All issues

In this work we consider Monteiro – Svaiter accelerated hybrid proximal extragradient (A-HPE) framework and accelerated Newton proximal extragradient (A-NPE) framework. The last framework contains an optimal method for rather smooth convex optimization problems with second-order oracle. We generalize A-NPE framework for higher order derivative oracle (schemes). We replace Newton’s type step in A-NPE that was used for auxiliary problem by Newton’s regularized (tensor) type step (Yu. Nesterov, 2018). Moreover we generalize large step A-HPE/A-NPE framework by replacing Monteiro – Svaiter’s large step condition so that this framework could work for high-order schemes. The main contribution of the paper is as follows: we propose optimal highorder methods for convex optimization problems. As far as we know for that moment there exist only zero, first and second order optimal methods that work according to the lower bounds. For higher order schemes there exists a gap between the lower bounds (Arjevani, Shamir, Shiff, 2017) and existing high-order (tensor) methods (Nesterov – Polyak, 2006; Yu.Nesterov, 2008; M. Baes, 2009; Yu.Nesterov, 2018). Asymptotically the ratio of the rates of convergences for the best existing methods and lower bounds is about 1.5. In this work we eliminate this gap and show that lower bounds are tight. We also consider rather smooth strongly convex optimization problems and show how to generalize the proposed methods to this case. The basic idea is to use restart technique until iteration sequence reach the region of quadratic convergence of Newton method and then use Newton method. One can show that the considered method converges with optimal rates up to a logarithmic factor. Note, that proposed in this work technique can be generalized in the case when we can’t solve auxiliary problem exactly, moreover we can’t even calculate the derivatives of the functional exactly. Moreover, the proposed technique can be generalized to the composite optimization problems and in particular to the constraint convex optimization problems. We also formulate a list of open questions that arise around the main result of this paper (optimal universal method of high order e.t.c.).

In this paper, we consider conditional gradient methods for optimizing strongly convex functions. These are methods that use a linear minimization oracle, which, for a given vector $p \in \mathbb{R}^n$, computes the solution of the subproblem

\[ \text{Argmin}_{x\in X}{\langle p,\,x \rangle}. \]There are a variety of conditional gradient methods that have a linear convergence rate in a strongly convex case. However, in all these methods, the dimension of the problem is included in the rate of convergence, which in modern applications can be very large. In this paper, we prove that in the strongly convex case, the convergence rate of the conditional gradient methods in the best case depends on the dimension of the problem $ n $ as $ \widetilde {\Omega} \left(\!\sqrt {n}\right) $. Thus, the conditional gradient methods may turn out to be ineffective for solving strongly convex optimization problems of large dimensions.

Also, the application of conditional gradient methods to minimization problems of a quadratic form is considered. The effectiveness of the Frank – Wolfe method for solving the quadratic optimization problem in the convex case on a simplex (PageRank) has already been proved. This work shows that the use of conditional gradient methods to solve the minimization problem of a quadratic form in a strongly convex case is ineffective due to the presence of dimension in the convergence rate of these methods. Therefore, the Shrinking Conditional Gradient method is considered. Its difference from the conditional gradient methods is that it uses a modified linear minimization oracle. It's an oracle, which, for a given vector $p \in \mathbb{R}^n$, computes the solution of the subproblem \[ \text{Argmin}\{\langle p, \,x \rangle\colon x\in X, \;\|x-x_0^{}\| \leqslant R \}. \] The convergence rate of such an algorithm does not depend on dimension. Using the Shrinking Conditional Gradient method the complexity (the total number of arithmetic operations) of solving the minimization problem of quadratic form on a $ \infty $-ball is obtained. The resulting evaluation of the method is comparable to the complexity of the gradient method.

In this paper we discuss lower bounds for convergence of convex optimization methods of high order and attainability of this bounds. We formulate a hypothesis that covers all the cases. It is noticeable that we provide this statement without a proof. Newton method is the most famous method that uses gradient and Hessian of optimized function. However, it converges locally even for strongly convex functions. Global convergence can be achieved with cubic regularization of Newton method [Nesterov, Polyak, 2006], whose iteration cost is comparable with iteration cost of Newton method and is equivalent to inversion of Hessian of optimized function. Yu.Nesterov proposed accelerated variant of Newton method with cubic regularization in 2008 [Nesterov, 2008]. R.Monteiro and B. Svaiter managed to improve global convergence of cubic regularized method in 2013 [Monteiro, Svaiter, 2013]. Y.Arjevani, O. Shamir and R. Shiff showed that convergence bound of Monteiro and Svaiter is optimal (cannot be improved by more than logarithmic factor with any second order method) in 2017 [Arjevani et al., 2017]. They also managed to find bounds for convex optimization methods of p-th order for $p ≥ 2$. However, they got bounds only for first and second order methods for strongly convex functions. In 2018 Yu.Nesterov proposed third order convex optimization methods with rate of convergence that is close to this lower bounds and with similar to Newton method cost of iteration [Nesterov, 2018]. Consequently, it was showed that high order methods can be practical. In this paper we formulate lower bounds for p-th order methods for $p ≥ 3$ for strongly convex unconstrained optimization problems. This paper can be viewed as a little survey of state of the art of high order optimization methods.

The paper is devoted to the problem of minimization of the non-smooth functional $f$ with a non-positive non-smooth Lipschitz-continuous functional constraint. We consider the formulation of the problem in the case of quasi-convex functionals. We propose new strategies of step-sizes and adaptive stopping rules in Mirror Descent for the considered class of problems. It is shown that the methods are applicable to the objective functionals of various levels of smoothness. Applying a special restart technique to the considered version of Mirror Descent there was proposed an optimal method for optimization problems with strongly convex objective functionals. Estimates of the rate of convergence for the considered methods are obtained depending on the level of smoothness of the objective functional. These estimates indicate the optimality of the considered methods from the point of view of the theory of lower oracle bounds. In particular, the optimality of our approach for Höldercontinuous quasi-convex (sub)differentiable objective functionals is proved. In addition, the case of a quasiconvex objective functional and functional constraint was considered. In this paper, we consider the problem of minimizing a non-smooth functional $f$ in the presence of a Lipschitz-continuous non-positive non-smooth functional constraint $g$, and the problem statement in the cases of quasi-convex and strongly (quasi-)convex functionals is considered separately. The paper presents numerical experiments demonstrating the advantages of using the considered methods.

An essential class of problems describes physical processes occurring in non-convex domains containing a corner greater than 180 degrees on the boundary. The solution in a neighborhood of a corner is singular and its finding using classical approaches entails a loss of accuracy. In the paper, we consider stationary, linearized by Picard’s iterations, Navier – Stokes equations governing the flow of a incompressible viscous fluid in the convection form in $L$-shaped domain. An $R_\nu$-generalized solution of the problem in special sets of weighted spaces is defined. A special finite element method to find an approximate $R_\nu$-generalized solution is constructed. Firstly, functions of the finite element spaces satisfy the law of conservation of mass in the strong sense, i.e. at the grid nodes. For this purpose, Scott – Vogelius element pair is used. The fulfillment of the condition of mass conservation leads to the finding more accurate, from a physical point of view, solution. Secondly, basis functions of the finite element spaces are supplemented by weight functions. The degree of the weight function, as well as the parameter $\nu$ in the definition of an $R_\nu$-generalized solution, and a radius of a neighborhood of the singularity point are free parameters of the method. A specially selected combination of them leads to an increase almost twice in the order of convergence rate of an approximate solution to the exact one in relation to the classical approaches. The convergence rate reaches the first order by the grid step in the norms of Sobolev weight spaces. Thus, numerically shown that the convergence rate does not depend on the corner value.

The article considers minimization of the expectation of convex function. Problems of this type often arise in machine learning and a variety of other applications. In practice, stochastic gradient descent (SGD) and similar procedures are usually used to solve such problems. We propose to use the ellipsoid method with mini-batching, which converges linearly and can be more efficient than SGD for a class of problems. This is verified by our experiments, which are publicly available. The algorithm does not require neither smoothness nor strong convexity of the objective to achieve linear convergence. Thus, its complexity does not depend on the conditional number of the problem. We prove that the method arrives at an approximate solution with given probability when using mini-batches of size proportional to the desired accuracy to the power −2. This enables efficient parallel execution of the algorithm, whereas possibilities for batch parallelization of SGD are rather limited. Despite fast convergence, ellipsoid method can result in a greater total number of calls to oracle than SGD, which works decently with small batches. Complexity is quadratic in dimension of the problem, hence the method is suitable for relatively small dimensionalities.

The paper is devoted to convex-concave saddle point problems where the objective is a sum of a large number of functions. Such problems attract considerable attention of the mathematical community due to the variety of applications in machine learning, including adversarial learning, adversarial attacks and robust reinforcement learning, to name a few. The individual functions in the sum usually represent losses related to examples from a data set. Additionally, the formulation admits a possibly nonsmooth composite term. Such terms often reflect regularization in machine learning problems. We assume that the dimension of one of the variable groups is relatively small (about a hundred or less), and the other one is large. This case arises, for example, when one considers the dual formulation for a minimization problem with a moderate number of constraints. The proposed approach is based on using Vaidya’s cutting plane method to minimize with respect to the outer block of variables. This optimization algorithm is especially effective when the dimension of the problem is not very large. An inexact oracle for Vaidya’s method is calculated via an approximate solution of the inner maximization problem, which is solved by the accelerated variance reduced algorithm Katyusha. Thus, we leverage the structure of the problem to achieve fast convergence. Separate complexity bounds for gradients of different components with respect to different variables are obtained in the study. The proposed approach is imposing very mild assumptions about the objective. In particular, neither strong convexity nor smoothness is required with respect to the low-dimensional variable group. The number of steps of the proposed algorithm as well as the arithmetic complexity of each step explicitly depend on the dimensionality of the outer variable, hence the assumption that it is relatively small.

The work is devoted to the construction of efficient and applicable to real tasks first-order methods of convex optimization, that is, using only values of the target function and its derivatives. Construction uses OGMG, fast gradient method which is optimal by complexity, but requires to know the Lipschitz constant for gradient and the strong convexity constant to determine the number of steps and step length. This requirement makes practical usage very hard. An adaptive on the constant for strong convexity algorithm ACGM is proposed, based on restarts of the OGM-G with update of the strong convexity constant estimate, and an adaptive on the Lipschitz constant for gradient ALGM, in which the use of OGM-G restarts is supplemented by the selection of the Lipschitz constant with verification of the smoothness conditions used in the universal gradient descent method. This eliminates the disadvantages of the original method associated with the need to know these constants, which makes practical usage possible. Optimality of estimates for the complexity of the constructed algorithms is proved. To verify the results obtained, experiments on model functions and real tasks from machine learning are carried out.

First-order optimization methods are workhorses in a wide range of modern applications in economics, physics, biology, machine learning, control, and other fields. Among other first-order methods accelerated and momentum ones obtain special attention because of their practical efficiency. The heavy-ball method (HB) is one of the first momentum methods. The method was proposed in 1964 and the first analysis was conducted for quadratic strongly convex functions. Since then a number of variations of HB have been proposed and analyzed. In particular, HB is known for its simplicity in implementation and its performance on nonconvex problems. However, as other momentum methods, it has nonmonotone behavior, and for optimal parameters, the method suffers from the so-called peak effect. To address this issue, in this paper, we consider an averaged version of the heavy-ball method (AHB). We show that for quadratic problems AHB has a smaller maximal deviation from the solution than HB. Moreover, for general convex and strongly convex functions, we prove non-accelerated rates of global convergence of AHB, its weighted version WAHB, and for AHB with restarts R-AHB. To the best of our knowledge, such guarantees for HB with averaging were not explicitly proven for strongly convex problems in the existing works. Finally, we conduct several numerical experiments on minimizing quadratic and nonquadratic functions to demonstrate the advantages of using averaging for HB. Moreover, we also tested one more modification of AHB called the tail-averaged heavy-ball method (TAHB). In the experiments, we observed that HB with a properly adjusted averaging scheme converges faster than HB without averaging and has smaller oscillations.

We consider one of the problems of transport modelling — searching the equilibrium distribution of traffic flows in the network. We use the classic Beckman’s model to describe time costs and flow distribution in the network represented by directed graph. Meanwhile agents’ behavior is not completely rational, what is described by the introduction of Markov logit dynamics: any driver selects a route randomly according to the Gibbs’ distribution taking into account current time costs on the edges of the graph. Thus, the problem is reduced to searching of the stationary distribution for this dynamics which is a stochastic Nash – Wardrope equilibrium in the corresponding population congestion game in the transport network. Since the game is potential, this problem is equivalent to the problem of minimization of some functional over flows distribution. The stochasticity is reflected in the appearance of the entropy regularization, in contrast to non-stochastic case. The dual problem is constructed to obtain a solution of the optimization problem. The universal primal-dual gradient method is applied. A major specificity of this method lies in an adaptive adjustment to the local smoothness of the problem, what is most important in case of the complex structure of the objective function and an inability to obtain a prior smoothness bound with acceptable accuracy. Such a situation occurs in the considered problem since the properties of the function strongly depend on the transport graph, on which we do not impose strong restrictions. The article describes the algorithm including the numerical differentiation for calculation of the objective function value and gradient. In addition, the paper represents a theoretical estimate of time complexity of the algorithm and the results of numerical experiments conducted on a small American town.