Les méthodes de Monte Carlo par chaînes de Markov (MCMC) utilisent généralement des chaînes de Markov réversibles. Jusqu’à récemment, une grande partie de la recherche théorique sur les chaînes de Markov concernait ce type de chaînes, notamment les théorèmes de Peskun (1973) et de Tierney (1998) qui permettent d’ordonner les variances asymptotiques de deux estimateurs issus de chaînes réversibles différentes. Dans ce mémoire nous analysons des algorithmes simulants des chaînes qui ne respectent pas cette condition. Nous parlons alors de chaînes non réversibles. Expérimentalement, ces chaînes produisent souvent des estimateurs avec une variance asymptotique plus faible et/ou une convergence plus rapide. Nous présentons deux algorithmes, soit l’algorithme de marche aléatoire guidée (GRW) par Gustafson (1998) et l’algorithme de discrete bouncy particle sampler (DBPS) par Sherlock et Thiery (2017). Pour ces deux algorithmes, nous comparons expérimentalement la variance asymptotique d’un estimateur avec la variance asymptotique en utilisant l’algorithme de Metropolis-Hastings. Récemment, un cadre théorique a été introduit par Andrieu et Livingstone (2019) pour ordonner les variances asymptotiques d’une certaine classe de chaînes non réversibles. Nous présentons leur analyse de GRW. De plus, nous montrons que le DBPS est inclus dans ce cadre théorique. Nous démontrons que la variance asymptotique d’un estimateur peut théoriquement diminuer en ajoutant des propositions à cet algorithme. Finalement, nous proposons deux modifications au DBPS. Tout au long du mémoire, nous serons intéressés par des chaînes issues de propositions déterministes. Nous montrons comment construire l’algorithme du delayed rejection avec des fonctions déterministes et son équivalent dans le cadre de Andrieu et Livingstone (2019). / Markov chain Monte Carlo (MCMC) methods commonly use chains that respect the detailed balance condition. These chains are called reversible. Most of the theory developed for MCMC evolves around those particular chains. Peskun (1973) and Tierney (1998) provided useful theorems on the ordering of the asymptotic variances for two estimators produced by two different reversible chains. In this thesis, we are interested in non-reversible chains, which are chains that don’t respect the detailed balance condition. We present algorithms that simulate non-reversible chains, mainly the Guided Random Walk (GRW) by Gustafson (1998) and the Discrete Bouncy Particle Sampler (DBPS) by Sherlock and Thiery (2017). For both algorithms, we compare the asymptotic variance of estimators with the ones produced by the Metropolis- Hastings algorithm. We present a recent theoretical framework introduced by Andrieu and Livingstone (2019) and their analysis of the GRW. We then show that the DBPS is part of this framework and present an analysis on the asymptotic variance of estimators. Their main theorem can provide an ordering of the asymptotic variances of two estimators resulting from nonreversible chains. We show that an estimator could have a lower asymptotic variance by adding propositions to the DBPS. We then present empirical results of a modified DBPS. Through the thesis we will mostly be interested in chains that are produced by deterministic proposals. We show a general construction of the delayed rejection algorithm using deterministic proposals and one possible equivalent for non-reversible chains.

The master's thesis aims to introduce and study the issue of frequency hopping with collsion avoidance (FH/CA). On this basis, design a computer program for simulating the operation of a radio systém FHCA, who works in the band used by other systems FH/CA . This simulation programm using MATLAB software to implement verify the correctness of programs. Use simulator to obtain date about the intensity of interference systems FH/CA for the chosen scenario. This work consists of five parts: the first part consists of describing the queuing system, the second part of the description of the radio frequency system with collision avoidence FH / CA, the third part of the description of the simulation model. The fourth part includes verification of the model in the fifth and last section inspects results are shown.

The diploma project is focused on investigating the reliability of multi-state technical systems. A summary of the basic conception of renewal theory and stochastic processes is given in this paper. The possibility of solving multi-state systems reliability by using Markov models or simulation is shown. The software Reliab. S. M. S. O. 1.0 was created for solving non-homogeneous series, parallel, series-parallel and parallel-series systems. Outputs of this software are described in chapter 10. This work also contains $\beta$-factor method and theory of common-cause failures. The diploma project was supported by project from MSMT of the Czech Republic no. 1M06047 "Centre for Quality and Reliability of Production" and by grant from Grant Agency of the Czech Republic (Czech Science Foundation) reg. no. 103/08/1658 "Advanced optimum design of composed concrete structures".

Les algorithmes MCMC sont difficiles à mettre à l'échelle, car ils doivent balayer l'ensemble des données à chaque itération, ce qui interdit leurs applications dans de grands paramètres de données. En gros, tous les algorithmes MCMC évolutifs peuvent être divisés en deux catégories: les méthodes de partage et de conquête et les méthodes de sous-échantillonnage. Le but de ce projet est de réduire le temps de calcul induit par des fonctions complexes ou à grande efficacité. / MCMC algorithms are difficult to scale, since they need to sweep over the whole data set at each iteration, which prohibits their applications in big data settings. Roughly speaking, all scalable MCMC algorithms can be divided into two categories: divide-and-conquer methods and subsampling methods. The aim of this project is to reduce the computing time induced by complex or largelikelihood functions.

We prove that every (nice) Feller process is an It^o process in the sense of Cinlar, Jacod, Protter and Sharpe (1980). Next we generalize the notion of the symbol and define it for this larger class of processes. As examples the solutions of stochastic differential equations are considered. The symbol is then used to derive a quick approach to the semimartingale characteristics as well as the generator of the process under consideration. Finally we give some examples of how our methods work for processes used in mathematical finance. / Wir haben gezeigt, dass jeder (nette) Feller Prozess ein It^o Prozess im Sinne von Cinlar, Jacod, Protter und Sharpe (1980) ist. Es stellt sich heraus, dass man den Begriff des Symbols, der für Feller Prozesse bekannt ist, auf diese größere Klasse verallgemeinern kann. Dieses Symbol haben wir für die Lösungen verschiedener stochastischer Differentialgleichungen berechnet. Außerdem haben wir gezeigt, dass das Symbol einen schnellen Zugang zur Berechnung der Semimartingal-Charakteristiken und des Erzeugers eines It^o Prozesses liefert. Zuletzt wurden die Ergebnisse auf Prozesse angewendet, die in der Finanzmathematik gebräuchlich sind. - (Die Dissertation ist veröffentlicht im Shaker Verlag GmbH, Postfach 101818, 52018 Aachen, Deutschland, http://www.shaker.de, ISBN: 978-3-8322-8244-8)

The interest in autonomous surveillance has increased due to advances in autonomous systems and sensor theory. This thesis is a preliminary study of the cooperation between UGVs and stationary sensors when monitoring a dedicated area. The primary focus is the path planning of a UGV for different initial intrusion alarms. Cell decomposition, i.e., spatial partitioning, of the area of surveillance was utilized, and the objective function is based on the probability of a present intruder in each cell. These probabilities were modeled through two different methods: ExpPlanner, utilizing an exponential decay function. Markov planner, utilizing a Markov chain to propagate the probabilities. The performance of both methods improves when a confident alarm system is utilized. By prioritizing the direction of the planned paths, the performances improved further. The Markov planner outperforms the ExpPlanner in finding a randomly walking intruder. The ExpPlanner is suitable for passive surveillance, and the Markov planner is suitable for ”aggressive target hunting”.

Um conjunto de modelos compostos de redes de filas em grades finitas servindo como ambientes aleatorios para um ou mais passeios aleatorios, que por sua vez podem afetar o comportamento das filas, e desenvolvido. Duas formas de interacao entre os passeios aleatorios sao consideradas. Para cada modelo, e provado que o processo Markoviano correspondente e recorrente positivo e reversivel. As equacoes de balanceamento detalhado sao analisadas para obter a forma funcional da medida invariante de cada modelo. Em todos os modelos analisados neste trabalho, a medida invariante em uma grade finita tem forma produto. Modelos de redes de filas como ambientes para multiplos passeios aleatorios sao estendidos a grades infinitas. Para cada modelo estendido, sao especificadas as condicoes para a existencia do processo estocastico na grade infinita. Alem disso, e provado que existe uma unica medida invariante na rede infinita cuja projecao em uma subgrade finita e dada pela medida correspondente de uma rede finita. Finalmente, e provado que essa medida invariante na rede infinita e reversivel. / A set of models composed of queueing networks serving as random environments for one or more random walks, which themselves can affect the behavior of the queues, is developed. Two forms of interaction between the random walkers are considered. For each model, it is proved that the corresponding Markov process is positive recurrent and reversible. The detailed balance equa- tions are analyzed to obtain the functional form of the invariant measure of each model. In all the models analyzed in the present work, the invariant measure on a finite lattice has product form. Models of queueing networks as environments for multiple random walks are extended to infinite lattices. For each model extended, the conditions for the existence of the stochastic process on the infinite lattice are specified. In addition, it is proved that there exists a unique invariant measure on the infinite network whose projection on a finite sublattice is given by the corresponding finite- network measure. Finally, it is proved that that invariant measure on the infinite lattice is reversible.

Um conjunto de modelos compostos de redes de filas em grades finitas servindo como ambientes aleatorios para um ou mais passeios aleatorios, que por sua vez podem afetar o comportamento das filas, e desenvolvido. Duas formas de interacao entre os passeios aleatorios sao consideradas. Para cada modelo, e provado que o processo Markoviano correspondente e recorrente positivo e reversivel. As equacoes de balanceamento detalhado sao analisadas para obter a forma funcional da medida invariante de cada modelo. Em todos os modelos analisados neste trabalho, a medida invariante em uma grade finita tem forma produto. Modelos de redes de filas como ambientes para multiplos passeios aleatorios sao estendidos a grades infinitas. Para cada modelo estendido, sao especificadas as condicoes para a existencia do processo estocastico na grade infinita. Alem disso, e provado que existe uma unica medida invariante na rede infinita cuja projecao em uma subgrade finita e dada pela medida correspondente de uma rede finita. Finalmente, e provado que essa medida invariante na rede infinita e reversivel. / A set of models composed of queueing networks serving as random environments for one or more random walks, which themselves can affect the behavior of the queues, is developed. Two forms of interaction between the random walkers are considered. For each model, it is proved that the corresponding Markov process is positive recurrent and reversible. The detailed balance equa- tions are analyzed to obtain the functional form of the invariant measure of each model. In all the models analyzed in the present work, the invariant measure on a finite lattice has product form. Models of queueing networks as environments for multiple random walks are extended to infinite lattices. For each model extended, the conditions for the existence of the stochastic process on the infinite lattice are specified. In addition, it is proved that there exists a unique invariant measure on the infinite network whose projection on a finite sublattice is given by the corresponding finite- network measure. Finally, it is proved that that invariant measure on the infinite lattice is reversible.

Health technology assessment: case study on breast carcinoma treatment in the Czech Republic Bc. Lenka Šlegerová January 4, 2019 Abstract This thesis proposes an original method for assessing total costs of med- ical treatment. It defines the semi-Markov model with four states that are associated with specific costs of the treatment, and not with patients' health statuses. This method is applied to individuals' treatment data drawn from the Czech clinical practice in the treatment of the metastatic HER2+ breast cancer. The aim is to assess the cost-effectiveness of adding medication per- tuzumab to the combination of trastuzumab+docetaxel within first-line therapy and to examine whether using individual data on Czech patients and the economic conditions leads to different results from foreign stud- ies. Furthermore, employing censored data from the clinical practice in the thesis complicates the estimation of patients' overall survival in compari- son to clinical-trials data that form random samples. Therefore, survival functions were not only estimated by the Kaplan-Meier estimator but also using the Cox proportional hazard model and the Accelerated failure time model that both control for the effects of included covariates. The addition of pertuzumab does not result in significantly longer pa- tients'...

本研究以確定給付型退休基金為對象，建構廣義隨機評價模型，以衡量不確定情況下退休基金之財務風險。希望藉著模型建構的過程，適切地描述基金評價過程中所應考量的各項要素。 為了強調基金評價時同時考量內外部精算假設的重要性，本研究將模型分為存活函數、經濟函數和給付函數三部份討論；存活函數利用離散時間非同質性半馬可夫過程（Discrete Time Non-Homogeneous semi-Markov Process）描述成員狀態轉移的機率，把成員工作年資、年齡和及狀態納入評價過程，有別於傳統僅以年齡為假設基礎之精算方法；經濟函數則以隨機過程表達外部環境的不確定性，結合上述假設資訊預估未來給付後，成為半馬可夫隨機精算評價模型，此一般性的模型能推展至基金評價時所需的各項流程。因此，本研究將模型應用於我國公務人員退撫基金，針對公務人員退撫基金的給付特性發展財務評價公式，完整地描述基金精算成本計算、未來人力與現金流量結構模擬以及敏感度分析等過程。 最後，本研究撰寫公務人員精算評價資訊系統，具體化半馬可夫隨機精算評價模型，實證公務人員退撫基金財務評價公式。實證結果也顯示，不論基金的性質或外部經濟環境，都將影響退休基金財務評價結果，為基金評價時不可忽略的精算假設。 / This study focuses on constructing a generalized valuation model for the defined benefit pension schemes. Financial soundness and funding stability are critical issues in pension fund management. In this study, a realistic stochastic model is built to monitor the uncertainty factors in affecting the financial risk and cash flow dynamics along the decision process. In order to evaluate the importance of the interior and exterior actuarial assumptions in pension valuation. Detailed models in describing the turnover patterns, economic uncertainties and benefit structures are explored. Semi-Markov process proposed by Dominicis, Manca and Granata (1991) and Janssen and Manca (1997) is extended in structuring the transition pattern of the plan’s population and the economic based factors are generated through stochastic processes. Modifications according to classification and movements of the plan member and the plan’s turnover pattern are employed to improve its practical usefulness. Then the actuarial valuations, cash flow analyses and workforce projection are performed and investigated. We has explicitly formulated the plan’s realistic phenomenon and implemented the proposed mechanism into a risk management framework for pension finance. By using this realistic approach, the cost factors could be monitored throughout the valuation. Typically these analyses involve substantial assumptions. This article has outlined the procedure of building the proposed model. Finally, Taiwan Public Employees Retirement System is simplified to illustrate the proposed methodology in pension valuation. The results from this study show that the structure of the pension schemes and the assumed economic factors are the significant factors in pension valuation. It also indicated that the fund manager can evaluate these impacts through the proposed model.