* discuss your vacuum world agents

 * Seaching

 ** perl program for searching link paths
 *** which search choices and what reasons ?
 ***  modification for bi-directional?

 chap 4

 * 4.1 - heuristic estimates of how good a node is
         basically methods of deciding when to expand which node

 * 4.2 - creating heuristic function aumatically from "relaxed"
         version of problem
       - pattern databases for sub-problems as heuristics
       - learning heuristics : adjusting parameters from known solutions

 * 4.3 - local search (especially if path is unimportant)
       - global vs local solution
       - hill-climbing algorithm
       - simulated annealing search
       - beam searches - which states to keep?  (like assexual reproduction)
       - genetic algorithms : combine several "parents" to continue search
         fitness function; mutations
       
 * 4.4 - local continuous search

 * 4.5 - "online" search
       -  alternates search and action
       - stochastic,