An accelerated first-order method with complexity analysis for solving cubic regularization subproblems

نویسندگان

چکیده

We propose a first-order method to solve the cubic regularization subproblem (CRS) based on novel reformulation. The reformulation is constrained convex optimization problem whose feasible region admits an easily computable projection. Our requires computing minimum eigenvalue of Hessian. To avoid expensive computation exact eigenvalue, we develop surrogate where replaced with approximate one. then apply methods such as Nesterov’s accelerated projected gradient (APG) and Barzilai-Borwein problem. As our main theoretical contribution, show that when $$\epsilon$$ -approximate computed by Lanczos approximately solved APG, approach returns solution CRS in $${\tilde{O}}(\epsilon ^{-1/2})$$ matrix-vector multiplications (where $${\tilde{O}}(\cdot )$$ hides logarithmic factors). Numerical experiments are comparable outperform Krylov subspace easy hard cases, respectively. further implement solvers adaptive methods, numerical results algorithms state-of-the-art algorithms.

برای دانلود باید عضویت طلایی داشته باشید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

An Accelerated First-Order Method for Solving Unconstrained SOS Polynomial Optimization Problems

Our interest lies in solving large-scale unconstrained SOS (sum of squares) polynomial optimization problems. Because interior-point methods for solving these problems are severely limited by the large-scale, we are motivated to explore efficient implementations of an accelerated first-order method to solve this class of problems. By exploiting special structural properties of this problem clas...

متن کامل

An Accelerated First-Order Method for Solving Unconstrained Polynomial Optimization Problems

Our interest lies in solving large-scale unconstrained polynomial optimization problems. Because interior-point methods for solving these problems are severely limited by the large-scale, we are motivated to explore efficient implementations of an accelerated first-order method to solve this class of problems. By exploiting special structural properties of this problem class, we greatly reduce ...

متن کامل

An accelerated first-order method for solving SOS relaxations of unconstrained polynomial optimization problems

An accelerated first-order method for solving SOS relaxations of unconstrained polynomial optimization problems Dimitris Bertsimas , Robert M. Freund & Xu Andy Sun To cite this article: Dimitris Bertsimas , Robert M. Freund & Xu Andy Sun (2013) An accelerated first-order method for solving SOS relaxations of unconstrained polynomial optimization problems, Optimization Methods and Software, 28:3...

متن کامل

An adaptive accelerated first-order method for convex optimization

In this paper, we present a new accelerated variant of Nesterov’s method for solving a class of convex optimization problems, in which certain acceleration parameters are adaptively (and aggressively) chosen so as to: preserve the theoretical iteration-complexity of the original method, and; substantially improve its practical performance in comparison to the other existing variants. Computatio...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Computational Optimization and Applications

سال: 2021

ISSN: ['0926-6003', '1573-2894']

DOI: https://doi.org/10.1007/s10589-021-00274-7