liveBook · Manning

Variation

As explained in Self-Improving Agents

The creation of a changed candidate from an existing candidate in an evolutionary search.

Variation creates a changed candidate from an existing one. In the framework, `mutate()` performs this step; the new artifact also records its parent, preserving its lineage. section 4.1.2

Changing an artifact

A parent artifact can produce a child with `parent.mutate(new_content=rewrite)`. Unlike the toy example that changes an integer, this keeps track of the artifact’s parent. section 4.1.2

Figure 4.2 The same four mechanics, in two vocabularies. Evolution runs down the left column and the framework down the right: variation maps to mutate(), heredity to the parent pointer, selection to the Signal, and time to the search round. Reading across each row is how the evolutionary loop becomes the real primitives, with nothing added but new names.

See how variation fits alongside heredity, selection, and time in the evolutionary loop. section 4.1.2

How Self-Improving Agents covers it

How deep it goes
Defines variation as producing a changed copy with `mutate()`, but does not specify a mutation strategy.
Examples
Shows an integer changed by one step and an artifact rewritten through `parent.mutate(new_content=rewrite)`.
What you can do after
Apply variation by calling `mutate()` to produce a changed artifact in the framework loop.
How it connects
Connects variation to mutation, heredity, selection, signals, lineage, and search rounds.

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.