Back to Search
Start Over
On the Non-ergodic Convergence Rate of the Directed Nonsmooth Composite Optimization
- Source :
- Parallel and Distributed Computing, Applications and Technologies ISBN: 9783030692438, PDCAT
- Publication Year :
- 2021
- Publisher :
- Springer International Publishing, 2021.
-
Abstract
- This paper considers the distributed “nonsmooth+nonsmooth” composite optimization problems for which n agents collaboratively minimize the sum of their local objective functions over the directed networks. In particular, we focus on the scenarios where the sought solutions are desired to possess some structural properties, e.g., sparsity. However, to ensure the convergence, most existing methods produce an ergodic solution via the averaging schemes as the output, which causes the desired structural properties of the output to be destroyed. To address this issue, we develop a new decentralized stochastic proximal gradient method, termed DSPG, in which the nonergodic (last) iteration acts as the output. We also show that the DSPG method achieves the nonergodic convergence rate \(O(\log (T)/\sqrt{T})\) for generally convex objective functions and \(O(\log (T)/T)\) for strongly convex objective functions. When the structure-enhancing regularization is absent and the simple and suffix averaging schemes are used, the convergence rates of DSPG reach \(O(1/\sqrt{T})\) for generally convex objective functions and O(1/T) for strongly convex objective functions, showing improvement relative to the rates \(O(\log (T)/\sqrt{T})\) and \(O(\log (T)/T)\) provided by the existing methods. Simulation examples further illustrate the effectiveness of the proposed method.
- Subjects :
- Discrete mathematics
0209 industrial biotechnology
Computer science
Regular polygon
020206 networking & telecommunications
02 engineering and technology
Directed graph
Regularization (mathematics)
020901 industrial engineering & automation
Rate of convergence
Convergence (routing)
0202 electrical engineering, electronic engineering, information engineering
Ergodic theory
Proximal Gradient Methods
Convex function
Subjects
Details
- ISBN :
- 978-3-030-69243-8
- ISBNs :
- 9783030692438
- Database :
- OpenAIRE
- Journal :
- Parallel and Distributed Computing, Applications and Technologies ISBN: 9783030692438, PDCAT
- Accession number :
- edsair.doi...........0fb249e3f0372b9637b4632b7433617d