18.405J | Spring 2016 | Graduate

Advanced Complexity Theory

Syllabus

Course Meeting Times

Lectures: 2 sessions / week, 1.5 hours / session

Prerequisites

The formal perquisite is 18.404J Theory of Computation which covers the early chapters of the required textbook book (see below).

However, students who have not taken the above specific course who still have a basic knowledge of complexity theory, either through the MIT undergraduate class 6.045J / 18.400J Automata, Computability, and Complexity, or through another basic course elsewhere, are likely to have enough background to register.

Description

This graduate-level course focuses on current research topics in computational complexity theory. Topics include:

  • Polynomial hierarchy and time-space lower bounds
  • Approximate counting and Toda’s Theorem
  • Relativization, Baker-Gill-Solovay
  • AC0 and switching lemma
  • Razborov-Smolensky
  • NEXP vs. ACC0
  • Communication complexity
  • Expanders and their applications
  • Derandomizing polynomial identity testing implies circuit lower bounds
  • PCP theorem intro: Constraint satisfaction vs games, hardness of approximation
  • Proof of PCP theorem: Hadamard code and an exponential PCP. Gap amplification.
  • Natural proofs

Textbook

The primary written reference for the course is

Arora, Sanjeev, and Boaz Barak. Computational Complexity: A Modern Approach. Cambridge University Press, 2009. ISBN: 9780521424264. [Preview with Google Books]

A draft of Computational Complexity: A Modern Approach is available online free of charge, but the latest edition contains corrections and improvements.

Problem Sets

There will be 4 (small) problem sets during the semester with two or three problems in each set. These will be due at the beginning of class roughly two weeks after their assignment. Please observe the following:

  • Write each problem in the problem set on a separate sheet (as each problem will be graded by separate graders) and write down your name on top of every sheet.
  • Specify the name of your collaborators (if done solely on our own write “Collaborators: None”) on the first page of the whole problem set.

Grading Assignment

Each registered student is expected to participate in grading one of the four problem sets. The graders will receive a model solution, and are expected to submit the graded solutions within a week.

Instructions

Each problem part should be marked as either: (3) Perfect solution; (2) Mostly correct solution but with minor mistakes; (1) A very flawed solution with some correct ideas; or (0) Missing / completely wrong solution.

Concrete mistakes should be marked and explained. The purpose is that the students get a fair and useful feedback on their solutions.

The grader will receive a score on their performance based the overall quality, precision, and fairness of their grading.

Final Project

Students will work individually or in pairs to complete a written final project. The final projects must be in the form of an in-depth literature review, highlighting a firm understanding of an important development or idea in complexity theory.

The choice of topic and relevant paper(s) must be done with coordination of the instructors, and receive a formal approval.

Timeline of the project is as follows:

DEADLINES ACTIVITIES
Second month of term Meet or email the instructors regarding the project
End of Second month of term Project Proposal due
Third month of term Write survey
Two weeks before end of term Final Project due

Project Presentations

There will be a limited number of slots available during the last week of the class for the presentation of the projects. Unlike the written survey (which is required), presentation of the final projects is voluntary. A satisfactory presentation of a well-done project counts as a strong bonus in our evaluation of students’ performance, and is strongly recommended to students with keen interest in further research in complexity theory.

Questions and Answers

We will use Piazza (a Q&A web service) for questions and answers during the semester. Note: Not available to OCW users.

Grading

The final grade will be based on four small problem sets, a grading assignment, and a final project.

The grading breakdown is as follows:

ACTIVITIES PERCENTAGES
4 Problem sets 40%
Grading and participation 20%
Final project 40%

Course Info

Learning Resource Types
Lecture Notes
Problem Sets
Projects with Examples