![[Lunds Universitet]](/icons/2radcs.gif){kind=icon}
Latest new, 2012
- 11/9. The lecture tomorrow will be in Lund in MH:227 from 10-12 and 13-16. Remember to sign up for the course no later than on 12/9.
- 28/8 The first lecture will take place on September 5th at the Technical University of Denmark/Danmarks tekniska Universitet (DTU). I will arrange for transportantion (cars/minibus(es)) based on the number of students that have signed up for the lecture at Tuesday noon. Please sign up for the first lecture using the doodle poll HERE
COURSE DESCRIPTION
Department: Division of Mathematical Statistics at the Centre for Mathematical Sciences, Lund Institute of Technology together with Informatics and Mathematical Modelling (IMM), Technical University of Denmark, Lyngby.
Credits: 7.5 ECTS credits.
Lecturers: Erik Lindström, phone: +46 46 222 45 78, email: erikl@maths.lth.se
Henrik Madsen, phone: +45 45 253408, email: hm@imm.dtu.dk
Prerequisites: Mathematical Statistics, basic course. Furthermore, it is recommendable to have taken a course on Stationary Processes, and necessary to have taken a basic course in Time Series Analysis, e.g. FMS051 Time Series Analysis / Tidsserieanalys in Lund or 02417 Time Series Analysis in Lyngby.
Course program: Can be found HERE. We start at wednesday Septmber 5th, and will have lectures and computer exercises on wednesdays. The lectures in Lund will be in MH:309a.
A short description of the contents of the course:
The graduate course in Advanced Time Series Analysis has its target audience amongst students with technical or natural science background and with adequate basic knowledge in mathematical statistics. The primary goal to give a thorough knowledge on modeling dynamic systems. A special attention is paid to non-linear and non-stationary systems, and the use of stochastic differential equations for modeling physical systems. In more detail:
- Non-linear time series models; Generalized transfer functions
- Kernel estimators and time series analysis
- Non-parametric models and modeling techniques
- Identification of non-linear models, cumulants and polyspectra
- Parameter estimation in non-linear models, Case study
- State space models and state filtering
- Stochastic differential equations (SDEs), Ito calculus, Exact and approximate filters
- Estimation of linear and (some) non-linear SDEs
- Modelling using SDEs
- Methods for tracking parameters in non-stationary time series.
- Experimental design for dynamic system identification.
- Prediction in non-linear models
Literature:
H. Madsen, J. Holst & E. Lindstrom (2010): Modelling Non-Linear and Non-Stationary Time Series
Exercises during the lectures:
In the afternoon during the lecture days, one hour will be used for you working with small exercises to be solved without computer. These exercises will be delivered during the lecture days.
Computer exercises: During the course the computer exercises will be
delivered in connection with the lectures. They are also downloadable from
the course homepage.
The computer exercises will in Lund be guided by
Erik Lindström, phone: +46 46 2224578.