­­­­­­­­CS395T: Approximability

51815 | ­Fall 2018 | M W 2-3:30 | RLP 0.104  

 

Instructor: Dana Moshkovitz

 

This class is about approximation algorithms and their limitations. It covers: combinatorial approximation algorithms; approximation algorithms based on linear programming (LP) and semidefinite programming (SDP); hierarchies of linear and semidefinite programming; limitations on LP and SDP; hardness of approximation and connections to multi-prover games and probabilistic checking of proofs; the proof of the PCP theorem, including combinatorial and algebraic techniques; sum-check; linearity testing; low degree testing; locally testable and decodable codes; composition; parallel repetition; long code, Fourier analysis and optimal inapproximability results; the Unique Games Conjecture; dictator tests and integrality gaps.

 

Format: Some of the meetings will be regular lectures, and some will be “workshops” in which the students give mini-lectures. Before each workshop there will be a schedule with the topics of the mini-lectures. Usually there will be problems to solve and a mini-lecture will present a solution, but sometimes mini-lectures will be surveys or discussions of open problems. Students should register to give mini-lectures after the release of each workshop schedule. Workshops will give students ample opportunity to give lectures and receive feedback. There are no exams or problem sets apart from workshop participation and presentation.

 

Syllabus

 

 

Date

Topic

1

Wednesday,

August 29

Overview; basic combinatorial approximation algorithms: vertex cover, set cover, 3SAT

 

Monday,

September 3

Labor Day

2

Wednesday, September 5

Approximation algorithms via Linear Programming

3

Monday, September  10

Workshop on approximation basics, including constraint satisfaction problems, label-cover, unique label-cover

4

Wednesday, Septemb­­­er 12

The primal-dual method

5

Monday, September 17

Workshop on approximation algorithms based on linear programming

6

Wednesday,

September 19,

Semidefinite programming (SDP), Max-Cut approximation and integrality gap

7

Monday,

September 24,

Workshop on SDP

8

Wednesday,

September 26,

Sum of Squares hierarchy of SDP

9

Monday,

October 1

Workshop on limitations of Sum of Squares

10

Wednesday,

October 3

Hardness of Approximation, Multi-prover games and probabilistic checking of proofs (PCP)

11

Monday, October 8

Workshop on PCP basics, including expanders and randomness efficient sequential repetition, transformation to two queries

12

Wednesday, October 10

Linearity testing

13

Monday, October 15

Workshop on linearity testing

14

Wednesday, October 17

Fourier analysis

15

Monday, October 22

Workshop on Fourier analysis

16

Wednesday, October 24

Hadamard-based PCP

17

Monday, October 29

Workshop on Hadamard-based PCP

18

Wednesday, October 21

Sum-Check and PCP with polylog queries

19

Monday, November 5

Workshop on sum-check and its many applications, low degree testing

20

Wednesday, November 7

Composition, PCP of proximity, robust PCP, decoding PCP

21

Monday, November 12

Workshop on composition, connection to code concate

22

Wednesday, November 14

Gap amplification (powering)

23

Monday, November 19

Parallel repetition

Wednesday, November 21

Thanksgiving

24

Monday, November 26

Workshop on gap amplification and parallel repetition

25

Wednesday, November 28

The long code and optimal inapproximability

26

Monday, December 3

Workshop on long code and optimal inapprox

27

Wednesday, December 5

The Unique Games Conjecture

28

Monday, December 10

Workshop on the Unique Games Conjecture