Nonlinear spectral decompositions by gradient flows of one-homogeneous functionals

@article{Bungert2019NonlinearSD,
  title={Nonlinear spectral decompositions by gradient flows of one-homogeneous functionals},
  author={Leon Bungert and Martin Burger and Antonin Chambolle and Matteo Novaga},
  journal={Analysis \& PDE},
  year={2019}
}
This paper establishes a theory of nonlinear spectral decompositions by considering the eigenvalue problem related to an absolutely one-homogeneous functional in an infinite-dimensional Hilbert space. This approach is both motivated by works for the total variation, where interesting results on the eigenvalue problem and the relation to the total variation flow have been proven previously, and by recent results on finite-dimensional polyhedral semi-norms, where gradient flows can yield spectral… Expand

Figures from this paper

Asymptotic profiles of nonlinear homogeneous evolution equations of gradient flow type
This work is concerned with the gradient flow of absolutely p -homogeneous convex functionals on a Hilbert space, which we show to exhibit finite ( $$p<2$$ p < 2 ) or infinite extinction time ( $$pExpand
Eigenvalue Problems in $\mathrm{L}^\infty$: Optimality Conditions, Duality, and Relations with Optimal Transport
In this article we characterize the L∞ eigenvalue problem associated to the Rayleigh quotient ‖∇u‖L∞/‖u‖∞ , defined on Lipschitz functions with homogeneous boundary conditions on a domain Ω. ForExpand
Introducing the p-Laplacian spectra
TLDR
A nonlinear decomposition, associated with nonlinear eigenfunctions of the p-Laplacian for p ∈ (1, 2) is developed, which can process signals of different degrees of smoothness and is shown to be extinct in finite time for γ ∈ R. Expand
Gradient Flows, Nonlinear Power Methods, and Computation of Nonlinear Eigenfunctions
TLDR
It is proved that Γ-convergence of functionals implies convergence of their ground states, and it is shown that the implicit Euler discretization of gradient flows gives rise to a nonlinear power method of the proximal operator and their convergence to nonlinear eigenfunctions. Expand
Iterative Methods for Computing Eigenvectors of Nonlinear Operators
TLDR
This chapter examines several iterative methods for solving nonlinear eigenvalue problems in variational image-processing, graph partition and classification, nonlinear physics and more, both classical algorithms and ones based on deep networks. Expand
Computing Nonlinear Eigenfunctions via Gradient Flow Extinction
TLDR
This work analyzes a scheme that recursively subtracts nonlinear eigenfunctions from given data and shows that this procedure yields a decomposition of the data into eigenFunctions in some cases as the 1-dimensional total variation, for instance. Expand
The Infinity Laplacian eigenvalue problem: reformulation and a numerical scheme
TLDR
This work presents an alternative formulation of the higher eigenvalue problem associated to the infinity Laplacian, which opens the door for numerical approximation of eigenfunctions and defines consistent numerical schemes for approximating infinity ground states and higher eigens. Expand
Numeric Solutions of Eigenvalue Problems for Generic Nonlinear Operators
TLDR
This paper proposes a generalized numerical method for solving eigenproblems for generic, nonlinear opera-tors, and solves the problem by a nonlinear adaptation of the power method, a well known linear eigensolver. Expand
Examining the Limitations of Dynamic Mode Decomposition through Koopman Theory Analysis
This work binds the existence of Koopman Eigenfunction (KEF), the geometric of the dynamics, and the validity of Dynamic Mode Decomposition (DMD) to one coherent theory. Viewing the dynamic as aExpand
Total-Variation Mode Decomposition
TLDR
This work formulate a relation between Dynamic Mode Decomposition (DMD), a dimensionality reduction method based on the Koopman operator, and the spectral TV decomposition, and a closed form solution of the rescaled DMD. Expand
...
1
2
...

References

SHOWING 1-10 OF 48 REFERENCES
Inverse Scale Space Decomposition
We investigate the inverse scale space flow as a decomposition method for decomposing data into generalised singular vectors. We show that the inverse scale space flow, based on convex and absolutelyExpand
Solution paths of variational regularization methods for inverse problems
We consider a family of variational regularization functionals for a generic inverse problem, where the data fidelity and regularization term are given by powers of a Hilbert norm and an absolutelyExpand
Spectral Decompositions Using One-Homogeneous Functionals
TLDR
Results on the orthogonality of the decomposition, a Parseval-type identity and the notion of generalized (nonlinear) eigenvectors closely link the nonlinear multiscale decompositions to the well-known linear filtering theory. Expand
Kurdyka–Łojasiewicz–Simon inequality for gradient flows in metric spaces
This paper is dedicated to providing new tools and methods for studying the trend to equilibrium of gradient flows in metric spaces in the entropy and metric sense, to establish decay rates, finiteExpand
Ground States and Singular Vectors of Convex Variational Regularization Methods
Singular value decomposition is the key tool in the analysis and understanding of linear regularization methods. In the last decade nonlinear variational approaches such as $\ell^1$ or totalExpand
Convergence rates of convex variational regularization
The aim of this paper is to provide quantitative estimates for the minimizers of non-quadratic regularization problems in terms of the regularization parameter, respectively the noise level. As usualExpand
Some qualitative properties for the total variation flow
We prove the existence of a finite extinction time for the solutions of the Dirichlet problem for the total variation flow. For the Neumann problem, we prove that the solutions reach the average ofExpand
Gauss-Green Theorem for Weakly Differentiable Vector Fields, Sets of Finite Perimeter, and Balance Laws
We analyze a class of weakly differentiable vector fields F W R N ! R N with the property that F 2 L 1 and div F is a (signed) Radon measure. These fields are called bounded divergence-measureExpand
Function Spaces and Potential Theory
The subject of this book is the interplay between function space theory and potential theory. A crucial step in classical potential theory is the identification of the potential energy of a chargeExpand
Convex Analysis and Monotone Operator Theory in Hilbert Spaces
This book provides a largely self-contained account of the main results of convex analysis and optimization in Hilbert space. A concise exposition of related constructive fixed point theory isExpand
...
1
2
3
4
5
...