Stochastic modelling for statistical applications
Stochastic modelling for statistical applications
Academic year 2023/2024
- Course ID
- Matteo Ruggiero
- 1st year
- Teaching period
- Second semester
- D.M. 270 TAF B - Distinctive
- Course disciplinary sector (SSD)
- MAT/06 - probability and statistics
- Type of examination
- PROBABILITY THEORY (MAT0034)
- Propedeutic for
- BAYESIAN STATISTICS (MAT0070)
Sommario del corso
The course introduces to the theory of Markov chains, in discrete and continuous time, and Lévy processes, considered essential probabilistic instruments which should be part of a modern statistician's toolbox. As an illustrative application, some time will be devoted to introduce the basics of Markov chain Monte Carlo simulation methods, with a few examples of the most widely used algorithms.
A module of the course, included in the overall courseload, will be taught by visiting professor Samuel Livingstone (University College London, UK) on Monte Carlo methods for statistical inference (cf. International visiting professors).
Results of learning outcomes
Knowledge and understanding: after the course, the student will have a sufficient knowledge of discrete time markov chains, continous time Markov chains, of the basic Monte Carlo sampling schemes and Markov chain Monte Carlo algorithms, and of Levy processes. The student will know the main properties of these objects and how to prove these formally.
Applying knowledge and understanding: the student will be able to manipulate the mathematical objects seen in the course in order to elicit and prove their properties, for example verifying its reversibility; finding its stationary distribution, etc. The student will also be able to construct a Markov chain Monte Carlo algorithm for a problem at hand (of comparable difficulty to those analysied in class) by choosing and appropriately adapting one of the strategies seen in class. The acquired knowledge will suffice for reading and understanding independently a research article on topics coherent with the course contents.
Making judgements: the student will be able to classify a Markov chain and a continuous time Markov chain in terms of irreducibility, periodicity, recurrence, stationarity and reversibility.
Communication skills: the student will be able to explain to a non expert the acquired concepts by using a formallly correct and rigourous exposition, and to discuss with experts about topics coherent with the course contents.
- Markov chains: transition matrices, Chapman-Kolmogorov equations, strong Markov property, classification of states and chains, invariant measures and stationarity, reversibility, convergence to equilibrium and convergence of ergodic averages.
- Elements of Monte Carlo and Markov chain Monte Carlo methods: Monte Carlo sampling; rejection sampling; importance sampling; Markov chain Monte Carlo principle; thinning; Metropolis-Hastings algorithm; Gibbs sampler; slice sampler; mixtures and cycles of MCMC.
- Continuous time Markov chains: transition functions, transition rates and infinitesimal generators; backward and forward Kolmogorov equations; embedded chains and holding times; uniform chains; stationarity; reversibility; scaling limits and diffusion approximations.
- Levy processes: definition; infinite divisibility; Levy-Khintchine formula; Levy-Ito decomposition; Poisson random measures; processes with infinite activity.
The material introduced will be throughly discussed and illustrated with numerous examples.
The course is composed of 24 lectures for a total of 48 hours. Classes are delivered in presence and with live streaming through the link published on the course's Moodle page (see button below).
Learning assessment methods
The final assessment consists in a mandatory written examination followed by an optional oral examination. The verification will evaluate the student's detailed knowledge and comprehension of the topics covered in class, together with the propriety of language, rigour of exposition and level of mathematical formalization used.
The written examination yields a maximum grade of 29/30. Students who obtain less than 26 in the written test can only accept or decline the grade. Students who obtain at least 26 in the written test can accept or decline the grade as is, or sustain the oral examination. As a result of the oral examination, the grade of the written test can increase or possibly decrease.
Suggested readings and bibliography
- NORRIS, J.R. Markov chains. Cambridge Series in Statistical and Probabilistic Mathematics.
- BREMAUD, P. Markov Chains. Springer.
Further suggested readings:
- BILLINGSLEY, P. Probability and measure. Wiley.
- GRIMMETT, G.R. and STIRZAKER, D.R. Probability and random processes. Oxford University Press.
- KARLIN and TAYLOR. A first Course in Stochastic Processes. Academic Press.
- KARLIN and TAYLOR. A second Course in Stochastic Processes. Academic Press.
- Enrollment opening date
- 01/09/2021 at 00:00
- Enrollment closing date
- 30/06/2022 at 00:00