1. General Information

Prereq: CSCE235, Discrete Math (exceptionally this year).

Course description: Constraint satisfaction has emerged as a powerful approach to articulate and solve many problems in computer science, engineering, and management.  It is now the basis for new programming languages and innovative commercial systems for production scheduling, product configuration, personnel planning and timetabling, etc.  The course will review the foundations of constraint satisfaction and the basic mechanisms for constraint propagation.  It will cover aspects of modeling and representation, and will examine islands of tractability and methods for theoretical and empirical evaluation of problem `difficulty.'  If time permits, we will examine new methods for decomposition and symmetry identification, designed to overcome the complexity barrier and to support interactions with users.

Lectures: Tuesday 5:00--6:15 pm and Friday 3:00 p.m.--4:15 p.m.
Location: Avery Hall, Room 352.

Instructor:   Prof.  Berthe Y. Choueiry
      Office location: Room 360, Avery Hall,
      choueiry AT cse.unl.edu, tel: (402)472-5444.
      Office hours: Tuesday/Thursday 2:00--3:00 p.m. or by appointment.

GTA:   Mr. Shant Karakashian
      email: shantk AT cse.unl.edu
      Office location: Room 123D, Avery Hall
    

Textbook:
"Constraint Processing" by Rina Dechter, 1^st edition. The textbook will not be followed sequestially, but should be used for reference.

For quick response, email cse421@cse.unl.edu.  Your message will be forwarded to both TA and instructor.

Important note: No official GTA was allocated to this class, but Shant has extensive experience in the subject matter and will try to help to the extent possible. All questions will have to be made to the instructor during office hours, make-up class, by email, or by appointment. We will follow a `peer-correction system' by which students grade each other's (glossaries and) homework with the help and close monitoring by the GTA and the instructor.

Topics include but are not restricted to:

• Properties, computational complexity, and practical importance.
• Global and local consistency: algorithms, properties and computational complexity.
• Islands of tractability for minimality and global consistency.
• Intelligent backtracking.
• Look-ahead techniques.
• Ordering heuristics.
• Theoretical and empirical comparison of hybrid search algorithms.
• Phase transition.
• Modeling and reduction methods between representations.
• If time permits: Decomposition. Symmetries and their approximations. Temporal constraint networks. Stochastic search. Reformulation and abstraction. Dynamic/conditional Constraint Satisfaction (CSP). Constrained Optimization Problem (COP).
• Other, depending on class interests.

Support:

• Rina Dechter, Constraint Processing, 2003, Morgan Kauffmann.  Available from bookstore.
• Edward Tsang, Foundations of Constraint Satisfaction. Available from instructor and on reserve at the Math Library in Avery.
• Technical papers given by instructor or available from the Electronic Reserve at the Love Library.

2. Protocol of the Course

The course consists of lectures by the instructor, three times per week.

Workload

Required and recommended reading:

Programming, theoretical, and library-search assignments:

• Programming assignments homework must be turned in using the web handin program on cse.unl.edu and turned in before class on the due date.
• Pen-and-paper assignments must be given to the instructor right before the lecture on the due date.
• Late homeworks are subject to a 20% deduction per day (including week-ends), any second after the due date counts as an entire day.
• Programming assignments can be done in any programming language but must run on cse.
• When code and data structures are provided for guidance, they will be mostly in Common Lisp (advice: use Allegro Common Lisp on Linux or a sun workstation).
• Students are kindly requested to indicate how much time they approximately spend on each exercise; this information will be aggregated and used for planning purposes and to balance your workload: it does not affect grading and the evaluation of individuals.

Surprise quizzes:

• There will be surprise quizzes throughout the semester (with a frequency inversely proportional to students' attendance).
• Quizzes will address all material covered during the lectures and/or by the required reading.
• No books or personal notes are allowed during the quizzes, unless explicitly specified.
• Quizzes cannot be made up.

Tests:

Project research, report, and presentation:

• The project can be either a literature study/survey or the implementation of algorithms.
• A presentation to the class is mandatory. The grade will take into account the quality of the research, report, and presentation.

Search Competition:

Attendance

• Because the course heavily relies on class discussion, attendance is mandatory.
• Attendance is taken at every lecture.
• Students who miss more than six sessions including make-up class/recitation may be asked to leave the course.
• Advanced notice to the instructor for each absence is required.
• Students are responsible for the material covered and announcements made during the class. Also, there will be surprise quizzes.
• Bonus points will be awarded to students who attend all lectures, interact lively, and participate in discussion in class.

Alerts

• It is the students responsibility to ensure an account on the department's computers.
• Discussions among students, instructor, and TA are encouraged. Homework however are a strictly individual activity: no sharing is permitted (unless when specified by instructor). Unethical behavior will be heavily sanctioned (e.g., a null grade on the task).
• Always acknowledge any help received from other individuals.
• Always fully reference material used (e.g., encyclopedia, book, paper, journal, web site).

3. Grading Policy

Grade Distribution

• (Surprise) quizzes: 37%
• Assignments: 28%
• Individual project
• Search competition: Solving a real-world optimization problem (in Lisp): 35% (Code and report: 25%, Presentation: 10%)
• Project: 35%.  (Research and report: 25%. Presentation: 10%.)

Grade Conversion

Reminder from the College of Arts & Sciences: a C (2.0) is the minimum passing grade in a PASS/NO PASS course (NOT a C-) as well as the lowest grade one can receive and still count the class toward a major.

4. How to Secure a Good Final Grade

Make a Presentation

Critical Summary

Write up

Attendance

Glossaries

• Students will be have to build an incremental and alphabetically sorted glossary of important terms.
• Terms to be included are the ones listed in the handouts distributed in class or sent my email.
• A glossary entry can be filled with: (1) its definition in Dechter's textbook, (2) its definition from another AI textbook, dictionary, or web resource, or (3) the student's own interpretation.
• All terms encountered during a week are due as a weekly glossary the following Monday, or as indicated in the schedule of the class.
• At the end of the course, the full alphabetically sorted glossary is due. The deadline for submitting final glossary is Friday, December 4, 2009 (Hint: choose a text editor that can sort entries alphabetically.)

Additional Work

5. Books on Reserve at the Math Library in Avery

Constraint Processing

AI

Lisp

Logic

6. (AI) References

7. Online Resources

Puzzles Built @ the ConSystLab

Benchmark Problems

Archives and On-Line Systems

Main publication venues

Conferences:

Journals:

Constraint Processing Online

• CP Online
• Yahoo Group: Constraints, Association for Constraint Programming
• AI Topics of the American Association for Artificial Intelligence
  Web search engines (Google (scholar), CiteSeer, Altavista, etc.)

Research Centers

Courses

• Rina Dechter: Constraint Networks ( 1999, 2001, 2003, 2007)
• Michela Milano: Constraint Programming Approach to AI Applications
• Roman Barták:
  • Foundations of Constraint Satisfaction, ESSLI 2002
  • Constraint Programming: An Online Textbook.
• Neil Yorke-Smith and Dan Bryce: CS227: Reasoning Methods in Artificial Intelligence.
• Fahiem Bacchus: CSC2512F Constraint Satisfaction Problems
• Barbara Smith: Tutorial on Constraint Programming.
• David McAllester: Lecture notes.
• Bacchus' Constraint Programming Bibliography

Academic Research Groups

• Constraint Systems Laboratory at UNL
• LIA@EPFL
• APES Group in the UK.
• Cork Constraint Computation Center (4C) of Freuder in University College Cork, Ireland.

Latex Resources

• From Shant Karakashian: LaTex on Wikibooks
• From Joel Gompert: Getting memory-efficient, pretty-looking, robust graphics with very fancy robust labels into LaTeX
• From Ryan Lim: The not so short introduction to LaTex2x