Abstract
The epsilon-subdifferential of convex univariate piecewise linear-quadratic (PLQ) functions can be computed in linear worst-case time complexity as the level-set of a convex function. Using binary search, we improve the complexity to logarithmic worst-case time, and prove such complexity is optimal. In addition, a new algorithm to compute the entire graph of the epsilon-subdifferential in (optimal) linear time is presented. Both algorithms are not limited to convex PLQ functions but are also applicable to any convex piecewise-defined functions with little restrictions.
Similar content being viewed by others
References
Bajaj, A., Hare, W., Lucet, Y.: Visualization of the đťś–-subdifferential of piecewise linear-quadratic functions, Comput. Optim. Appl. Accepted for publication (2016)
Bauschke, H.H., Combettes, P.L.: Convex analysis and monotone operator theory in Hilbert spaces, CMS Books in Mathematics/Ouvrages de Mathématiques de la SMC, 2nd edn. Springer, Cham (2017). With a foreword by Hédy Attouch
Bauschke, H.H., Goebel, R., Lucet, Y., Wang, X.: The proximal average: Basic theory. SIAM J. Optim. 19, 768–785 (2008)
Bauschke, H.H., Moffat, S.M., Wang, X.: Self-dual smooth approximations of convex functions via the proximal average. In: Bauschke, H. H., Burachik, R. S., Combettes, P. L., Elser, V., Luke, D. R., Wolkowicz, H. (eds.) Fixed-Point Algorithms for Inverse Problems in Science and Engineering, vol. 49 of Springer Optimization and Its Applications, pp. 23–32. Springer, New York (2011)
Bauschke, H.H., Mohrenschildt, M.v.: Symbolic computation of Fenchel conjugates. ACM Commun. Comput. Algebra 40, 18–28 (2006)
Borwein, J.M., Hamilton, C.H.: Symbolic computation of multidimensional Fenchel conjugates. In: ISSAC 2006, pp. 23–30. ACM, New York (2006)
Borwein, J.M.: Symbolic Fenchel conjugation. Math. Program. 116, 17–35 (2008)
Brøndsted, A., Rockafellar, R.T.: On the subdifferentiability of convex functions. Proc. Amer. Math. Soc. 16, 605–611 (1965)
Contento, L., Ern, A., Vermiglio, R.: A linear-time approximate convex envelope algorithm using the double Legendre-Fenchel transform with application to phase separation. Comput. Optim. Appl. 60, 231–261 (2015)
Gardiner, B., Lucet, Y.: Graph-matrix calculus for computational convex analysis. In: Bauschke, H. H., Burachik, R. S., Combettes, P. L., Elser, V., Luke, D. R., Wolkowicz, H. (eds.) Fixed-point Algorithms for Inverse Problems in Science and Engineering, vol. 49 of Springer Optimization and Its Applications, pp. 243–259. Springer, New York (2011)
Goebel, R.: The proximal average for saddle functions and its symmetry properties with respect to partial and saddle conjugacy. J. Nonlinear Convex Anal. 11, 1–11 (2010)
Goebel, R., Hare, W., Wang, X.: The optimal value and optimal solutions of the proximal average of convex functions. Nonlinear Anal. Theory Methods Appl. 75, 1290–1304 (2012)
Hare, W.: A proximal average for nonconvex functions: A proximal stability perspective. SIAM J. Optim. 20, 650–666 (2009)
Hiriart-Urruty, J.-B.: From convex optimization to nonconvex optimization. Necessary and sufficient conditions for global optimality. In: Nonsmooth Optimization and Related Topics (Erice, 1988), vol. 43 of Ettore Majorana Internat. Sci. Ser. Phys. Sci., pp. 219–239. Plenum, New York (1989)
Hiriart-Urruty, J.-B., Lemaréchal, C.: Convex Analysis and Minimization Algorithms I, vol. 305 of Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences]. Springer-Verlag, Berlin (1993). Vol I: Fundamentals
Hiriart-Urruty, J.-B.: Convex Analysis and Minimization Algorithms II, vol. 306 of Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences]. Springer-Verlag, Berlin (1993). Vol II: Advanced theory and bundle methods
Hiriart-Urruty, J.-B., Lucet, Y.: Parametric computation of the Legendre–Fenchel conjugate. J. Convex Anal. 14, 657–666 (2007)
Ioffe, A.D.: Approximate subdifferentials and applications. I. The finite-dimensional theory. Trans. Amer. Math Soc. 281, 389–416 (1984)
Johnstone, J., Koch, V., Lucet, Y.: Convexity of the proximal average. J. Optim. Theory Appl. 148, 107–124 (2011)
Lucet, Y.: A fast computational algorithm for the Legendre–Fenchel transform. Comput. Optim. Appl. 6, 27–57 (1996)
Lucet, Y.: Computational Convex Analysis library, 1996-2013
Lucet, Y.: Faster than the fast legendre transform, the linear-time legendre transform. Numer. Algorithms 16, 171–185 (1997)
Lucet, Y.: Fast Moreau envelope computation I: Numerical algorithms. Numer. Algorithms 43, 235–249 (2006)
Lucet, Y.: What shape is your conjugate? A survey of computational convex analysis and its applications. SIAM Rev. 52, 505–542 (2010)
Lucet, Y.: Techniques and open questions in computational convex analysis. In: Computational and Analytical Mathematics, vol. 50 of Springer Proc. Math. Stat., pp. 485–500. Springer, New York (2013)
Lucet, Y., Bauschke, H.H., Trienis, M.: The piecewise linear-quadratic model for computational convex analysis. Comput. Optim. Appl. 43, 95–118 (2009)
Rockafellar, R.T., Wets, R.J.-B.: Variational Analysis. Springer-Verlag, Berlin (1998)
Trienis, M.: Computational convex analysis: From continuous deformation to finite convex integration, Master’s thesis, University of British Columbia (2007)
Tseng, P., Luo, Z.-Q.: On computing the nested sums and infimal convolutions of convex piecewise-linear functions. J. Algorithms 21, 240–266 (1996)
Acknowledgements
This work was supported in part by Discovery Grant #298145-2013 (Lucet) from the Natural Sciences and Engineering Research Council of Canada (NSERC). Part of the research was performed in the Computer-Aided Convex Analysis (CA2) laboratory funded by a Leaders Opportunity Fund (LOF, John R. Evans Leaders Fund – Funding for research infrastructure) from the Canada Foundation for Innovation (CFI) and by a British Columbia Knowledge Development Fund (BCKDF).
This work was started at the end of Anuj Bajaj MSc research under the guidance of Dr. Warren Hare. Their preliminary efforts, ideas, and interest motivated the authors to pursue more efficient algorithms. The authors thank them for their initial contribution without which this work would not have been possible.
The authors are very grateful for several suggestions by the reviewers, which greatly improve the presentation of the paper.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Kumar, D., Lucet, Y. Computation of the Epsilon-Subdifferential of Convex Piecewise Linear-Quadratic Functions in Optimal Worst-Case Time. Set-Valued Var. Anal 27, 623–641 (2019). https://doi.org/10.1007/s11228-018-0476-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11228-018-0476-5
Keywords
- Subdifferential
- đťś–-Subdifferentials
- Piecewise linear-quadratic functions
- Convex function
- Computational convex analysis (CCA)
- Computer-aided convex analysis
- Visualization