Abstract
Given an infeasible, unbounded, or pathological convex optimization problem, a natural question to ask is: what is the smallest change we can make to the problem’s parameters such that the problem becomes solvable? In this paper, we address this question by posing it as an optimization problem involving the minimization of a convex regularization function of the parameters, subject to the constraint that the parameters result in a solvable problem. We propose a heuristic for approximately solving this problem that is based on the penalty method and leverages recently developed methods that can efficiently evaluate the derivative of the solution of a convex cone program with respect to its parameters. We illustrate our method by applying it to examples in optimal control and economics.
Similar content being viewed by others
References
Agrawal A, Amos B, Barratt S, Boyd S, Diamond S, Kolter J.Z (2019a) Differentiable convex optimization layers. In: Advances in neural information processing systems, pp 9558–9570
Agrawal A, Barratt S, Boyd S, Busseti E, Moursi W (2019b) Differentiating through a cone program. J Appl Numer Optim 1(2):107–115
Amaldi E (1994) From finding maximum feasible subsystems of linear systems to feedforward neural network design. Ph.D. thesis, Citeseer
Amaldi E, Pfetsch M, Trotter L (1999) Some structural and algorithmic properties of the maximum feasible subsystem problem. In: International conference on integer programming and combinatorial optimization. Springer, pp 45–59
Amaral P, Júdice J, Sherali H (2008) A reformulation–linearization–convexification algorithm for optimal correction of an inconsistent system of linear constraints. Comput Oper Res 35(5):1494–1509
Barratt S, Boyd S (2019) Least squares auto-tuning. arXiv preprint arXiv:1904.05460
Ben-Tal A, Nemirovski A (2001) Lectures on modern convex optimization: analysis, algorithms, and engineering applications, vol 2. SIAM
Boyd S, Vandenberghe L (2004) Convex optimization. Cambridge University Press, Cambridge
Carver W (1922) Systems of linear inequalities. Ann Math 212–220
Chinneck J (1996) An effective polynomial-time heuristic for the minimum-cardinality IIS set-covering problem. Ann Math Artif Intell 17(1):127–144
Chinneck J (1997) Finding a useful subset of constraints for analysis in an infeasible linear program. INFORMS J Comput 9(2):164–174
Chinneck J (2001) Analyzing mathematical programs using MProbe. Ann Oper Res 104(1–4):33–48
Chinneck J (2007) Feasibility and infeasibility in optimization: algorithms and computational methods, vol 118. Springer, Berlin
Chinneck J, Dravnieks E (1991) Locating minimal infeasible constraint sets in linear programs. ORSA J Comput 3(2):157–168
Diamond S, Boyd S (2016) CVXPY: a Python-embedded modeling language for convex optimization. J Mach Learn Res 17(83):1–5
Gambella C, Marecek J, Mevissen M (2019) Projections onto the set of feasible inputs and the set of feasible solutions. In: Allerton conference on communication, control, and computing. IEEE, pp 937–943
Greenberg H (1987) ANALYZE: a computer-assisted analysis system for linear programming models. Oper Res Lett 6(5):249–255
Greenberg H (1987) Computer-assisted analysis for diagnosing infeasible or unbounded linear programs. In: Computation mathematical programming. Springer, pp 79–97
Greenberg H, Murphy F (1991) Approaches to diagnosing infeasible linear programs. ORSA J Comput 3(3):253–261
GUROBI Optimization (2019) Gurobi optimizer reference manual
IBM (2016) IBM ILOG CPLEX optimization studio CPLEX user’s manual
Karp R (1972) Reducibility among combinatorial problems. In: Complexity of computer computations, pp 85–103
Kellner K, Pfetsch M, Theobald T (2019) Irreducible infeasible subsystems of semidefinite systems. J Optim Theory Appl 181(3):727–742
Kurator W, O’Neill R (1980) PERUSE: an interactive system for mathematical programs. ACM Trans Math Softw 6(4):489–509
Martinet B (1970) Brève communication. régularisation d’inéquations variationnelles par approximations successives. ESAIM: Mathematical Modelling and Numerical Analysis-Modélisation Mathématique et Analyse Numérique 4(R3):154–158
MOSEK Aps (2020) MOSEK optimizer API for Python. https://docs.mosek.com
Nesterov Y (2013) Gradient methods for minimizing composite functions. Math Program 140(1):125–161
Obuchowska W (1998) Infeasibility analysis for systems of quadratic convex inequalities. Eur J Oper Res 107(3):633–643
Obuchowska W (1999) On infeasibility of systems of convex analytic inequalities. J Math Anal Appl 234(1):223–245
O’Donoghue B, Chu E, Parikh N, Boyd S (2016) Conic optimization via operator splitting and homogeneous self-dual embedding. J Optim Theory Appl 169(3):1042–1068
Parikh N, Boyd S (2014) Proximal algorithms. Found Trends® Optim 1(3):127–239. https://doi.org/10.1561/2400000003
Pfetsch M (2003) The maximum feasible subsystem problem and vertex-facet incidences of polyhedra. Ph.D. thesis
Pfetsch M (2008) Branch-and-cut for the maximum feasible subsystem problem. SIAM J Optim 19(1):21–38
Roodman G (1979) Note–post-infeasibility analysis in linear programming. Manage Sci 25(9):916–922
Sankaran J (1993) A note on resolving infeasibility in linear programs by constraint relaxation. Oper Res Lett 13(1):19–20
Tamiz M, Mardle S, Jones D (1995) Resolving inconsistency in infeasible linear programmes. Technical report, School of Mathematical Studies, University of Portsmouth, UK
Tamiz M, Mardle S, Jones D (1996) Detecting IIS in infeasible linear programmes using techniques from goal programming. Comput Oper Res 23(2):113–119
Van Loon JNM (1981) Irreducibly inconsistent systems of linear inequalities. Eur J Oper Res 8(3):283–288
Acknowledgements
S. Boyd is an Engineering Subject Editor for the Optimization and Engineering journal. S. Barratt is supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1656518.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflicts of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: Convex formulation
Appendix: Convex formulation
In the case that A is a constant while b and c are affine functions of \(\theta\), we can write (9) as an equivalent convex optimization problem. In the linear case (i.e., when \(\mathcal {K}= {\text{ R }}_+^n\)), we can simply drop the strong duality requirement (which always holds in this case) and express (9) as
For more general cones \(\mathcal {K}\) (such as, e.g., the second order cone), a sufficient condition for strong duality is that there exist a feasible point in the interior of the cone. We can write this as, for example,
(We could similarly constrain \(y\in \mathop {{\mathbf{int }}}\limits \mathcal {K}^{*}\) and \(s\in \mathcal {K}\).)
In general, optimizing over open constraint sets is challenging and these problems may not even have an optimal point, but, in practice (and for well-enough behaved r, e.g., r continuous) we can approximate the true optimal value of (6) by approximating the open set \(\mathop {{\mathbf{int }}}\limits \mathcal {K}\) as a sequence of closed sets \(\mathcal {K}_\varepsilon \subseteq \mathop {{\mathbf{int }}}\limits \mathcal {K}\) such that \(\mathcal {K}_\varepsilon \rightarrow \mathop {{\mathbf{int }}}\limits \mathcal {K}\) as \(\varepsilon \downarrow 0\).
Rights and permissions
About this article
Cite this article
Barratt, S., Angeris, G. & Boyd, S. Automatic repair of convex optimization problems. Optim Eng 22, 247–259 (2021). https://doi.org/10.1007/s11081-020-09508-9
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11081-020-09508-9