liveBook · Manning

Local optimum

As explained in Self-Improving Agents

A local peak is a point in a search landscape that scores higher than nearby alternatives but may be lower than points elsewhere.

A local optimum is a candidate that scores higher than every one-step neighbor, even though a better candidate exists farther away. A greedy search stops there because reaching the better candidate requires a downhill move first. section 4.1.3 section 3.6

Keeping a population of candidates spread across the search landscape can help: some may already be in the valley between peaks, where climbing toward a taller peak is possible. A population is not a guarantee; it can still wander without finding a better candidate. section 4.1.3

Figure 4.3 A local optimum, and the population that escapes it. A single greedy climber (solid) reaches the near peak and stops, since the only way onward is down. A scattered population (hollow) keeps members out in the valley, and one of them crosses to the taller peak; the whole argument of this chapter is in that single crossing.

Compare a greedy climber that halts on the near peak with a scattered population that can cross the valley. section 4.1.3

How Self-Improving Agents covers it

How deep it goes
Defines the local-optimum trap with a two-peak landscape and compares greedy search with a population.
Examples
Uses a lower nearby peak, a valley, and a taller distant peak to illustrate how a population can escape.
What you can do after
Understand why a greedy search may halt at a local optimum and how population search can help it cross a valley.
How it connects
Connects local optima to search landscapes, greedy search, hill climbing, and population search.

Also explained in

Compare how 11 books explain local optimum →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Self-Improving Agents. The section links point to the parts of the book it draws on.