Exact Markov Chain-based Runtime Analysis of a Discrete Particle Swarm Optimization Algorithm on Sorting and OneMax

02/05/2019
by   Moritz Mühlenthaler, et al.
0

We present a comprehensive analysis of a discrete particle swarm optimization (PSO) algorithm that can be adapted to work on a large class of discrete optimization problems. For this purpose, we model the algorithm's behavior by Markov chains with different choices of transition probabilities and prove, in most cases, matching lower and upper bounds on the expected return times. These new results on Markov chains enable us to derive lower and upper bounds on the expected number of function evaluations performed by the PSO algorithm with a single particle solving the sorting problem and OneMax. Additionally, we are able to prove for the first time upper bounds on the runtime of PSO with multiple particles. Our new techniques in the analysis of Markov chains include a useful indistinguishability property and a novel calculation of bounds by integration. Implicitly, we extend in this work the fitness level method to a non-elitist setting. Additionally, we present a guide that shows how our technical tools may be used in other settings.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/23/2021

Generalization Error Bounds on Deep Learning with Markov Datasets

In this paper, we derive upper bounds on generalization errors for deep ...
research
08/09/2014

Sensitivity analysis for finite Markov chains in discrete time

When the initial and transition probabilities of a finite Markov chain i...
research
07/17/2017

A Discrete Bouncy Particle Sampler

Markov Chain Monte Carlo (MCMC) algorithms are statistical methods desig...
research
06/18/2016

Hitting times of local and global optima in genetic algorithms with very high selection pressure

The paper is devoted to upper bounds on the expected first hitting times...
research
09/28/2022

From Understanding the Population Dynamics of the NSGA-II to the First Proven Lower Bounds

Due to the more complicated population dynamics of the NSGA-II, none of ...
research
11/19/2009

Evolutionary estimation of a Coupled Markov Chain credit risk model

There exists a range of different models for estimating and simulating c...

Please sign up or login with your details

Forgot password? Click here to reset