Abstract
Functional constrained optimization is becoming more and more important in machine learning and operations research. Such problems have potential applications in risk-averse machine learning, semisupervised learning and robust optimization among others. In this paper, we first present a novel Constraint Extrapolation (ConEx) method for solving convex functional constrained problems, which utilizes linear approximations of the constraint functions to define the extrapolation (or acceleration) step. We show that this method is a unified algorithm that achieves the best-known rate of convergence for solving different functional constrained convex composite problems, including convex or strongly convex, and smooth or nonsmooth problems with stochastic objective and/or stochastic constraints. Many of these rates of convergence were in fact obtained for the first time in the literature. In addition, ConEx is a single-loop algorithm that does not involve any penalty subproblems. Contrary to existing primal-dual methods, it does not require the projection of Lagrangian multipliers into a (possibly unknown) bounded set. Second, for nonconvex functional constrained problems, we introduce a new proximal point method which transforms the initial nonconvex problem into a sequence of convex problems by adding quadratic terms to both the objective and constraints. Under certain MFCQ-type assumption, we establish the convergence and rate of convergence of this method to KKT points when the convex subproblems are solved exactly or inexactly. For large-scale and stochastic problems, we present a more practical proximal point method in which the approximate solutions of the subproblems are computed by the aforementioned ConEx method. Under a strong feasibility assumption, we establish the total iteration complexity of ConEx required by this inexact proximal point method for a variety of problem settings, including nonconvex smooth or nonsmooth problems with stochastic objective and/or stochastic constraints. To the best of our knowledge, most of these convergence and complexity results of the proximal point method for nonconvex problems also seem to be new in the literature.
Similar content being viewed by others
Notes
This x, y is required to be non-random because we are dropping the inner product terms of the left hand side of (2.32).
References
Allen-Zhu, Z., Hazan, E.: Variance reduction for faster non-convex optimization. In: International Conference on Machine Learning, pp. 699–707 (2016)
Andreani, R., Haeser, G., Martínez, J.M.: On sequential optimality conditions for smooth constrained optimization. Optimization 60(5), 627–641 (2011)
Andreani, R., Martínez, J.M., Ramos, A., Silva, P.J.S.: Strict constraint qualifications and sequential optimality conditions for constrained optimization. Math. Oper. Res. 43, 693–717 (2018)
Aravkin, A.Y., Burke, J.V., Drusvyatskiy, D., Friedlander, M.P., Roy, S.: Level-set methods for convex optimization. Math. Program. 174, 359–390 (2018)
Ben-Tal, A., Nemirovski, A.: Non-Euclidean restricted memory level method for large-scale convex optimization. Math. Program. 102, 407–456 (2005)
Bertsekas, D.P.: Nonlinear Programming. Athena Scientific, Belmont (1999)
Bertsekas, D.P.: Convex Optimization Algorithms. Athena Scientific, Belmont (2015)
Cartis, C., Gould, N.I., Toint, P.L.: On the complexity of finding first-order critical points in constrained nonlinear optimization. Math. Program. 144(1), 93–106 (2014)
Chambolle, A., Pock, T.: A first-order primal-dual algorithm for convex problems with applications to imaging. J. Math. Imaging Vis. 40(1), 120–145 (2011)
Chen, Y., Lan, G., Ouyang, Y.: Optimal primal-dual methods for a class of saddle point problems. SIAM J. Optim. 24(4), 1779–1814 (2014)
Davis, D., Grimmer, B.: Proximally guided stochastic subgradient method for nonsmooth, nonconvex problems. arXiv:1707.03505v4 (2017)
Dinh, Q.T., Gumussoy, S., Michiels, W., Diehl, M.: Combining convex-concave decompositions and linearization approaches for solving BMIS, with application to static output feedback. arXiv:1109.3320 (2011)
Facchinei, F., Kungurtsev, V., Lampariello, L., Scutari, G.: Ghost penalties in nonconvex constrained optimization: diminishing stepsizes and iteration complexity. arXiv:1709.03384 (2017)
Fang, C., Li, C.J., Lin, Z., Zhang, T.: Spider: Near-optimal non- convex optimization via stochastic path-integrated differential estimator. Adv. Neural Inf. Process. Syst. 687–697 (2018)
Frostig, R., Ge, R., Kakade, S., Sidford, A.: Un-regularizing: approximate proximal point and faster stochastic algorithms for empirical risk minimization. In: International Conference on Machine Learning, pp. 2540–2548 (2015)
Ghadimi, S., Lan, G.: Stochastic first- and zeroth-order methods for nonconvex stochastic programming. SIAM J. Optim. 23(4), 2341–2368 (2013)
Ghadimi, S., Lan, G.: Accelerated gradient methods for nonconvex nonlinear and stochastic programming. Math. Program. 156(1–2), 59–99 (2016)
Güler, O.: New proximal point algorithms for convex minimization. SIAM J. Optim. 2(4), 649–664 (1992)
Hamedani, E.Y., Aybat, N.S.: A primal-dual algorithm for general convex-concave saddle point problems. arXiv:1803.01401 (2018)
Kong, W., Melo, J.G., Monteiro, R.D.: Complexity of a quadratic penalty accelerated inexact proximal point method for solving linearly constrained nonconvex composite programs. arXiv:1802.03504 (2018)
Lan, G.: First-order and Stochastic Optimization Methods for Machine Learning. Springer, Berlin (2020)
Lan, G., Lee, S., Zhou, Y.: Communication-efficient algorithms for decentralized and stochastic optimization. Math. Program. 180, 237–284 (2018)
Lan, G., Monteiro, R.D.C.: Iteration-complexity of first-order penalty methods for convex programming. Math. Program. 138, 115–139 (2013)
Lan, G., Monteiro, R.D.C.: Iteration-complexity of first-order augmented Lagrangian methods for convex programming. Math. Program. 155(1–2), 511–547 (2016)
Lan, G., Yang, Y.: Accelerated stochastic algorithms for nonconvex finite-sum and multi-block optimization. arXiv:1805.05411 (2018)
Lan, G., Zhou, Y.: An optimal randomized incremental gradient method. Math. Program. 171(1–2), 167–215 (2018)
Lan, G., Zhou, Z.: Algorithms for stochastic optimization with expectation constraints. arXiv:1604.03887 (2016)
Lemaréchal, C., Nemirovski, A.S., Nesterov, Y.E.: New variants of bundle methods. Math. Program. 69, 111–148 (1995)
Lin, Q., Ma, R., Yang, T.: Level-set methods for finite-sum constrained convex optimization. In: Proceedings of the 35th International Conference on Machine Learning, vol. 80, pp. 3112–3121 (2018)
Lin, Q., Nadarajah, S., Soheili, N.: A level-set method for convex optimization with a feasible solution path. SIAM J. Optim. 28(4), 3290–3311 (2018)
Ma, R., Lin, Q., Yang, T.: Proximally constrained methods for weakly convex optimization with weakly convex constraints. arXiv:1908.01871 (2019)
Mangasarian, O., Fromovitz, S.: The fritz john necessary optimality conditions in the presence of equality and inequality constraints. J. Math. Anal. Appl. 17, 37–47 (1967)
Martínez, J.M., Svaiter, B.F.: A practical optimality condition without constraint qualifications for nonlinear programming. J. Optim. Theory Appl. 118(1), 117–133 (2003)
Nemirovski, A.: Prox-method with rate of convergence o(1/t) for variational inequalities with Lipschitz continuous monotone operators and smooth convex-concave saddle point problems. SIAM J. Optim. 15(1), 229–251 (2004)
Nesterov, Y.: Lectures on Convex Optimization. Springer, Berlin (2018)
Nguyen, L.M., Liu, J., Scheinberg, K., c, M.T.: A novel method for machine learning problems using stochastic recursive gradient. In: Proceedings of the 34th International Conference on Machine Learning, vol. 70, pp. 2613–2621 (2017)
Nouiehed, M., Sanjabi, M., Lee, J.D., Razaviyayn, M.: Solving a class of non-convex min-max games using iterative first order methods. arXiv:1902.08297 (2019)
Pham, N.H., Nguyen, L.M., Phan, D.T., Tran-Dinh, Q.: Proxsarah: an efficient algorithmic framework for stochastic composite nonconvex optimization. arXiv:1902.05679 (2019)
Polyak, B.: A general method of solving extremum problems. Sov. Math. Doklady 8(3), 593–597 (1967)
Rafique, H., Liu, M., Lin, Q., Yang, T.: Non-convex min-max optimization: provable algorithms and applications in machine learning. arXiv:1810.02060 (2018)
Reddi, S.J., Hefny, A., Sra, S., Póczós, B., Smola, A.J.: Stochastic variance reduction for nonconvex optimization. In: International Conference on Machine Learning, pp. 314–323 (2016)
Rockafellar, R.T., Uryasev, S.: Optimization of conditional value-at-risk. J. Risk 2, 21–41 (2000)
Shapiro, A., Dentcheva, D., Ruszczynski, A.: Lectures on Stochastic Programming: Modeling and Theory, 2nd edn. Society for Industrial and Applied Mathematics, Philadelphia (2014)
Wang, X., Ma, S., Yuan, Y.: Penalty methods with stochastic approximation for stochastic nonlinear programming. Math. Comput. 86(306), 1793–1820 (2017)
Wang, Z., Ji, K., Zhou, Y., Liang, Y., Tarokh, V.: Spiderboost: A class of faster variance-reduced algorithms for nonconvex optimization. arXiv:1810.10690 (2018)
Xu, Y.: Iteration complexity of inexact augmented Lagrangian methods for constrained convex programming. Math. Program. (2019)
Yu, H., Neely, M., Wei, X.: Online convex optimization with stochastic constraints. Adv. Neural Inf. Process. Syst. pp. 1428–1438 (2017)
Zhou, D., Xu, P., Gu, Q.: Stochastic nested variance reduction for nonconvex optimization. In Proceedings of the 32nd International Conference on Neural Information Processing Systems (USA, 2018), NIPS’18, Curran Associates Inc., pp. 3925–3936
Acknowledgements
The authors would like to thank Dr. Qihang Lin for a few inspiring discussions that help to improve the initial version of this work. The authors would also like to thank an anonymous reviewer whose comments helped in a significant streamlining of the presentation of the paper.
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.
Boob and Lan was partially supported by the NSF Grant CCF 1909298. Deng was partially supported by the NSFC Grant 11831002.
Appendices
Proof of Proposition 1
Let us denote
It is easy to see that \({\bar{\psi }}_0(x)\) and \({\bar{\psi }}_i(x),\ i \in [m]\), are convex functions. Moreover, their respective subdifferentials can be written as
where \(\nabla W\) is the gradient in the first variable. Note that \(\nabla W(x^*, x^*) = {\varvec{0}}\). Consider the constrained convex optimization problem:
Note that \(x^*\) is a feasible solution of this problem. For sake of this proof, define \(\varPsi _k(x) := {\bar{\psi }}_0(x) + \tfrac{k}{2}\sum _{i=1}^m \left[ {\bar{\psi }}_i(x)\right] _+^2 +\tfrac{1}{2}\left\Vert x-x^*\right\Vert ^{2}_{2}\). Let \(S=\{x \in X: \left\Vert x-x^*\right\Vert ^{}_{2} \leqslant \varepsilon \}\) for some \(\varepsilon > 0\) such that any \(x \in S\) which is feasible for (A.2) satisfies \({\bar{\psi }}_0(x) \geqslant {\bar{\psi }}_0(x^*)\). Let \(x_k :=\hbox {argmin}_{x\in S} \varPsi _k(x)\). This is well-defined since \(\varPsi _k\) is a strongly convex function. Note that
where second inequality follows from the fact that \(x_k\) is optimal and \(x^* \in S\) is a feasible point. Note that as \(k \rightarrow \infty \), we have \(\limsup _{k\rightarrow \infty } \varPsi _k(x_k) < \infty \Rightarrow \limsup _{k\rightarrow \infty }{\bar{\psi }}(x_k) \leqslant 0\).
Moreover, note that \({{\,\mathrm{dom}\,}}(\liminf _{k\rightarrow \infty } \varPsi _k) \subseteq \{x: \psi _{i}(x) \leqslant 0, i \in [m]\}\). Also note that \({{\,\mathrm{dom}\,}}(\liminf _{k\rightarrow \infty } \varPsi _k) \cap S \ne \emptyset \) since both sets contain \(x^*\). Then, definition of set S implies \({\bar{\psi }}_0(x) \geqslant {\bar{\psi }}_0(x^*)\) for all \(x \in {{\,\mathrm{dom}\,}}(\liminf _{k\rightarrow \infty } \varPsi _k) \cap S\). Hence, \( \liminf _{k\rightarrow \infty } \varPsi _k(x_k) \geqslant \liminf _{k\rightarrow \infty } {\bar{\psi }}_0(x_k) \geqslant {\bar{\psi }}_0(x^*)\). This inequality with (A.3) implies that \(\lim _{k\rightarrow \infty } \varPsi _k(x_k) = \psi _0(x^*)\) and \(x_k \rightarrow x^*\). Hence, there exists \({\bar{k}}\) such that for all \(k > {\bar{k}}\), \(x_k \in {{\,\mathrm{int}\,}}(S)\). So for such k we can write the following first-order criterion for convex optimization (\(\left[ {\bar{\psi }}_{i}\right] _+^2\) is a convex function):
This implies that \(x_k\) is also the optimal solution of
For simplicity, let us denote \(v_k=k\,\left[ {\bar{\psi }}(x_k)\right] _+\). Due to the optimality of \(x_k\) of solving the above, we have
We claim that \(\{v_k\}\) is a bounded sequence. Indeed, if this is true, then we can find a convergent subsequence \(\{i_k\}\) with \(\lim _{k\rightarrow \infty } v_{i_k}=v^*\). Taking \(k\rightarrow \infty \) in (A.4), we have
Placing \(x=x^*\), we have \({\bar{\psi }}_0(x^*)\geqslant \limsup {\bar{\psi }}_0(x_{i_k})\), thus \(\lim _{k\rightarrow \infty }{\bar{\psi }}_0(x_{i_k})={\bar{\psi }}_0(x^*)\) based on the lower semicontinuity of \({\bar{\psi }}_0\). In view of this discussion, \(x^*\) optimizes the right side of (A.5). Thus, applying the first order criterion, we have
It remains to apply \(\partial {\bar{\psi }}_0(x^*)=\partial \psi _0(x^*)\) and \(\partial {\bar{\psi }}_i(x^*)=\partial \psi _i(x^*)\).
In addition, to prove complimentary slackness, it suffices to show when \({\bar{\psi }}_i(x^*)=\psi _i(x^*)<0\), we must have \({v^{(i)}}^*=0\). Since \(x_k\) converges to \(x^*\) and \({\bar{\psi }}_i\) is continuous, there exists some \(\exists k_0>0\), such that \({\bar{\psi }}_i(x_{i_k})<0\) when \(k>k_0\). Hence \({v^{(i)}_{i_k}}^*=0\) by its definition. Taking the limit, we have \({v^{(i)}}^*=0\).
It remains to show the missing piece, that \(\{v_k\}\) is a bounded sequence. We will prove by contradiction. If this is not true, we may assume \(\lim _{k\rightarrow \infty }\Vert v_k\Vert = \infty \), passing to a subsequence if necessary. Moreover, define \(y_k=v_k/\Vert v_k\Vert \), since \(y_k\) is a unit vector, it has some limit point, let us assume \(\lim _{k\rightarrow \infty }y_{j_k}=y^*\) for a subsequence \(\{j_k\}\). Dividing both sides of (A.4) by \(\Vert v_k\Vert \) and then passing it to the subsequence \(\{j_k\}\), we have
Taking \(k\rightarrow \infty \), we have
Since subsequence \(x_{j_k}\) converges to \(x^*\) and \({\bar{\psi }}_i\) is continuous, we see that \({\bar{\psi }}_i(x_{j_k}) < 0\) for any \(i \notin {\mathcal {A}}(x^*)\) for \(k \geqslant k_0\). This implies \(y_{j_k}= j_k \left[ {\bar{\psi }}_i(x_{j_k})\right] _+ = 0\) for all \( k \geqslant k_0\) and for all \(i \notin {\mathcal {A}}(x^*)\). So we must have \({\varvec{0}}\in N_X(x^*) + \sum _{i \in {\mathcal {A}}(x^*)} y^{*\left( i\right) } \partial \psi _i(x^*)\). Here, we have used the fact that \(\nabla W(x^*, x^*) = {\varvec{0}}\), implying that \(\partial {\bar{\psi }}_i(x^*) = \partial \psi _{i}(x^*)\) for all \(i = 0, \ldots , m\). Let \(u \in N_X(x^*)\) and \(g_i(x^*) \in \partial \psi _i(x^*), i \in {\mathcal {A}}(x^*)\) be such that
Then we can derive a contradiction by using MFCQ (Definition 3). Assume that z satisfies MFCQ (3). Therefore, we have
where first inequality follows since \(z \in -N_X^*(x^*)\) and \(u \in N_X(x^*)\) hence \(z^Tu \leqslant 0\), second inequality follows due to the fact that \(y^{*\left( i\right) } \geqslant 0\) and \(g_i(x^*) \in \partial \psi _i(x^*)\) and last strict inequality follows since (3) and \(y^{*\left( i\right) } > 0\) for at least one \(i \in {\mathcal {A}}(x^*)\).
Proof of Proposition 4
Let us define \({\bar{\psi }}_i, i = 0, \ldots , m\) as in (A.1) where \(x^*\) is a local solution of (1.1) then,
where the first implication follows from the fact that \({\bar{\psi }}_{i}(x) \geqslant \psi _{i}(x)\) for all \( i \in [m]\) or equivalently \(\{x \in X: {\bar{\psi }}_i(x) \leqslant 0, i \in [m],\ \left\Vert x-x^*\right\Vert ^{}_{}< \varepsilon \} \subseteq \{x \in X: \psi _i(x) \leqslant 0, i \in [m],\ \left\Vert x-x^*\right\Vert ^{}_{} < \varepsilon \}\), and second implication follows from the fact that \({\bar{\psi }}_i(x) \geqslant \psi _{i}(x)\).
The last statement implies that \(x^*\) is a local optimal solution for the convex problem (A.2). Hence, it is also a global optimal solution. Based on (3.39) from Assumption 3, we have,
Hence, by Slater condition, we have that there exists \(y^* \geqslant {\varvec{0}}\) such that \((x^*, y^*)\) satisfy first order KKT-condition for the convex problem (A.2). Thus, we have
It remains to apply \(\partial {\bar{\psi }}_i(x^*)=\partial \psi _i(x^*)\) and \({\bar{\psi }}_i(x^*) = \psi _i(x^*)\) for all \(i \in 0, \ldots , m\). Hence, we conclude the proof.
Rights and permissions
About this article
Cite this article
Boob, D., Deng, Q. & Lan, G. Stochastic first-order methods for convex and nonconvex functional constrained optimization. Math. Program. 197, 215–279 (2023). https://doi.org/10.1007/s10107-021-01742-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10107-021-01742-y
Keywords
- Functional constrained optimization
- Stochastic algorithms
- Convex and nonconvex optimization
- Acceleration