Fast convergence of generalized forward-backward algorithms for structured monotone inclusions
arXiv:2107.10107
Abstract
In this paper, we develop rapidly convergent forward-backward algorithms for computing zeroes of the sum of finitely many maximally monotone operators. A modification of the classical forward-backward method for two general operators is first considered, by incorporating an inertial term (closed to the acceleration techniques introduced by Nesterov), a constant relaxation factor and a correction term. In a Hilbert space setting, we prove the weak convergence to equilibria of the iterates , with worst-case rates of in terms of both the discrete velocity and the fixed point residual, instead of the classical rates of established so far for related algorithms. Our procedure is then adapted to more general monotone inclusions and a fast primal-dual algorithm is proposed for solving convex-concave saddle point problems.
arXiv comment25 pages
References in corpus (5)
- An inertial forward-backward algorithm for monotone inclusions
- Generalized Forward-Backward Splitting
- The rate of convergence of Nesterov's accelerated forward-backward method is actually faster than
- A Relaxed Inertial Forward-Backward-Forward Algorithm for Solving Monotone Inclusions with Application to GANs
- Accelerated Proximal Point Method for Maximally Monotone Operators