STANFORD EE 292E
Analysis and Control of Markov Chains
Spring 2008

Course Information

Class Times and Locations

Tuesdays and Thursdays 4:15-5:30PM in Education 128

Course Description

An introduction to dynamic optimization with an emphasis on problem formulation. Problems studied include Viterbi decoding, control of queues, revenue management, caching, optimal stopping, scheduling tasks, linear-quadratic control, multi-armed bandit problems, admission control, and network switching.

Prerequisites

Prerequisites: EE 263 and EE 278, or equivalent background in probability, linear algebra, and dynamic systems. Experience with Matlab.


Instructors


Grading

Course grades will be based 15% on homework assignments, 35% on a midterm exam, and 50% on a final exam.

Students are encouraged to work on homework problems in groups but must write up their own solutions. When writing up solutions, students should write the names of people with whom they discussed the assignment.


Course Readings

The primary textbook will be Dynamic Programming and Optimal Control, Vol. 1 (3rd edition), by Dimitri Bertsekas. The instructor will also provide some supplementary notes over the course of the quarter.


Announcements
Tentative Lectures


Date Topic Read Sections Notes
1 April Deterministic Systems 2.1 Note
3 April Viterbi Decoding, Caching 2.2 Note
8 April Caching 4.1 Note
10 April Linear-Quadratic Control, Stochastic Systems 4.2 Note
15 April Dynamic Pricing Note
17 April Inventory Management Note
22 April Portfolio Management 4.3 Note
24 April Optimal Stopping, Scheduling 4.4, 4.5 Note
29 April Imperfect State Information 5.1 Note
1 May Separation Principle 5.2 Note
6 May Belief States 5.4 Note
8 May Infinite Horizon Problems Note
13 May Algorithms for Infinite Horizon Problems Note
15 May Algorithms for Infinite Horizon Problems Note
20 May Algorithms for Infinite Horizon Problems Note 1  Note 2
22 May Queueing and Uniformization Note
27 May Multi-Armed Bandits Note 1  Note 2
29 May Approximate Dynamic Programming Note
3 June Review


Homework

Homework is due at 5PM on Thursdays. Please place it in a file cabinet labeled "EE292E" on the second floor of Packard. Late homework will not be accepted.

Homework 1 Solution  Matlab code
Homework 2 Solution
Homework 3 Solution  Matlab code
Homework 4 Solution
Homework 5 Solution
Homework 6 Solution  Matlab code 1  Matlab code 2
Homework 7 Solution
Homework 8 Solution  Matlab code


Exams

Students can choose one of two time slots to take the five-hour, open-book, open-notes take-home midterm: May 2 from 12PM to 5PM, and May 3 from 12PM to 5PM.
Midterm 2005-2006: Exam  Solution
Midterm 2007-2008: Exam  Solution

Students can choose one of two time slots to take the five-hour, open-book, open-notes take-home final: June 4 from 12PM to 5PM, and June 5 from 12PM to 5PM.
Final 2005-2006: Exam  Solution
Final 2007-2008: Exam  Solution