Abstract
In semidefinite programming the dual may fail to attain its optimal value and there could be a duality gap, i.e., the primal and dual optimal values may differ. In a striking paper, Ramana (Math. Program. Ser. B 77, 129–162, 1997) proposed a polynomial size extended dual that does not have these deficiencies and yields a number of fundamental results in complexity theory. In this work we walk the reader through a concise and self-contained derivation of Ramana’s dual, relying mostly on elementary linear algebra.
Notes
Proposition 3 is usually stated in the space \({\mathbb {R}}^{n}.\)
That is, K is closed, convex, and \(\lambda x \in K\) for all \(x \in K\) and \(\lambda \ge 0.\)
This means two things: (i) it is a convex subset of \({{\mathcal {S}}}_+^{n}\) and (ii) if X, Y are in \({{\mathcal {S}}}_+^{n},\) and the open line segment \(\{ \lambda X + (1-\lambda )Y : \lambda \in (0,1) \}\) intersects \(F, \, \) then both X and Y must be in F.
Afterwards \(U_1\) may not look like in Eq. (3.39) anymore, but for our purposes this does not matter.
The following argument may better explain the role of the \(Y_i\) matrices. If S is any slack, then \(\langle S, Y_1 \rangle = \langle S, Y_2 \rangle = 0. \, \) We invite the reader to check that these equations lead to the same argument that we gave in paragraph (1) that show the last two rows and columns of S are zero.
We can use any matrix norm, for example the spectral norm or the Frobenius norm.
References
Ben-Tal, A., Nemirovskii, A.: Lectures on Modern Convex Optimization. MPS/SIAM Series on Optimization. SIAM, Philadelphia (2001)
Borwein, J.M., Lewis, A.S.: Convex Analysis and Nonlinear Optimization: Theory and Examples. CMS Books in Mathematics, 2nd edn. Springer, Berlin (2005)
Borwein, J.M., Wolkowicz, H.: Facial reduction for a cone-convex programming problem. J. Aust. Math. Soc. 30, 369–380 (1981)
Borwein, J.M., Wolkowicz, H.: Regularizing the abstract convex program. J. Math. Anal. Appl. 83, 495–530 (1981)
De Klerk, E.: Aspects of Semidefinite Programming: Interior Point Algorithms and Selected Applications. Springer, Cham (2006)
De Klerk, E., Terlaky, T., Roos, K.: Self-dual embeddings. In: Handbook of Semidefinite Programming, pp 111–138. Springer (2000)
Drusvyatskiy, D., Wolkowicz, H., et al.: The many faces of degeneracy in conic optimization. Found. Trends® Opt. 3(2), 77–170 (2017)
Klep, I., Schweighofer, M.: An exact duality theory for semidefinite programming based on sums of squares. Math. Oper. Res. 38(3), 569–590 (2013)
Laurent, M., Rendl, F.: Semidefinite programming and integer programming. Handb. Oper. Res. Manag. Sci. 12, 393–514 (2005)
Liu, M., Pataki, G.: Exact duals and short certificates of infeasibility and weak infeasibility in conic linear programming. Math. Program. 167(2), 435–480 (2018)
Lourenço, B.F., Muramatsu, M., Tsuchiya, T.: A structural geometrical analysis of weakly infeasible SDPs. J. Oper. Res. Soc. Jpn. 59(3), 241–257 (2016)
Lovász, L.: Graphs and Geometry, vol. 65. American Mathematical Society, Providence (2019)
Luo, Z.-Q., Sturm, J., Zhang, S.: Duality Results for Conic Convex Programming. Technical Report Report 9719/A, Erasmus University Rotterdam, Econometric Institute, The Netherlands (1997)
Nemirovski, A.: Advances in convex optimization: conic programming. Int. Congr. Math. 1, 413–444 (2007)
O’Donnell, R.: SOS is not obviously automatizable, even approximately. In: 8th Innovations in Theoretical Computer Science Conference (ITCS 2017). Schloss Dagstuhl-Leibniz-Zentrum fuer Informatik (2017)
Pataki, G.: Strong duality in conic linear programming: facial reduction and extended duals. In: Bailey, D., Bauschke, H.H., Garvan, F., Théra, M., Vanderwerff, J.D., Wolkowicz, H. (eds.) Proceedings of Jonfest: A Conference in Honour of the 60th Birthday of Jon Borwein. Springer (2013). Also available from http://arxiv.org/abs/1301.7717
Pataki, G.: Characterizing bad semidefinite programs: normal forms and short proofs. SIAM Rev. 61(4), 839–859 (2019)
Pataki, G.: Bad semidefinite programs: they all look the same. SIAM J. Opt. 27(1), 146–172 (2017)
Permenter, F., Parrilo, P.: Partial facial reduction: simplified, equivalent SDPs via approximations of the PSD cone. Math. Program. 171, 1–54 (2014)
Ramana, M.V.: An exact duality theory for semidefinite programming and its complexity implications. Math. Program. Ser. B 77, 129–162 (1997)
Ramana, M.V., Freund, R.: On the ELSD Duality Theory for SDP. Technical Report, MIT (1996)
Ramana, M.V., Tunçel, L., Wolkowicz, H.: Strong duality for semidefinite programming. SIAM J. Opt. 7(3), 641–662 (1997)
Renegar, J.: A Mathematical View of Interior-Point Methods in Convex Optimization. MPS-SIAM Series on Optimization, SIAM, Philadelphia (2001)
Rockafellar, T.R.: Convex Analysis. Princeton University Press, Princeton (1970)
Vandenberghe, L., Boyd, S.: Semidefinite programming. SIAM Rev. 38(1), 49–95 (1996)
Waki, H., Muramatsu, M.: Facial reduction algorithms for conic optimization problems. J. Optim. Theory Appl. 158(1), 188–215 (2013)
Zhu, Y., Pataki, G., Tran-Dinh, Q.: Sieve-SDP: a simple facial reduction algorithm to preprocess semidefinite programs. Math. Program. Comput. 11(3), 503–586 (2019)
Acknowledgements
Part of this paper was written during a visit to Chapel Hill by the first author and he would like to express his gratitude to Prof. Takashi Tsuchiya, for creating conditions that made this visit possible. The work of the first author was partially supported by the JSPS KAKENHI Grant Numbers JP15H02968 and JP19K20217. The second author thanks Pravesh Kothari and Ryan O’ Donnell for helpful discussions on SDP. The work of the second author was supported by the National Science Foundation, award DMS-1817272. Both authors are grateful to Siyuan Chen, Alex Touzov, and Yuzixuan Zhu for their careful reading of the manuscript and their helpful comments. Most importantly, we are very grateful to the anonymous referees whose comments and suggestions greatly improved the manuscript.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Lourenço, B.F., Pataki, G. A simplified treatment of Ramana’s exact dual for semidefinite programming. Optim Lett 17, 219–243 (2023). https://doi.org/10.1007/s11590-022-01898-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11590-022-01898-2