TR2021-150

Homogeneous Formulation of Convex Quadratic Programs for Infeasibility Detection


    •  Raghunathan, A., "Homogeneous Formulation of Convex Quadratic Programs for Infeasibility Detection", IEEE Conference on Decision and Control (CDC), DOI: 10.1109/​CDC45484.2021.9682847, December 2021, pp. 968-973.
      BibTeX TR2021-150 PDF
      • @inproceedings{Raghunathan2021dec,
      • author = {Raghunathan, Arvind},
      • title = {Homogeneous Formulation of Convex Quadratic Programs for Infeasibility Detection},
      • booktitle = {IEEE Conference on Decision and Control (CDC)},
      • year = 2021,
      • pages = {968--973},
      • month = dec,
      • doi = {10.1109/CDC45484.2021.9682847},
      • issn = {2576-2370},
      • isbn = {978-1-6654-3659-5},
      • url = {https://www.merl.com/publications/TR2021-150}
      • }
  • MERL Contact:
  • Research Areas:

    Control, Optimization

Abstract:

Convex Quadratic Programs (QPs) have come to play a central role in the computation of control action for constrained dynamical systems. Convex QPs arise in Model Predictive Control (MPC) for linear systems, switched linear systems and in sequential linearization of nonlinear systems. A number algorithms have been developed in recent years for the solution of such QPs. However not all algorithms are capable of computing an optimal solution if feasible or producing a certificate of infeasibility. In this paper, we present a novel Homogeneous QP (HQP) formulation which is obtained by embedding the original QP in a larger space. The key properties of the HQP are: (i) is always feasible, (ii) an optimal solution to QP can be readily obtained from a solution to HQP, and (iii) infeasibility of QP corresponds to a particular solution of HQP. An immediate consequence is that all the existing algorithms for QP are now also capable of robustly detecting infeasibility. In particular, we present an Infeasible Interior Point Method (IIPM) for the HQP and show polynomial iteration complexity when applied to HQP. A key distinction with prior IPM approaches is that we do not need to solve second-order cone programs. Numerical experiments on the formulation are provided using existing codes.

 

  • Related News & Events

    •  AWARD    Arvind Raghunathan receives Roberto Tempo Best CDC Paper Award at 2022 IEEE Conference on Decision & Control (CDC)
      Date: December 8, 2022
      Awarded to: Arvind Raghunathan
      MERL Contact: Arvind Raghunathan
      Research Areas: Control, Optimization
      Brief
      • Arvind Raghunathan, Senior Principal Research Scientist in the Data Analytics group, received the IEEE Control Systems Society Roberto Tempo Best CDC Paper Award. The award was presented at the 2022 IEEE Conference on Decision & Control (CDC).

        The award is given annually in honor of Roberto Tempo, the 44th President of the IEEE Control Systems Society (CSS). The Tempo Award Committee selects the best paper from the previous year's CDC based on originality, potential impact on any aspect of control theory, technology, or implementation, and for the clarity of writing. This year's award committee was headed by Prof. Patrizio Colaneri, Politecnico di Milano. Arvind's paper was nominated for the award by Prof. Lorenz Biegler, Carnegie Mellon University, with supporting letters from Prof. Andreas Waechter, Northwestern University, and Prof. Victor Zavala, University of Wisconsin-Madison.
    •  
  • Related Publication

  •  Raghunathan, A., "Homogeneous Formulation of Convex Quadratic Programs for Infeasibility Detection", arXiv, pp. 12, January 2022.
    BibTeX arXiv
    • @article{Raghunathan2022jan,
    • author = {Raghunathan, Arvind},
    • title = {Homogeneous Formulation of Convex Quadratic Programs for Infeasibility Detection},
    • journal = {arXiv},
    • year = 2022,
    • pages = 12,
    • month = jan,
    • url = {https://arxiv.org/abs/2112.15316}
    • }