With the hill-climbing search algorithm, I do not quite understand how the algorithm can get stuck at "plateaus/ridges." Moving one step back, what exactly denotes a "maximum" or "minimum" in a problem? For instance, in the 8-queens problem, it is said that it hits a local minima 86% of the time. Can someone please elaborate more on how and what denotes maximas and minimas and given this, how the algorithm can get stuck at a plateau.
What I know so far is the following:
- Expand the current state, find the best neighbor, and that becomes the new current state.
- No looking beyond one step from where you are, no memory