Set-based Search

Model

→ Transition exists where there is an ext rule to go from state with facts in A to facts in B

Process

→ process selects transition rule A->B based on minimizing function $f_{wert}$ and tie breaker $f_{select}$

Instance

Design

• Measuring a fact B to compare with a fact A may be expensive → just thrown away old fact
• If not expensive, define $f_{wert}$
• If want randomness, $f_{wert}$ assigns constant val to fact then let $f_{select}$ to randomly select

Questions

What are the problem specific components of set-based search?

Fact: set of facts Ext: extension rules

Where to we need to make sure we handle tie-break in set based search? Why do we need to do so?

If multiple extensions have the same $f_{wert}$ score, we need $f_{wert}$ to pick one.

What is the purpose of a problem specific component? (Ex What does the declaration of F create for us in a Set-Based model?)

• Declaration of F create S for free
• Declaration of Ext create T for free