A relation between Γ-convergence of functionals and their associated gradient flows

Huaiyu Jian

1999-01-01

Abstract:Giorgi conjectured in 1979 that if a sequence of functionals converges in the sense of Γ-convergence to a limiting functional, then the corresponding gradient flows will converge as well after changing timescale appropriately. It is shown that this conjecture holds true for a rather wide kind of functionals.

What problem does this paper attempt to address?

A Relation Between Г-convergence of Functionals and Their Associated Gradient Flows

Huaiyu Jian

DOI: https://doi.org/10.1007/bf02876564

1999-01-01

Science in China Series A Mathematics

Abstract:Giorgi conjectured in 1979 that if a sequence of functionals converges in the sense of Г-convergence to a limiting functional, then the corresponding gradient flows will converge as well after changing timescale appropriately. It is shown that this conjecture holds true for a rather wide kind of functionals.
On a Complex Hessian Flow

Weimin Sheng,Jiaxiang Wang

DOI: https://doi.org/10.2140/pjm.2019.300.159

2019-01-01

Pacific Journal of Mathematics

Abstract:We consider a gradient flow generated by a complex Hessian functional which is defined on compact Kahler manifolds. By setting up the a priori estimates of the admissible solutions, we prove the long-time existence of the solution to the flow and its convergence. Thus we show the functional admits a local minimal point in the space of admissible functions. As its application, we show the solvability of a class of complex Hessian equations.
Γ-CONVERGENCE OF INTEGRAL FUNCTIONALS DEPENDING ON VECTOR-VALUED FUNCTIONS OVER PARABOLIC DOMAINS

HY Jian

DOI: https://doi.org/10.1142/s0252959900000285

2000-01-01

Chinese Annals of Mathematics Series B

Abstract:This paper studies Gamma-convergence for a sequence of parabolic functionals,F-epsilon (u) = integral(0)(T) integral(Omega) f(x/epsilon, t, del u)dxdt as epsilon -->0,where the integrand f is nonconvex, and periodic on the first variable. The author obtains the representation formula of the Gamma-limit. The results in this paper support a conclusion which relates Gamma-convergence of parabolic functionals to the associated gradient flows and confirms one of De Giorgi's conjectures partially.
Gamma-convergence of a gradient-flow structure to a non-gradient-flow structure

Mark A. Peletier,Mikola C. Schlottke

DOI: https://doi.org/10.1007/s00526-022-02190-y

2022-04-07

Calculus of Variations and Partial Differential Equations

Abstract:Abstract We study the asymptotic behaviour of a gradient system in a regime in which the driving energy becomes singular. For this system gradient-system convergence concepts are ineffective. We characterize the limiting behaviour in a different way, by proving $$\Gamma $$ Γ -convergence of the so-called energy-dissipation functional , which combines the gradient-system components of energy and dissipation in a single functional. The $$\Gamma $$ Γ -limit of these functionals again characterizes a variational evolution, but this limit functional is not the energy-dissipation functional of any gradient system. The system in question describes the diffusion of a particle in a one-dimensional double-well energy landscape, in the limit of small noise. The wells have different depth, and in the small-noise limit the process converges to a Markov process on a two-state system, in which jumps only happen from the higher to the lower well. This transmutation of a gradient system into a variational evolution of non-gradient type is a model for how many one-directional chemical reactions emerge as limit of reversible ones. The $$\Gamma $$ Γ -convergence proved in this paper both identifies the ‘fate’ of the gradient system for these reactions and the variational structure of the limiting irreversible reactions.
Gradient flows on graphons: existence, convergence, continuity equations

Sewoong Oh,Soumik Pal,Raghav Somani,Raghavendra Tripathi

2023-06-30

Abstract:Wasserstein gradient flows on probability measures have found a host of applications in various optimization problems. They typically arise as the continuum limit of exchangeable particle systems evolving by some mean-field interaction involving a gradient-type potential. However, in many problems, such as in multi-layer neural networks, the so-called particles are edge weights on large graphs whose nodes are exchangeable. Such large graphs are known to converge to continuum limits called graphons as their size grow to infinity. We show that the Euclidean gradient flow of a suitable function of the edge-weights converges to a novel continuum limit given by a curve on the space of graphons that can be appropriately described as a gradient flow or, more technically, a curve of maximal slope. Several natural functions on graphons, such as homomorphism functions and the scalar entropy, are covered by our set-up, and the examples have been worked out in detail.

Probability,Machine Learning
Local Conditions for Global Convergence of Gradient Flows and Proximal Point Sequences in Metric Spaces

Lorenzo Dello Schiavo,Jan Maas,Francesco Pedrotti

DOI: https://doi.org/10.1090/tran/9156

2023-04-11

Abstract:This paper deals with local criteria for the convergence to a global minimiser for gradient flow trajectories and their discretisations. To obtain quantitative estimates on the speed of convergence, we consider variations on the classical Kurdyka--Łojasiewicz inequality for a large class of parameter functions. Our assumptions are given in terms of the initial data, without any reference to an equilibrium point. The main results are convergence statements for gradient flow curves and proximal point sequences to a global minimiser, together with sharp quantitative estimates on the speed of convergence. These convergence results apply in the general setting of lower semicontinuous functionals on complete metric spaces, generalising recent results for smooth functionals on $\mathbb{R}^n$. While the non-smooth setting covers very general spaces, it is also useful for (non)-smooth functionals on $\mathbb{R}^n$.

Optimization and Control
Non-sufficiency of smoothness in the gradient conjecture

Florian Gruen

2024-08-20

Abstract:It is well known that for analytic cost functions, gradient flow trajectories have finite length and converge to a single critical point. The gradient conjecture of R. Thom states that, again for analytic cost functions, whenever the gradient flow trajectory converges, the limit of its unit secants exists. One might think that already the convergence of the gradient flow trajectory to a critical point is enough to ensure that the unit secants have a limit, but this does not hold in general - the gradient conjecture is to a certain extend sharp. We provide a counterexample in case of the missing analyticity assumption, that is a smooth (non-analytic) cost function $f$, where the limit of unit secants does not exist. In addition, $f$ satisfies even a strong geometric length-distance convergence property.

Dynamical Systems
CONVERGENCE OF THE FAMILY OF EULER-HALLEY TYPE METHODS ON RIEMANNIAN MANIFOLDS UNDER THE Γ-Condition

Jin-Hua Wang,Chong Li

DOI: https://doi.org/10.11650/twjm/1500405357

2009-01-01

Taiwanese Journal of Mathematics

Abstract:A convergence criterion of the family of Euler-Halley type methods for the vector fields on Riemannian manifolds whose covariant derivatives satisfy the $\gamma$-condition is established. The corresponding results due to \cite{H} are extended. An application to analytic vector fields is provided.
Existence, uniqueness and L2t(H2x) ∩ L∞t(H1x) ∩ H1t(L2x) regularity of the gradient flow of the Ambrosio–Tortorelli functional

Tommaso Cortopassi

DOI: https://doi.org/10.1051/cocv/2024060

2024-10-07

ESAIM Control Optimisation and Calculus of Variations

Abstract:We consider the gradient flow of the Ambrosio–Tortorelli functional, proving existence, uniqueness and L 2 t( H 2 x) ∩ L ∞ t( H 1 x) ∩ H 1 t( L 2 x) regularity of the solution in dimension 2. Such functional is an approximation in the sense of Γ-convergence of the Mumford–Shah functional often used in problems of image segmentation and fracture mechanics. The strategy of the proof essentially follows the one of []Feng and Prohl, M2AN Math. Model. Numer. Anal. 38 (2004) 291–320] but the crucial estimate is attained employing a different technique, and in the end it allows to prove better estimates than the ones obtained in [Feng and Prohl, M2AN Math. Model. Numer. Anal. 38 (2004) 291–320]. In particular we prove that if U ⊂ R 2 is a bounded C 2 domain, the initial data ( u 0, z 0) ∈ [ H 1( U)] 2 with 0 ≤ z 0 ≤ 1, then for every T > 0 there exists a unique gradient flow ( u( t), z( t)) of the Ambrosio–Tortorelli functional such that ( u, z) ∈ [ L 2(0, T; H 2( U)) ∩ L ∞(0, T; H 1( U)) ∩ H 1(0, T; L 2( U)] 2 while previously such regularity was known only for short times.

mathematics, applied,automation & control systems
A qualitative difference between gradient flows of convex functions in finite- and infinite-dimensional Hilbert spaces

Jonathan W. Siegel,Stephan Wojtowytsch

2023-10-27

Abstract:We consider gradient flow/gradient descent and heavy ball/accelerated gradient descent optimization for convex objective functions. In the gradient flow case, we prove the following:

Machine Learning,Classical Analysis and ODEs,Numerical Analysis
Convergence and non-convergence in a nonlocal gradient flow

Sangmin Park,Robert L. Pego

2024-09-13

Abstract:We study the asymptotic convergence of solutions as $t\rightarrow\infty$ of $\partial_t u=-f(u)+\int f(u)$, a nonlocal differential equation that is formally a gradient flow in a constant-mass subspace of $L^2$ arising from simplified models of phase transitions. In case the solution takes finitely many values, we provide a new proof of stabilization that uses a Łojasiewicz-type gradient inequality near a degenerate curve of equilibria. Solutions with infinitely many values in general need not converge to equilibrium, however, which we demonstrate by providing counterexamples for piecewise linear and cubic functions $f$. Curiously, the exponential rate of convergence in the finite-value case can jump from order $O(1)$ to arbitrarily small values upon perturbation of parameters.

Classical Analysis and ODEs,Analysis of PDEs,Dynamical Systems
Gradient descent for unbounded convex functions on Hadamard manifolds and its applications to scaling problems

Hiroshi Hirai,Keiya Sakabe

2024-04-15

Abstract:In this paper, we study asymptotic behaviors of continuous-time and discrete-time gradient flows of a ``lower-unbounded" convex function $f$ on a Hadamard manifold $M$, particularly, their convergence properties to the boundary $M^{\infty}$ at infinity of $M$. We establish a duality theorem that the infimum of the gradient-norm $\|\nabla f(x)\|$ of $f$ over $M$ is equal to the supremum of the negative of the recession function $f^{\infty}$ of $f$ over the boundary $M^{\infty}$, provided the infimum is positive. Further, the infimum and the supremum are obtained by the limits of the gradient flows of $f$, Our results feature convex-optimization ingredients of the moment-weight inequality for reductive group actions by Georgoulas, Robbin, and Salamon,and are applied to noncommutative optimization by Bürgisser et al. FOCS 2019. We show that the gradient descent of the Kempf-Ness function for an unstable orbit converges to a 1-parameter subgroup in the Hilbert-Mumford criterion, and the associated moment-map sequence converges to the mimimum-norm point of the moment polytope. We show further refinements for operator scaling -- the left-right action on a matrix tuple $A= (A_1,A_2,\ldots,A_N)$. We characterize the gradient-flow limit of operator scaling via a vector-space generalization of the classical Dulmage-Mendelsohn decomposition of a bipartite graph. Also, for a special case of $N = 2$, we reveal that this limit determines the Kronecker canonical form of matrix pencils $s A_1+A_2$.

Optimization and Control,Differential Geometry
Gamma-convergence of quadratic functionals perturbed by bounded linear functionals

Gianni Dal Maso,Davide Donati

DOI: https://doi.org/10.1016/j.na.2023.113395

IF: 1.743

2024-01-01

Nonlinear Analysis

Abstract:Given a bounded open set Ω ⊂ R n , we study sequences of quadratic functionals on the Sobolev space H 0 1 ( Ω ) , perturbed by sequences of bounded linear functionals. We prove that their Γ -limits, in the weak topology of H 0 1 ( Ω ) , can always be written as the sum of a quadratic functional, a linear functional, and a non-positive constant. The classical theory of G - and H -convergence completely characterises the quadratic and linear parts of the Γ -limit and shows that their coefficients do not depend on Ω . The constant, which instead depends on Ω and will be denoted by − ν ( Ω ) , plays an important role in the study of the limit behaviour of the energies of the solutions. The main result of this paper is that, passing to a subsequence, we can prove that ν coincides with a non-negative Radon measure on a sufficiently large collection of bounded open sets Ω . Moreover, we exhibit an example that shows that the previous result cannot be obtained for every bounded open set. The specific form of this example shows that the compactness theorem for the localisation method in Γ -convergence cannot be easily improved.

mathematics, applied
Gradient flows in asymmetric metric spaces and applications

Shin-ichi Ohta,Wei Zhao

2023-07-20

Abstract:This paper is devoted to the investigation of gradient flows in asymmetric metric spaces (for example, irreversible Finsler manifolds and Minkowski normed spaces) by means of discrete approximation. We study basic properties of curves and upper gradients in asymmetric metric spaces, and establish the existence of a curve of maximal slope, which is regarded as a gradient curve in the non-smooth setting. { Introducing} a natural convexity assumption on the potential function, { which is called the $(p,\lambda)$-convexity,} we also obtain some regularizing effects on the asymptotic behavior of curves of maximal slope. Applications include several existence results for gradient flows in Finsler manifolds, doubly nonlinear differential evolution equations on { infinite-dimensional Funk spaces}, and heat flow on compact Finsler manifolds.

Differential Geometry
Convergence Analysis of Gradient Algorithms on Riemannian Manifolds Without Curvature Constraints and Application to Riemannian Mass

Jinhua Wang,Xiangmei Wang,Chong Li,Jen-Chih Yao

DOI: https://doi.org/10.1137/19m1289285

IF: 2.763

2021-01-01

SIAM Journal on Optimization

Abstract:We study the convergence issue for the gradient algorithm (employing general step sizes) for optimization problems on general Riemannian manifolds (without curvature constraints). Under the assumption of the local convexity/quasi-convexity (resp. weak sharp minima), local/global convergence (resp. linear convergence) results are established. As an application, the linear convergence properties of the gradient algorithm employing the constant step sizes and the Armijo step sizes for finding the Riemannian $L^p$ ($p\in[1,+\infty)$) centers of mass are explored, respectively, which in particular extend and/or improve the corresponding results in \cite{Afsari2013}.
Gradient Flows and Riemannian Structure in the Gromov-Wasserstein Geometry

Zhengxin Zhang,Ziv Goldfeld,Kristjan Greenewald,Youssef Mroueh,Bharath K. Sriperumbudur

2024-07-16

Abstract:The Wasserstein space of probability measures is known for its intricate Riemannian structure, which underpins the Wasserstein geometry and enables gradient flow algorithms. However, the Wasserstein geometry may not be suitable for certain tasks or data modalities. Motivated by scenarios where the global structure of the data needs to be preserved, this work initiates the study of gradient flows and Riemannian structure in the Gromov-Wasserstein (GW) geometry, which is particularly suited for such purposes. We focus on the inner product GW (IGW) distance between distributions on $\mathbb{R}^d$. Given a functional $\mathsf{F}:\mathcal{P}_2(\mathbb{R}^d)\to\mathbb{R}$ to optimize, we present an implicit IGW minimizing movement scheme that generates a sequence of distributions $\{\rho_i\}_{i=0}^n$, which are close in IGW and aligned in the 2-Wasserstein sense. Taking the time step to zero, we prove that the discrete solution converges to an IGW generalized minimizing movement (GMM) $(\rho_t)_t$ that follows the continuity equation with a velocity field $v_t\in L^2(\rho_t;\mathbb{R}^d)$, specified by a global transformation of the Wasserstein gradient of $\mathsf{F}$. The transformation is given by a mobility operator that modifies the Wasserstein gradient to encode not only local information, but also global structure. Our gradient flow analysis leads us to identify the Riemannian structure that gives rise to the intrinsic IGW geometry, using which we establish a Benamou-Brenier-like formula for IGW. We conclude with a formal derivation, akin to the Otto calculus, of the IGW gradient as the inverse mobility acting on the Wasserstein gradient. Numerical experiments validating our theory and demonstrating the global nature of IGW interpolations are provided.

Analysis of PDEs,Optimization and Control,Machine Learning
On the global convergence of Wasserstein gradient flow of the Coulomb discrepancy

Siwan Boufadène,François-Xavier Vialard

2024-01-29

Abstract:In this work, we study the Wasserstein gradient flow of the Riesz energy defined on the space of probability measures. The Riesz kernels define a quadratic functional on the space of measure which is not in general geodesically convex in the Wasserstein geometry, therefore one cannot conclude to global convergence of the Wasserstein gradient flow using standard arguments. Our main result is the exponential convergence of the flow to the minimizer on a closed Riemannian manifold under the condition that the logarithm of the source and target measures are H{ö}lder continuous. To this goal, we first prove that the Polyak-Lojasiewicz inequality is satisfied for sufficiently regular solutions. The key regularity result is the global in-time existence of H{ö}lder solutions if the initial and target data are H{ö}lder continuous, proven either in Euclidean space or on a closed Riemannian manifold. For general measures, we prove using flow interchange techniques that there is no local minima other than the global one for the Coulomb kernel. In fact, we prove that a Lagrangian critical point of the functional for the Coulomb (or Energy distance) kernel is equal to the target everywhere except on singular sets with empty interior. In addition, singular enough measures cannot be critical points.

Analysis of PDEs
Stability of the heat flow under convergence in concentration and consequences

Nicola Gigli,Simone Vincini

2024-10-07

Abstract:We extend to the framework of convergence in concentration virtually all the results concerning stability of Sobolev functions and differential operators known to be in place under the stronger measured-Gromov-Hausdorff convergence. These include, in particular: i) A general $\Gamma$--$\varlimsup$ inequality for the Cheeger energy, ii) Convergence of the heat flow under a uniform ${\sf CD}(K,\infty)$ condition on the spaces. As we will show, building on ideas developed for mGH-convergence, out of this latter result we can obtain clean stability statements for differential, flows of vector fields, Hessian, eigenvalues of the Laplacian and related objects in presence of uniform lower Ricci bounds. At the technical level, one of the tools we develop to establish these results is the notion of convergence of maps between metric measure spaces converging in concentration. Previous results in the field concerned the stability of the ${\sf CD}$ (Funano-Shioya '13) and ${\sf RCD}$ (Ozawa-Yokota '19) conditions. The latter was obtained proving $\Gamma$-convergence for the Cheeger energies. This is not sufficient to pass to the limit in the heat flow; one of the byproducts of our work is the improvement of this to Mosco-convergence.

Metric Geometry,Analysis of PDEs
Discrete-time gradient flows in Gromov hyperbolic spaces

Shin-ichi Ohta

2023-06-22

Abstract:We investigate fundamental properties of the proximal point algorithm for Lipschitz convex functions on (proper, geodesic) Gromov hyperbolic spaces. We show that the proximal point algorithm from an arbitrary initial point can find a point close to a minimizer of the function. Moreover, we establish contraction estimates (akin to trees) for the proximal (resolvent) operator. Our results can be applied to small perturbations of trees.

Optimization and Control,Metric Geometry
Unilateral gradient flow of the Ambrosio-Tortorelli functional by minimizing movements

Jean-Francois Babadjian,Vincent Millot

DOI: https://doi.org/10.48550/arXiv.1207.3687

2013-10-25

Abstract:Motivated by models of fracture mechanics, this paper is devoted to the analysis of unilateral gradient flows of the Ambrosio-Tortorelli functional, where unilaterality comes from an irreversibility constraint on the fracture density. In the spirit of gradient flows in metric spaces, such evolutions are defined in terms of curves of maximal unilateral slope, and are constructed by means of implicit Euler schemes. An asymptotic analysis in the Mumford-Shah regime is also carried out. It shows the convergence towards a generalized heat equation outside a time increasing crack set.

Analysis of PDEs

A relation between Γ-convergence of functionals and their associated gradient flows

A Relation Between Г-convergence of Functionals and Their Associated Gradient Flows

On a Complex Hessian Flow

Γ-CONVERGENCE OF INTEGRAL FUNCTIONALS DEPENDING ON VECTOR-VALUED FUNCTIONS OVER PARABOLIC DOMAINS

Gamma-convergence of a gradient-flow structure to a non-gradient-flow structure

Gradient flows on graphons: existence, convergence, continuity equations

Local Conditions for Global Convergence of Gradient Flows and Proximal Point Sequences in Metric Spaces

Non-sufficiency of smoothness in the gradient conjecture

CONVERGENCE OF THE FAMILY OF EULER-HALLEY TYPE METHODS ON RIEMANNIAN MANIFOLDS UNDER THE Γ-Condition

Existence, uniqueness and L2t(H2x) ∩ L∞t(H1x) ∩ H1t(L2x) regularity of the gradient flow of the Ambrosio–Tortorelli functional

A qualitative difference between gradient flows of convex functions in finite- and infinite-dimensional Hilbert spaces

Convergence and non-convergence in a nonlocal gradient flow

Gradient descent for unbounded convex functions on Hadamard manifolds and its applications to scaling problems

Gamma-convergence of quadratic functionals perturbed by bounded linear functionals

Gradient flows in asymmetric metric spaces and applications

Convergence Analysis of Gradient Algorithms on Riemannian Manifolds Without Curvature Constraints and Application to Riemannian Mass

Gradient Flows and Riemannian Structure in the Gromov-Wasserstein Geometry

On the global convergence of Wasserstein gradient flow of the Coulomb discrepancy

Stability of the heat flow under convergence in concentration and consequences

Discrete-time gradient flows in Gromov hyperbolic spaces

Unilateral gradient flow of the Ambrosio-Tortorelli functional by minimizing movements