Examples of global search

This example is from Wikipedia and may be reused under a CC BY-SA license.
Then more than a dozen software packages for complete globalsearch are described.
It is therefore evident that a globalsearch of the entire solution space has a tendency to be computationally intractable.
This same example is estimated to require over 20,000 years to solve by a globalsearch on a similar machine.
The basic idea of globalsearch is to represent and manipulate sets of candidate solutions or solution spaces.
In the rest of the paper we illustrate the process of deriving a globalsearch algorithm by applying it to an example.
The efficiency of globalsearch methods enables us to keep only small solution fragments, which leads to a larger neighbourhood leaving more room for improvement.
The population-based genetic algorithm can be regarded as a globalsearch algorithm whereas representative-agent-type search algorithms are necessarily local.
The procedure has shown to be competitive compared to both the pure globalsearch procedure and simple local search methods based on simpler neighbourhoods.
Each node in the globalsearch tree consists of a solution to a conjunctive problem plus the set of remaining disjunctions.
In particular, we explore the distinction between globalsearch and local search.
The key idea is that at each step, a given neighbourhood of a certain solution is searched using a globalsearch approach.
This globalsearch approach is used for completing fragments into solutions in the local search approach (same branching scheme).
Globalsearch is used to find the best solution within a neighbourhood.
The algorithm consistently displays drastic improvements over a globalsearch with demonstrated effort reduction of up to six orders of magnitude.
They are effective at performing globalsearch (in probability) instead of local optima.
Constraint programming is generally considered as a globalsearch technique.
Finally, we show how ad hoc algebras can considerably enhance the design of heuristics based on local and globalsearch within a constraint-programming environment.
Restricting communication to a finite subset of the population ensures that, over some period of time, there is sufficient heterogeneity in the population to allow for a good globalsearch.
While the algorithm makes no efficiency guarantees, experience with robot design problems has consistently shown dramatic improvement over a globalsearch with low sensitivity to initial conditions.
Our constraint-satisfaction algorithms fall into the class of globalsearch algorithms.
As in biological bee colonies, a small number of scouts keeps exploring the solution space looking for new regions of high fitness (globalsearch).
From
Wikipedia