Thesis (Ph.D.)(Industrial and Systems Eng.)--University of Pretoria, 2004. / Includes bibliographical references.

Thesis (M. Phil.)--University of Hong Kong, 1983.

published_or_final_version / Mathematics / Master / Master of Philosophy

No description available.

No description available.

Thesis (M.Sc.)--York University, 2006. Graduate Programme in Computer Science. / Typescript. Includes bibliographical references (leaves 117-130). Also available on the Internet. MODE OF ACCESS via web browser by entering the following URL: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&res_dat=xri:pqdiss&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&rft_dat=xri:pqdiss:MR29613

Markov decision processes (MDPs) are finite-state probabilistic systems with both strategic and random choices, hence well-established to model the interactions between a controller and its randomly responding environment. An MDP can be mathematically viewed as a one and half player stochastic game played in rounds when the controller chooses an action, and the environment chooses a successor according to a fixed probability distribution.<p><p>There are two incomparable views on the behavior of an MDP, when the strategic choices are fixed. In the traditional view, an MDP is a generator of sequence of states, called the state-outcome; the winning condition of the player is thus expressed as a set of desired sequences of states that are visited during the game, e.g. Borel condition such as reachability. The computational complexity of related decision problems and memory requirement of winning strategies for the state-outcome conditions are well-studied.<p><p>Recently, MDPs have been viewed as generators of sequences of probability distributions over states, called the distribution-outcome. We introduce synchronizing conditions defined on distribution-outcomes, which intuitively requires that the probability mass accumulates in some (group of) state(s), possibly in limit. A probability distribution is p-synchronizing if the probability mass is at least p in some state, and a sequence of probability distributions is always, eventually, weakly, or strongly p-synchronizing if respectively all, some, infinitely many, or all but finitely many distributions in the sequence are p-synchronizing.<p><p>For each synchronizing mode, an MDP can be (i) sure winning if there is a strategy that produces a 1-synchronizing sequence; (ii) almost-sure winning if there is a strategy that produces a sequence that is, for all epsilon > 0, a (1-epsilon)-synchronizing sequence; (iii) limit-sure winning if for all epsilon > 0, there is a strategy that produces a (1-epsilon)-synchronizing sequence.<p><p>We consider the problem of deciding whether an MDP is winning, for each synchronizing and winning mode: we establish matching upper and lower complexity bounds of the problems, as well as the memory requirement for optimal winning strategies.<p><p>As a further contribution, we study synchronization in probabilistic automata (PAs), that are kind of MDPs where controllers are restricted to use only word-strategies; i.e. no ability to observe the history of the system execution, but the number of choices made so far. The synchronizing languages of a PA is then the set of all synchronizing word-strategies: we establish the computational complexity of the emptiness and universality problems for all synchronizing languages in all winning modes.<p><p>We carry over results for synchronizing problems from MDPs and PAs to two-player turn-based games and non-deterministic finite state automata. Along with the main results, we establish new complexity results for alternating finite automata over a one-letter alphabet. In addition, we study different variants of synchronization for timed and weighted automata, as two instances of infinite-state systems. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

The classic theorems of Büchi and Kleene state the expressive equivalence of finite automata to monadic second order logic and regular expressions, respectively. These fundamental results enjoy applications in nearly every field of theoretical computer science. Around the same time as Büchi and Kleene, Rabin investigated probabilistic finite automata. This equally well established model has applications ranging from natural language processing to probabilistic model checking. Here, we give probabilistic extensions Büchi\\\'s theorem and Kleene\\\'s theorem to the probabilistic setting. We obtain a probabilistic MSO logic by adding an expected second order quantifier. In the scope of this quantifier, membership is determined by a Bernoulli process. This approach turns out to be universal and is applicable for finite and infinite words as well as for finite trees. In order to prove the expressive equivalence of this probabilistic MSO logic to probabilistic automata, we show a Nivat-theorem, which decomposes a recognisable function into a regular language, homomorphisms, and a probability measure. For regular expressions, we build upon existing work to obtain probabilistic regular expressions on finite and infinite words. We show the expressive equivalence between these expressions and probabilistic Muller-automata. To handle Muller-acceptance conditions, we give a new construction from probabilistic regular expressions to Muller-automata. Concerning finite trees, we define probabilistic regular tree expressions using a new iteration operator, called infinity-iteration. Again, we show that these expressions are expressively equivalent to probabilistic tree automata. On a second track of our research we investigate Constraint LTL over multidimensional data words with data values from the infinite tree. Such LTL formulas are evaluated over infinite words, where every position possesses several data values from the infinite tree. Within Constraint LTL on can compare these values from different positions. We show that the model checking problem for this logic is PSPACE-complete via investigating the emptiness problem of Constraint Büchi automata.

Les Markov Decision Process (MDP) sont des systèmes finis probabilistes avec à la fois des choix aléatoires et des stratégies, et sont ainsi reconnus comme de puissants outils pour modéliser les interactions entre un contrôleur et les réponses aléatoires de l'environment. Mathématiquement, un MDP peut être vu comme un jeu stochastique à un joueur et demi où le contrôleur choisit à chaque tour une action et l'environment répond en choisissant un successeur selon une distribution de probabilités fixée.Il existe deux incomparables représentations du comportement d'un MDP une fois les choix de la stratégie fixés.Dans la représentation classique, un MDP est un générateur de séquences d'états, appelées state-outcome; les conditions gagnantes du joueur sont ainsi exprimées comme des ensembles de séquences désirables d'états qui sont visités pendant le jeu, e.g. les conditions de Borel telles que l'accessibilité. La complexité des problèmes de décision ainsi que la capacité mémoire requise des stratégies gagnantes pour les conditions dites state-outcome ont été déjà fortement étudiées.Depuis peu, les MDPs sont également considérés comme des générateurs de séquences de distributions de probabilités sur les états, appelées distribution-outcome. Nous introduisons des conditions de synchronisation sur les distributions-outcome, qui intuitivement demandent à ce que la masse de probabilité s'accumule dans un (ensemble d') état, potentiellement de façon asymptotique.Une distribution de probabilités est p-synchrone si la masse de probabilité est d'au moins p dans un état; et la séquence de distributions de probabilités est toujours, éventuellement, faiblement, ou fortement p-synchrone si, respectivement toutes, certaines, infiniment plusieurs ou toutes sauf un nombre fini de distributions dans la séquence sont p-synchrones.Pour chaque type de synchronisation, un MDP peut être(i) assurément gagnant si il existe une stratégie qui génère une séquence 1-synchrone;(ii) presque-assurément gagnant si il existe une stratégie qui génère une séquence (1-epsilon)-synchrone et cela pour tout epsilon strictement positif;(iii) asymptotiquement gagnant si pour tout epsilon strictement positif, il existe une stratégie produisant une séquence (1-epsilon)-synchrone.Nous considérons le problème consistant à décider si un MDP est gagnant, pour chaque type de synchronisation et chaque mode gagnant: nous établissons les limites supérieures et inférieures de la complexité de ces problèmes ainsi que la capacité mémoire requise pour une stratégie gagnante optimale.En outre, nous étudions les problèmes de synchronisation pour les automates probabilistes (PAs) qui sont en fait des instances de MDP où les contrôleurs sont restreint à utiliser uniquement des stratégies-mots; c'est à dire qu'ils n'ont pas la possibilité d'observer l'historique de l'exécution du système et ne peuvent connaitre que le nombre de choix effectués jusque là. Les langages synchrones d'un PA sont donc l'ensemble des stratégies-mots synchrones: nous établissons la complexité des problèmes des langages synchrones vides et universels pour chaque mode gagnant.Nous répercutons nos résultats obtenus pour les problèmes de synchronisation sur les MDPs et PAs aux jeux tour à tour à deux joueurs ainsi qu'aux automates finis non-déterministes. En plus de nos résultats principaux, nous établissons de nouveaux résultats de complexité sur les automates finis alternants avec des alphabets à une lettre. Enfin, nous étudions plusieurs variations de synchronisation sur deux instances de systèmes infinis que sont les automates temporisés et pondérés. / Markov decision processes (MDPs) are finite-state probabilistic systems with bothstrategic and random choices, hence well-established to model the interactions between a controller and its randomly responding environment.An MDP can be mathematically viewed as a one and half player stochastic game played in rounds when the controller chooses an action,and the environment chooses a successor according to a fixedprobability distribution.There are two incomparable views on the behavior of an MDP, when thestrategic choices are fixed. In the traditional view, an MDP is a generator of sequence of states, called the state-outcome; the winning condition of the player is thus expressed as a set of desired sequences of states that are visited during the game, e.g. Borel condition such as reachability.The computational complexity of related decision problems and memory requirement of winning strategies for the state-outcome conditions are well-studied.Recently, MDPs have been viewed as generators of sequences of probability distributions over states, calledthe distribution-outcome. We introduce synchronizing conditions defined on distribution-outcomes,which intuitively requires that the probability mass accumulates insome (group of) state(s), possibly in limit.A probability distribution is p-synchronizing if the probabilitymass is at least p in some state, anda sequence of probability distributions is always, eventually,weakly, or strongly p-synchronizing if respectively all, some, infinitely many, or all but finitely many distributions in the sequence are p-synchronizing.For each synchronizing mode, an MDP can be (i) sure winning if there is a strategy that produces a 1-synchronizing sequence; (ii) almost-sure winning if there is a strategy that produces a sequence that is, for all epsilon > 0, a (1-epsilon)-synchronizing sequence; (iii) limit-sure winning if for all epsilon > 0, there is a strategy that produces a (1-epsilon)-synchronizing sequence.We consider the problem of deciding whether an MDP is winning, for each synchronizing and winning mode: we establish matching upper and lower complexity bounds of the problems, as well as the memory requirementfor optimal winning strategies.As a further contribution, we study synchronization in probabilistic automata (PAs), that are kind of MDPs where controllers are restricted to use only word-strategies; i.e. no ability to observe the history of the system execution, but the number of choices made so far.The synchronizing languages of a PA is then the set of all synchronizing word-strategies: we establish the computational complexity of theemptiness and universality problems for all synchronizing languages in all winning modes.We carry over results for synchronizing problems from MDPs and PAs to two-player turn-based games and non-deterministic finite state automata. Along with the main results, we establish new complexity results foralternating finite automata over a one-letter alphabet.In addition, we study different variants of synchronization for timed andweighted automata, as two instances of infinite-state systems.

The classic theorems of Büchi and Kleene state the expressive equivalence of finite automata to monadic second order logic and regular expressions, respectively. These fundamental results enjoy applications in nearly every field of theoretical computer science. Around the same time as Büchi and Kleene, Rabin investigated probabilistic finite automata. This equally well established model has applications ranging from natural language processing to probabilistic model checking. Here, we give probabilistic extensions Büchi\\\''s theorem and Kleene\\\''s theorem to the probabilistic setting. We obtain a probabilistic MSO logic by adding an expected second order quantifier. In the scope of this quantifier, membership is determined by a Bernoulli process. This approach turns out to be universal and is applicable for finite and infinite words as well as for finite trees. In order to prove the expressive equivalence of this probabilistic MSO logic to probabilistic automata, we show a Nivat-theorem, which decomposes a recognisable function into a regular language, homomorphisms, and a probability measure. For regular expressions, we build upon existing work to obtain probabilistic regular expressions on finite and infinite words. We show the expressive equivalence between these expressions and probabilistic Muller-automata. To handle Muller-acceptance conditions, we give a new construction from probabilistic regular expressions to Muller-automata. Concerning finite trees, we define probabilistic regular tree expressions using a new iteration operator, called infinity-iteration. Again, we show that these expressions are expressively equivalent to probabilistic tree automata. On a second track of our research we investigate Constraint LTL over multidimensional data words with data values from the infinite tree. Such LTL formulas are evaluated over infinite words, where every position possesses several data values from the infinite tree. Within Constraint LTL on can compare these values from different positions. We show that the model checking problem for this logic is PSPACE-complete via investigating the emptiness problem of Constraint Büchi automata.