Corpus ID: 209377022

A Control-Theoretic Perspective on Optimal High-Order Optimization

  title={A Control-Theoretic Perspective on Optimal High-Order Optimization},
  author={Tianyi Lin and Michael I. Jordan},
  • Tianyi Lin, Michael I. Jordan
  • Published 2019
  • Mathematics, Computer Science
  • ArXiv
  • In this paper, we provide a control-theoretic perspective on optimal tensor optimization algorithms for minimizing a convex function in a finite-dimensional Euclidean space. Given a function $\Phi: \mathbb{R}^d \rightarrow \mathbb{R}$ that is convex and twice-continuously differentiable, we study an ordinary differential equation (ODE) that is governed by the gradient operator $\nabla \Phi$ and a positive control parameter $\lambda(t)$ that tends to infinity as $t \rightarrow +\infty$. The… CONTINUE READING


    Publications referenced by this paper.
    Fast convex optimization via inertial dynamics with Hessian driven damping
    • 31
    • PDF
    Fast convergence of inertial dynamics and algorithms with asymptotic vanishing viscosity
    • 95
    • PDF
    Second Order Forward-Backward Dynamical Systems For Monotone Inclusion Problems
    • 37
    • Highly Influential
    • PDF
    Accelerated Regularized Newton Methods for Minimizing Composite Convex Functions
    • 18
    • PDF
    Tensor Methods for Minimizing Functions with H\"{o}lder Continuous Higher-Order Derivatives
    • 19
    • PDF
    Tensor Methods for Finding Approximate Stationary Points of Convex Functions
    • 7
    • PDF
    The Second-order in Time Continuous Newton Method
    • 13
    • Highly Influential
    The Differential Inclusion Modeling FISTA Algorithm and Optimality of Convergence Rate in the Case b $\leq3$
    • 14
    • PDF
    On damped second-order gradient systems
    • 29
    • Highly Influential
    • PDF