Genetic Algorithm and Graph Partitioning Bui Moon 1996

"Genetic algorithms are known to be not so good at fine tuning around local optima, as pointed out in the previous section (J Grefenstette, "Incorporating Problem Specific Knowledge into Genetic Algorithms 1987")"

"we believe that clustered vertices (in a relative sense) are usually more prone to be participants of a high-quality schema on a chromosome than an arbitrary set of vertices in most graphs"

"they mention how neat it would be if an algorithm adaptively figured out the right number of clusters. is that where Newman comes in?

Notes:

• "Understanding of high dim. spaces for gauging pop size"
• "GAs in very large dim. Who does it?"
• "schemas reduce the dimensionality of the crossover component of search"