Since Quadratic Assignment Problem (QAP) belongs to NP-hard problems,
it is extremely hard to find exact solutions except for small size.
In terms of precision, to obtain  preferable solutions 
for larger problems in practical time,
approximate searching methods are flourishly studied
and many Meta Strategies such
as Genetic Algorithm(GA), Tabu Search(TS),   are designed.

Since QAP is a generalized problem containig other NP-hard problems such as Traveling Salesman
Problem (TSP) and Graph Partitioning Problem (GPP) as its special cases,
it is very difficult to use some properties of the problem.
For example, it is hard to define efficent neighborhood.
These Meta Strategies for QAP, however,
can be considered as generalized methods for these various contained 
problems. Therefore, it is worth studying methods for QAP, since
they lead to consider characteristics common to
other various specific problems.

In this thesis, these Meta Strategies for QAP are implemented to analyze
their performance.
Based on this analysis, The general framework to implove these performance are proposed, 
and its effectiveness is inspected.
Similurly, the issues of method proposed so far are also refered, and
concrete measures are given.

On the other hand, 
these generalized strategies for QAP are also applied to the instances of 
TSP and GPP. From these results, applicability of strategies for QAP
are discussed.