following command, assuming a conda environment is loaded:

conda install bioptim=2.1.0 -cconda-forge

Moreover, to run the ACADOS examples, one must compile ACADOS=0.1.4. To automatically download and install it (again assuming the conda environment is loaded) the script external/acados_install.sh can be run.

Please note that the Ipopt solver used is ma57 that must be manually downloaded and installed from the hsl website https://www.hsl.rl.ac.uk/. The default solver (mumps) will be a bit slower and will solve in more iterations.

Running the examples

Once everything is properly installed, to generate the table of the paper (except for gait which is skipped because it can be a bit RAM extensive!), you can run the table_generation.py script. You can also explore each individual optimal control programs by having a look a main.py in each folder.

Please note that the generate_table.py files use ma57, while main.py files use mumps, that is so one which does not have ma57 still can run the examples.

If anything goes wrong (weird behavior of the solver or you are not able to reproduce the results of the paper), please increase the print level of the solver up to 5 (in pointing/pointing/ocp.py for instance) to catch the probable errors that you are missing with the optimized settings by default.

Thanks for using Bioptim!