Στοχαστική Βελτιστοποίηση-Stochastic Optimization
12/16/2017

Περιεχόμενο Μαθήματος

Objectives of the course:
Stochastic models and algorithms are pertinent in many applications. The students are exposed to several of the stochastic aspects in several courses but, due to the diverse character of the focus of these courses, they are not provided with an essentially unifying framework. After taking this course, they have been exposed to the underlying techniques and methodologies and are able to cast and study their problems in a much deeper fashion. They are also given the information and tools to tackle these problems in new ways which were available in the literature not directly pertaining to their own application. What we emphasize in this course is the basic underlying dynamic stochastic dynamical system and the several techniques available for studying its limiting behavior.

List of topics to be covered and time spent on each:

Dynamic Programming 9 hours
Stochastic Stability 9 hours
Markovian Learning 9 hours
Stochastic Approximation 9 hours
Simulated Annealing 6 hours

Ημ/νία τελευταίας τροποποίησης: Wednesday, July 22, 2009
Course readings, papers, projects, out-of-class activities, etc:

For each part a bibliography is provided. For each part we cover in class, the following material is also distributed in class.

1. Dynamic Programing
Reference 13: Chapters 1,2,3,7

2. Stochastic Stability
References 3,4,5

3. Markovian Learning
Reference 4: Chapters 1,2,3

4. Stochastic Approximation
Reference 1: Chapters 1,2,3,6
References 1,2,4 (papers)

5. Simulated Annealing
References 1,4,5,9


Several problems from the material distributed in class are assigned as homework. The students also have to make a final presentation in class of some paper which addresses an application and uses the ideas and methodologies covered in class. The papers to be presented are chosen by me and the student, and are related to the student’s area of specialization.

Ημ/νία τελευταίας τροποποίησης: Wednesday, July 22, 2009