#### Abstract

Many problems arising in graph theory are difficult by nature, and finding solutions to large or complex instances of them often require the use of computers. As some such problems are $NP$-hard or lie even higher in the polynomial hierarchy, it is unlikely that efficient, exact algorithms will solve them. Therefore, alternative computational methods are used. Combinatorial computing is a branch of mathematics and computer science concerned with these methods, where algorithms are developed to generate and search through combinatorial structures in order to determine certain properties of them. In this thesis, we explore a number of such techniques, in the hopes of solving specific problem instances of interest.

Three separate problems are considered, each of which is attacked with different methods of combinatorial computing and optimization. The first, originally proposed by ErdH{o}s and Hajnal in 1967, asks to find the Folkman number $F_e(3,3;4)$, defined as the smallest order of a $K_4$-free graph that is not the union of two triangle-free graphs. A notoriously difficult problem associated with Ramsey theory, the best known bounds on it prior to this work were $19 leq F_e(3,3;4) leq 941$. We improve the upper bound to $F_e(3,3;4) leq 786$ using a combination of known methods and the Goemans-Williamson semi-definite programming relaxation of MAX-CUT. The second problem of interest is the Ramsey number $R(C_4,K_m)$, which is the smallest $n$ such that any $n$-vertex graph contains a cycle of length four or an independent set of order $m$. With the help of combinatorial algorithms, we determine $R(C_4,K_9)=30$ and $R(C_4,K_{10})=36$ using large-scale computations on the Open Science Grid. Finally, we explore applications of the well-known Lenstra-Lenstra-Lov'{a}sz (LLL) algorithm, a polynomial-time algorithm that, when given a basis of a lattice, returns a basis for the same lattice with relatively short vectors. The main result of this work is an application to graph domination, where certain hard instances are solved using this algorithm as a heuristic.

#### Publication Date

6-2013

#### Document Type

Thesis

#### Student Type

Graduate

#### Degree Name

Computer Science (MS)

#### Department, Program, or Center

Computer Science (GCCIS)

#### Advisor

Stanislaw Radziszowski

#### Advisor/Committee Member

Ivona Bezakova

#### Advisor/Committee Member

Darren A. Narayan

#### Recommended Citation

Lange, Alexander R., "Solving Hard Graph Problems with Combinatorial Computing and Optimization" (2013). Thesis. Rochester Institute of Technology. Accessed from

https://scholarworks.rit.edu/theses/5946

#### Campus

RIT – Main Campus

#### Plan Codes

COMPSCI-MS

## Comments

Physical copy available from RIT's Wallace Library at QA186 .L36 2013