This repository accompagnies the Wiley paper "A Generalization of the Riccati Recursion for Equality-Constrained Linear Quadratic Optimal Control". The purpose of the repository is:

• reproducable benchmark results and possibility to test out performance on other platforms and for other problems

• possible re-use of code, specifically the gen_riccati part of this repo

• clarifications for details of our implementation

This repository contains

• gen_riccati: a library that implements the generalization of Riccati recursion from the paper, the only dependency here is blasfeo

dependencies: blasfeo

• symspals: a library named that allows setting up sparse linear systems symbolically, the library is interfaced to three sparse linear solvers: mumps, ma57 and pardiso

dependencies: sparse linear solvers: mumps, ma57 and pardiso

to compile, make sure to obtain the source and binary files of ma57 and pardiso, respectively. Place them at the following locations: external/ma57_src/ddeps.f external/ma57_src/fakemetis.f external/ma57_src/ma57d.f and external/pardiso_shared_lib/libpardiso.so.

• the code for performing the benchmark of the paper: here we have a dependency on fatrop for setting up the problems

dependencies: symspals, gen_riccati and fatrop

citing:

if you use this code, please cite the paper as follows:

@article{vanroye2024generalization,
  title={A generalization of the Riccati recursion for equality-constrained linear quadratic optimal control},
  author={Vanroye, Lander and De Schutter, Joris and Decr{\'e}, Wilm},
  journal={Optimal Control Applications and Methods},
  volume={45},
  number={1},
  pages={436--454},
  year={2024},
  publisher={Wiley Online Library}
}

Known issues: not possible to use ma57 and mumps with metis library at the same time because ma57