|
0368.3168
|
Spring 2008 |
Lectures |
Group 1: Monday 10:00-13:00, Orenstein 103 |
||||||||||||||
Instructors |
Oded Regev |
Schreiber 103 | 6407996 |
||||||||||||||
Lecture notes on the web |
Previous courses in TAU by Muli Safra,
Amnon Ta-Shma |
||||||||||||||
Textbooks |
Main references:
Other textbooks:
|
||||||||||||||
Requirements |
Attendance, homework assignments, quiz |
||||||||||||||
Prerequisites |
Computational models, Algorithms |
||||||||||||||
Other links |
A compendium
of NP-complete problems and what's known about them. |
Week |
Date |
Class Topics |
1 |
May 5,6 |
Administration; Historical remarks; Computability [AB1, S
part II]: languages, the Turing machine model of computation and its
robustness, the universal turing machine, uncomputable functions; Time
complexity [AB2,S7]: the class P. |
|
May 6 |
Circuit Complexity: The classes SIZE(T(n)) and P/poly; Every
language is in SIZE(n2n), but there exist functions that require
circuits of size at least 2n/10n [AB6, S9.3, P4.3]. |
2 |
May 12,13 |
The class NP, problems in NP, alternative definitions of NP,
reductions and NP-completeness, TMSAT is NP-complete [AB2]; The Cook-Levin
theorem [S7.4]. |
|
May 13 |
If there exists a unary NP-complete language then P=NP (Berman's
Theorem); NP-completeness of MAX-CUT (via a reduction from NAE-3SAT) [P 9.3]. |
3 |
May 19,20 |
How to deal with NP-complete problems; Decision vs. Search [AB2.5]; what if P=NP?;
More complexity classes: coNP [AB2.6]; EXP, NEXP; Σ2P,
Π2P; the polynomial-time hierarchy [AB5.1]. |
|
May 20 |
NP and coNP; The classes P, NP and coNP are closed under the
Kleene star. |
4 |
May 23 (Fri), 27 |
The Karp-Lipton
Theorem [AB6.2]; the time hierarchy theorem [AB3.1]; oracle machines
and the limits of diagonalization (oracle A such that PA = NPA
and oracle A such that PA ≠ NPA) [AB3.5]. |
|
May 27 |
MINIMUM-CIRCUIT is in Π2P [P 17.2]; Σ2P = NPSAT [AB5.5]; At least one of
P=NP and NP=EXP is false. |
5 |
May 26, June 3 |
Space complexity [AB4,S8]; PSPACE, NPSPACE,
L, NL; PATH is in NL; Savitch's theorem [AB4.3,S8.1]; TQBF is
PSPACE-complete [AB4.3]. |
|
June 3 |
Space
Complexity: An example of a language in L; An alternative definition of NL [AB4.4.1]; 2SAT is in coNL [P9.2]. |
6 |
June 2,10 |
Logspace Reductions, Composition in L [AB4,S8.1]; PATH is NL-complete [AB4.4]; The class ZNL; Immerman-Szelepcsényi Theorem: NL=coNL [AB4.4.2]
(to be continued). |
|
June 10 |
Padding argument: NL ≠ DTIME(n); 2SAT is NL-complete; Reciting Immerman-Szelepcsényi Theorem. |
7 |
June 16,17 |
Optimization
problems, approximation algorithms, 2-approximation for Vertex Cover
(VC) via maximal matching [P13.1]. Linear
programming, Integer programming, VC ≤p IP. LP is in P and IP is NP-hard. Relaxations. Rounding. A
2-approximation for (weighted) VC via LP relaxation. A lnn-approximation for Set Cover [V2.2.1]. |
|
June 17 |
TSP with
triangle inequality is NP-complete; 3/2-approximation for TSP with triangle
inequality [V3.2.2]. |
8 |
June 23,24 |
Quiz; Proof by the probabilistic method: Every graph
has a cut of size |E|/2. Integer program for MAX-CUT. Goemans-Williamson
approximation algorithm for MAX-CUT and its analysis. Sampling a unit vector
in Rn (informal). |
|
June 24 |
Gap problems; Hardness of Gap-Problems implies Hardness of
approximation. Applied to E4SAT. |
9 |
June 30, July 1 |
Primality testing: groups and subgroups, Fermat's test, Carmichael numbers,
fast exponentiation, Miller-Rabin
test. |
|
July 1 |
Randomized Algorithms: Karger's algorithm
for MIN-CUT [MR1.1]; Every graph has O(|V|2) minimum
cuts. |
10 |
July 7,8 |
Probabilistic complexity classes: BPP, RP (and coRP), ZPP [AB7,P11]; Amplification
of BPP using Chernoff’s
inequality [AB7.4,P11.2] ; BPP is contained
in P/poly [AB7.6]; BPP is contained
in Σ2P [AB7.7,P17.2]. (Oded Goldreich's notes) |
|
July 8 |
If SAT is in BPP then NP=RP; Equivalence of two definitions of ZPP. |
11 |
July 14,15 |
OWF [AB10.1] Discrete Log and El-Gamal [MOV8.9.1]. Summary. |
|
July 15 |
The conditional expectation technique, applied on MAX-E3SAT and
MAX-CUT. |