The thesis is about finding upper bounds on the "distance" between the distributions of two Markov chains, given that one, the original, is perturbed to the other. After providing some basic definitions e properties about Markov chains and their ergodicities, the work is composed by two parts. In the first part, I analyze the case of finite, irreducible and homogeneous Markov chains. The aim is to find condition numbers which lead to define upper bounds on the "distance" between the stationary distribution vectors of the two Markov chains. In the second part, I consider the case of a Markov chain with values in a Polish space. Here I make use of the Wasserstain distance between measures to find upper bounds on the distance between the probability measures of the two Markov chains.
Il lavoro di tesi consiste nella ricerca di maggioranti della "distanza" tra le distribuzioni di due catene di Markov, ipotizzando che una, l'originale, sia perturbata nell'altra. Dopo aver fornito alcune definizioni e proprietà basilari sulle catene di Markov e ergodicità, il lavoro è composto da due parti. Nella prima parte, analizzo il caso di catene di Markov finite, irriducibili e omogenee. L'obiettivo è quello di trovare numeri di condizionamento che portino a definire maggioranti sulla distanza tra vettori stazionari di due catene di Markov. Nella seconda parte, considero il caso di una catena di Markov a valori in uno spazio polacco. Qui faccio uso della distanza di Wasserstein tra misure per trovare maggioranti della distanza tra misure di probabilità delle due catene di Markov.
Perturbation theory for Markov chains
PALMIOTTO, MATTIA
2021/2022
Abstract
The thesis is about finding upper bounds on the "distance" between the distributions of two Markov chains, given that one, the original, is perturbed to the other. After providing some basic definitions e properties about Markov chains and their ergodicities, the work is composed by two parts. In the first part, I analyze the case of finite, irreducible and homogeneous Markov chains. The aim is to find condition numbers which lead to define upper bounds on the "distance" between the stationary distribution vectors of the two Markov chains. In the second part, I consider the case of a Markov chain with values in a Polish space. Here I make use of the Wasserstain distance between measures to find upper bounds on the distance between the probability measures of the two Markov chains.File | Dimensione | Formato | |
---|---|---|---|
Tesi Palmiotto Mattia.pdf
accesso riservato
Dimensione
1.01 MB
Formato
Adobe PDF
|
1.01 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/42088