Markov chains and applications in the generation of combinatorial designs

This Thesis deals with discrete Markov chains and their applications in the generation of combinatorial designs. A conjecture on the generation of proper edge colorings of the complete graph K_n, for n even, is tackled. A proper edge coloring is an edge coloring such that no two adjacent edges have...

Descripción completa

Detalles Bibliográficos
Autor: Calatayud Gregori, Julia
Tipo de recurso: tesis de maestría
Fecha de publicación:2018
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/113466
Acceso en línea:https://hdl.handle.net/2117/113466
Access Level:acceso abierto
Palabra clave:Markov processes
Markov chain
Metropolis
K_n
Combinatorial design
Proper coloring
Potential
Entropy
Markov, Processos de
Classificació AMS::60 Probability theory and stochastic processes::60J Markov processes
Àrees temàtiques de la UPC::Matemàtiques i estadística::Estadística matemàtica
Descripción
Sumario:This Thesis deals with discrete Markov chains and their applications in the generation of combinatorial designs. A conjecture on the generation of proper edge colorings of the complete graph K_n, for n even, is tackled. A proper edge coloring is an edge coloring such that no two adjacent edges have the same color. Proper edge colorings are characterized by minimizing the potential and maximizing the entropy. We implement an algorithm in the software R to generate proper colorings from any arbitrary coloring of K_n, by identifying the colorings as nodes in a Markov chain, where transition probabilities are defined so that the potential decreases or, alternatively, the entropy increases. The conjecture states that the algorithm converges in polynomial time. We give original proofs of the conjecture in K_4 and K_6, and we provide new results and ideas that could be used to prove the conjecture in the general case K_n.