Linear convergence of forward-backward accelerated algorithms without knowledge of the modulus of strong convexity

DOI10.1137/23M158111XMaRDI QIDQ6561383zbMATH OpenFDO

Authors Bo-wen Li, Bin Shi, Yaxiang Yuan

Publication date 25 June 2024

Published in SIAM Journal on Optimization (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1137/23M158111X

zbMATH Keywords

\(\mu\)-strongly convex function\texttt{FISTA}\texttt{NAG}Lyapunov functionphase-space representation

Mathematics Subject Classification ID

Ridge regression; shrinkage estimators (Lasso) (62J07) Numerical aspects of computer graphics, image analysis, and computational geometry (65D18) Analysis of algorithms and problem complexity (68Q25) Convex programming (90C25) Nonlinear programming (90C30)

Cites work

• A differential equation for modeling Nesterov's accelerated gradient method: theory and insights
• A dynamical approach to an inertial forward-backward algorithm for convex minimization
• A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems
• A Lyapunov analysis of accelerated methods in optimization
• A second-order differential system with Hessian-driven damping; application to non-elastic shock laws
• A variational perspective on accelerated methods in optimization
• An introduction to continuous optimization for imaging
• Fast convex optimization via inertial dynamics with Hessian driven damping
• From differential equation solvers to accelerated first-order methods for convex optimization
• On the convergence of the iterates of the ``fast iterative shrinkage/thresholding algorithm
• The rate of convergence of Nesterov's accelerated forward-backward method is actually faster than 1/k^2
• Understanding the acceleration phenomenon via high-resolution differential equations

Cited in

(5)

• Acceleration via perturbations on low-resolution ordinary differential equations
• Symplectic discretization approach for developing new proximal point algorithm
• A Lyapunov analysis of accelerated PDHG algorithms
• Modern Theory of Gradient-Based Optimization
• Understanding the ADMM algorithm via high-resolution differential equations

This page was built for publication: Linear convergence of forward-backward accelerated algorithms without knowledge of the modulus of strong convexity

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6561383)