Geodesic Convexity Research Articles

Geodesic Convexity of the Symmetric Eigenvalue Problem and Convergence of Steepest Descent

We study the convergence of the Riemannian steepest descent algorithm on the Grassmann manifold for minimizing the block version of the Rayleigh quotient of a symmetric matrix. Even though this problem is non-convex in the Euclidean sense and only very locally convex in the Riemannian sense, we discover a structure for this problem that is similar to geodesic strong convexity, namely, weak-strong convexity. This allows us to apply similar arguments from convex optimization when studying the convergence of the steepest descent algorithm but with initialization conditions that do not depend on the eigengap δ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\delta $$\\end{document}. When δ>0\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\delta >0$$\\end{document}, we prove exponential convergence rates, while otherwise the convergence is algebraic. Additionally, we prove that this problem is geodesically convex in a neighbourhood of the global minimizer of radius O(δ)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\\mathcal {O}}(\\sqrt{\\delta })$$\\end{document}.

Graph convexity impartial games: Complexity and winning strategies

Accordingly to Duchet (1987), the first paper of convexity on general graphs, in english, is the 1981 paper “Convexity in graphs”. One of its authors, Frank Harary, introduced in 1984 the first graph convexity games, focused on the geodesic convexity, which were investigated in a sequence of five papers that ended in 2003. In this paper, we continue this research line, extend these games to other graph convexities, and obtain winning strategies and complexity results. Among them, we obtain winning strategies for general convex geometries and winning strategies for trees from the Sprague-Grundy theory on impartial games. We also obtain the first PSPACE-hardness results on convexity games, by proving that the normal play and the misère play of the impartial hull game on the geodesic convexity is PSPACE-complete even in graphs with diameter two.

Some convexity criteria for differentiable functions on the 2‐Wasserstein space

AbstractWe show that a differentiable function on the 2‐Wasserstein space is geodesically convex if and only if it is also convex along a larger class of curves which we call ‘acceleration‐free’. In particular, the set of acceleration‐free curves includes all generalised geodesics. We also show that geodesic convexity can be characterised through first‐ and second‐order inequalities involving the Wasserstein gradient and the Wasserstein Hessian. Subsequently, such inequalities also characterise convexity along acceleration‐free curves.

Differentially private Riemannian optimization

In this paper, we study the differentially private empirical risk minimization problem where the parameter is constrained to a Riemannian manifold. We introduce a framework for performing differentially private Riemannian optimization by adding noise to the Riemannian gradient on the tangent space. The noise follows a Gaussian distribution intrinsically defined with respect to the Riemannian metric on the tangent space. We adapt the Gaussian mechanism from the Euclidean space to the tangent space compatible to such generalized Gaussian distribution. This approach presents a novel analysis as compared to directly adding noise on the manifold. We further prove privacy guarantees of the proposed differentially private Riemannian (stochastic) gradient descent using an extension of the moments accountant technique. Overall, we provide utility guarantees under geodesic (strongly) convex, general nonconvex objectives as well as under the Riemannian Polyak-Łojasiewicz condition. Empirical results illustrate the versatility and efficacy of the proposed framework in several applications.

Optimal Transport and Generalized Ricci Flow

We prove results relating the theory of optimal transport and generalized Ricci flow. We define an adapted cost functional for measures using a solution of the associated dilaton flow. This determines a formal notion of geodesics in the space of measures, and we show geodesic convexity of an associated entropy functional. Finally, we show monotonicity of the cost along the backwards heat flow, and use this to give a new proof of the monotonicity of the energy functional along generalized Ricci flow.

Optimality conditions and duality for nonsmooth multiobjective semi-infinite programming problems with vanishing constraints on Hadamard manifolds

This article is concerned with nonsmooth multiobjective semi-infinite programming problems with vanishing constraints in the setting of Hadamard manifolds (abbreviated as, (NMSIPVC)). We present the Abadie constraint qualification (abbreviated as, (ACQ)) and a modified version of (ACQ), namely, (NMSIPVC)-tailored Abadie constraint qualification (abbreviated as, (NMSIPVC-ACQ)), for (NMSIPVC). The Karush-Kuhn-Tucker (abbreviated as, KKT) type necessary criteria of optimality for (NMSIPVC) are derived by employing (ACQ) and (NMSIPVC-ACQ). Moreover, sufficient criteria of optimality for (NMSIPVC) are derived under generalized geodesic convexity hypotheses and certain mild restrictions on the index sets. Further, we formulate Wolfe type and Mond-Weir type dual models related to (NMSIPVC) and derive several duality results that relate (NMSIPVC) and the corresponding dual models. Non-trivial numerical examples are incorporated to demonstrate the validity of the results established in this paper. To the best of our knowledge, this is for the first time that constraint qualifications, optimality conditions, and duality results have been investigated for (NMSIPVC) in the setting of Hadamard manifolds.

Integer programming models and polyhedral study for the geodesic classification problem on graphs

We study a discrete version of the classical classification problem in Euclidean space, to be called the geodesic classification problem. It is defined on a graph, where some vertices are initially assigned a class and the remaining ones must be classified. This vertex partition into classes is grounded on the concept of geodesic convexity on graphs, as a replacement for Euclidean convexity in the multidimensional space. We propose two new integer programming models along with branch-and-bound algorithms to solve them. We also carry out a polyhedral study of the associated polyhedra, which produced families of facet-defining inequalities and separation algorithms. Finally, we run computational experiments to evaluate the computational efficiency and the classification accuracy of the proposed approaches by comparing them with classic solution methods for the Euclidean convexity classification problem.

Pareto Efficiency Criteria and Duality for Multiobjective Fractional Programming Problems with Equilibrium Constraints on Hadamard Manifolds

This article deals with multiobjective fractional programming problems with equilibrium constraints in the setting of Hadamard manifolds (abbreviated as MFPPEC). The generalized Guignard constraint qualification (abbreviated as GGCQ) for MFPPEC is presented. Furthermore, the Karush–Kuhn–Tucker (abbreviated as KKT) type necessary criteria of Pareto efficiency for MFPPEC are derived using GGCQ. Sufficient criteria of Pareto efficiency for MFPPEC are deduced under some geodesic convexity hypotheses. Subsequently, Mond–Weir and Wolfe type dual models related to MFPPEC are formulated. The weak, strong, and strict converse duality results are derived relating MFPPEC and the respective dual models. Suitable nontrivial examples have been furnished to demonstrate the significance of the results established in this article. The results derived in the article extend and generalize several notable results previously existing in the literature. To the best of our knowledge, optimality conditions and duality for MFPPEC have not yet been studied in the framework of manifolds.

Second-Order Optimality Conditions and Duality for Multiobjective Semi-Infinite Programming Problems on Hadamard Manifolds

This paper is devoted to the study of multiobjective semi-infinite programming problems on Hadamard manifolds. We consider a class of multiobjective semi-infinite programming problems (abbreviated as MSIP) on Hadamard manifolds. We use the concepts of second-order Karush–Kuhn–Tucker stationary point and second-order Karush–Kuhn–Tucker geodesic pseudoconvexity of the considered problem to derive necessary and sufficient second-order conditions of efficiency, weak efficiency and proper efficiency for MSIP along with certain generalized geodesic convexity assumptions. Moreover, we formulate the second-order Mond–Weir-type dual problem related to MSIP and deduce weak and strong duality theorems relating MSIP and the dual problem. The significance of our results is demonstrated with the help of non-trivial examples. To the best of our knowledge, this is the first time that second-order optimality conditions for MSIP have been studied in Hadamard manifold setting.

Duality for Multiobjective Programming Problems with Equilibrium Constraints on Hadamard Manifolds under Generalized Geodesic Convexity

This article is devoted to the study of a class of multiobjective mathematical programming problems with equilibrium constraints on Hadamard manifolds (in short, (MPPEC)). We consider (MPPEC) as our primal problem and formulate two different kinds of dual models, namely, Wolfe and Mond-Weir type dual models related to (MPPEC). Further, we deduce the weak, strong as well as strict converse duality relations that relate (MPPEC) and the corresponding dual problems employing geodesic pseudoconvexity and geodesic quasiconvexity restrictions. Several suitable numerical examples are incorporated to demonstrate the significance of the deduced results. The results derived in this article generalize and extend several previously existing results in the literature.

Characterization of solution sets of geodesic convex semi-infinite programming on Riemannian manifolds

Characterization of solution sets of geodesic convex semi-infinite programming on Riemannian manifolds

New Generalization of Geodesic Convex Function

As a generalization of a geodesic function, this paper introduces the notion of the geodesic φE-convex function. Some properties of the φE-convex function and geodesic φE-convex function are established. The concepts of a geodesic φE-convex set and φE-epigraph are also given. The characterization of geodesic φE-convex functions in terms of their φE-epigraphs, are also obtained.

On Geodesic Convexity in Mycielskian of Graphs

The convexity induced by the geodesics in a graph G is called the geodesic convexity of G. Mycielski graphs preserve the property of being triangle-free and many parameters such as power domination number, coloring number, determining number and recently general position number have been determined for them. In this work, we determine the geodesic convexity parameters viz., convexity, geodetic iteration, geodetic, and hull numbers for Mycielski graphs for which the underlying graphs considered are path, cycle, star, and complete graph.

Transport Type Metrics on the Space of Probability Measures Involving Singular Base Measures

We develop the theory of a metric, which we call the $$\nu $$ -based Wasserstein metric and denote by $$W_\nu $$ , on the set of probability measures $${\mathcal {P}}(X)$$ on a domain $$X \subseteq \mathbb {R}^m$$ . This metric is based on a slight refinement of the notion of generalized geodesics with respect to a base measure $$\nu $$ and is relevant in particular for the case when $$\nu $$ is singular with respect to m-dimensional Lebesgue measure; it is also closely related to the concept of linearized optimal transport. The $$\nu $$ -based Wasserstein metric is defined in terms of an iterated variational problem involving optimal transport to $$\nu $$ ; we also characterize it in terms of integrations of classical Wasserstein metric between the conditional probabilities when measures are disintegrated with respect to optimal transport to $$\nu $$ , and through limits of certain multi-marginal optimal transport problems. We also introduce a class of metrics which are dual in a certain sense to $$W_\nu $$ , defined relative to a fixed based measure $$\mu $$ , on the set of measures which are absolutely continuous with respect to a second fixed based measure $$\sigma $$ . As we vary the base measure $$\nu $$ , the $$\nu $$ -based Wasserstein metric interpolates between the usual quadratic Wasserstein metric (obtained when $$\nu $$ is a Dirac mass) and a metric associated with the uniquely defined generalized geodesics obtained when $$\nu $$ is sufficiently regular (eg, absolutely continuous with respect to Lebesgue). When $$\nu $$ concentrates on a lower dimensional submanifold of $$\mathbb {R}^m$$ , we prove that the variational problem in the definition of the $$\nu $$ -based Wasserstein metric has a unique solution. We also establish geodesic convexity of the usual class of functionals, and of the set of source measures $$\mu $$ such that optimal transport between $$\mu $$ and $$\nu $$ satisfies a strengthening of the generalized nestedness condition introduced in McCann and Pass (Arch Ration Mech Anal 238(3):1475–1520, 2020). We finally introduce a slight variant of the dual metric mentioned above in order to prove convergence of an iterative scheme to solve a variational problem arising in game theory.

Complexity and winning strategies of graph convexity games (Brief Announcement)

Accordingly to Duchet (1987), the first paper of convexity on general graphs, in english, is the 1981 paper “Convexity in graphs”. One of its authors, Frank Harary, introduced in 1984 the first graph convexity games, focused on the geodesic convexity, which were investigated in a sequence of five papers that ended in 2003. In this paper, we continue this research line, extend these games to other graph convexities, and obtain winning strategies and complexity results. Among them, we obtain winning strategies for general convex geometries in graphs. We also obtain the first PSPACE-hardness results on convexity games, by proving that the normal play and the misère play of the hull game on the geodesic and the monophonic convexities are PSPACE-complete.

Geodesic Convexity of graphs: partition and geodetic sets

This work describes the contribution of Artigas, Dantas, Dourado, and Szwarcfiter on the topic of convexity in the context of geodesic convex partitions, covers and geodetic sets from 2008 to 2013.

Optimality Conditions for Multiobjective Mathematical Programming Problems with Equilibrium Constraints on Hadamard Manifolds

In this paper, we consider a class of multiobjective mathematical programming problems with equilibrium constraints on Hadamard manifolds (in short, (MMPEC)). We introduce the generalized Guignard constraint qualification for (MMPEC) and employ it to derive Karush–Kuhn–Tucker (KKT)-type necessary optimality criteria. Further, we derive sufficient optimality criteria for (MMPEC) using geodesic convexity assumptions. The significance of the results deduced in the paper has been demonstrated by suitable non-trivial examples. The results deduced in this article generalize several well-known results in the literature to a more general space, that is, Hadamard manifolds, and extend them to a more general class of optimization problems. To the best of our knowledge, this is the first time that generalized Guignard constraint qualification and optimality conditions have been studied for (MMPEC) in manifold settings.

Stochastic Zeroth-Order Riemannian Derivative Estimation and Optimization

We consider stochastic zeroth-order optimization over Riemannian submanifolds embedded in Euclidean space, where the task is to solve Riemannian optimization problems with only noisy objective function evaluations. Toward this, our main contribution is to propose estimators of the Riemannian gradient and Hessian from noisy objective function evaluations, based on a Riemannian version of the Gaussian smoothing technique. The proposed estimators overcome the difficulty of nonlinearity of the manifold constraint and issues that arise in using Euclidean Gaussian smoothing techniques when the function is defined only over the manifold. We use the proposed estimators to solve Riemannian optimization problems in the following settings for the objective function: (i) stochastic and gradient-Lipschitz (in both nonconvex and geodesic convex settings), (ii) sum of gradient-Lipschitz and nonsmooth functions, and (iii) Hessian-Lipschitz. For these settings, we analyze the oracle complexity of our algorithms to obtain appropriately defined notions of ϵ-stationary point or ϵ-approximate local minimizer. Notably, our complexities are independent of the dimension of the ambient Euclidean space and depend only on the intrinsic dimension of the manifold under consideration. We demonstrate the applicability of our algorithms by simulation results and real-world applications on black-box stiffness control for robotics and black-box attacks to neural networks. Funding: J. Li and S. Ma acknowledge the support of the National Science Foundation (NSF) [Grants DMS-1953210, CCF-1934568, and CCF-2007797]. K. Balasubramanian acknowledges the support of the NSF [Grant DMS-2053918]. K. Balasubramanian and S. Ma acknowledge the support of the UC Davis CeDAR (Center for Data Science and Artificial Intelligence Research) Innovative Data Science Seed Funding Program.

New Hadamard-type inequality for new class of geodesic convex functions

This paper aims to introduce the concept of (\( E,\mu,\kappa \))-convex function by using special inequality. Hadamard integral inequality for this new class of geodesic convex function in the case of Lebesgue and Sugeno integrals is given.

A multiscale analysis of multi-agent coverage control algorithms

This paper presents a theoretical framework for the design and analysis of gradient descent-based algorithms for coverage control tasks involving robot swarms. We adopt a multiscale approach to analysis and design to ensure consistency of the algorithms in the large-scale limit. First, we represent the macroscopic configuration of the swarm as a probability measure and formulate the macroscopic coverage task as the minimization of a convex objective function over probability measures. We then construct a macroscopic dynamics for swarm coverage, which takes the form of a proximal descent scheme in the L2-Wasserstein space. Our analysis exploits the generalized geodesic convexity of the coverage objective function, proving convergence in the L2-Wasserstein sense to the target probability measure. We then obtain a consistent gradient descent algorithm in the Euclidean space that is implementable by a finite collection of agents, via a “variational” discretization of the macroscopic coverage objective function. We establish the convergence properties of the gradient descent and its behavior in the continuous-time and large-scale limits. Furthermore, we establish a connection with well-known Lloyd-based algorithms, seen as a particular class of algorithms within our framework, and demonstrate our results via numerical experiments.