The code of RandomForKnapsack is not particularly good, as it creates
new arrays on the heap at each step.
Write a new RandomForKnapsackOptimized class, in which ONLY 2 arrays
are instantiated: one to keep the best solution seen so far, and one as
a buffer to randomize at each step to evaluate a new configuration.
When a new, better configuration is found, just swap the two array pointers.
Create a new RandomForKnapsackOptimizedTest, based on RandomForKnapsackTest,
to test this new implementation.
Consider the class HillClimbingForQueens.
Modify the Hill Climbing algorithm to use Steepest Ascent:
instead of climbing to the first better element in the neighbourhood,
first look at all the candidate solutions in the neighbourhood, and
then move toward the best.
Implement such new variant in a class called SteepestAscentHCforQueens.
Write a test class SteepestAscentHCforQueensTest in which you test your
steepest ascent version on boards of size 8, 16, 20 and 100.
Solutions to this exercise can be found in the solutions
module, under the org.pg4200.sol10 package.