LARA

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
Next revision Both sides next revision
sav08:non-ground_instantiation_and_resolution [2008/04/01 16:34]
vkuncak
sav08:non-ground_instantiation_and_resolution [2008/04/02 10:49]
vkuncak
Line 31: Line 31:
 Illustration of rules. Illustration of rules.
  
-===== Reducing the Search Space =====+===== Combining Instantiation and Resolution ​=====
  
-More powerful rules.+These two are more powerful rules.
  
-But more choices at each step.+But we therefore have more choices at each step.
  
 If we try to do the proof, how do we know what to instantiate with? ++|we instantiate to enable subsequent resolution rule.++ If we try to do the proof, how do we know what to instantiate with? ++|we instantiate to enable subsequent resolution rule.++
  
-**Resolution with instantiation**+**Instantiation followed by resolution:**
 \[ \[
 \frac{C \cup \{\lnot A_1\}\ \ \ D \cup \{A_2\}} \frac{C \cup \{\lnot A_1\}\ \ \ D \cup \{A_2\}}
Line 45: Line 45:
 \] \]
 such that $subst(\sigma_1)(A_1) = subst(\sigma_2)(A_2)$. such that $subst(\sigma_1)(A_1) = subst(\sigma_2)(A_2)$.
 +
 +This rule generalizes the above non-ground resolution and ground resolution.
 +
 +Note: if we apply instantiation that renames variables in each clause, then $\sigma_1$ and $\sigma_2$ can have disjoint domains and we let $\sigma = \sigma_1 \cup \sigma_2$, obtaining
 +
 +One complete proof system contains:
 +  * instantiation followed by resolution
 +  * instantiation
  
 Note: $\sigma$ such that $subst(\sigma)(A_1) = subst(\sigma)(A_2)$ is called a **unifier** for $\{A_1,​A_2\}$. Note: $\sigma$ such that $subst(\sigma)(A_1) = subst(\sigma)(A_2)$ is called a **unifier** for $\{A_1,​A_2\}$.
  
-**Factoring with instantiation** +Note: a //​renaming//​ is substitution whose domain and codomain are variables and which is injective. ​ Thatis, it renames variables without clashing variable names. 
-\[ + 
-\frac{C \cup \{A_1A_2\}} +Further step: do we need to consider all possible unifiers?
-     {subst(\sigma)(C)} +
-\] +
-where $subst(\sigma)(A_1)=subst(\sigma)(A_2)$.+
  
-Further step: do we need to consider all unifiers?+Most general unifier for $\{A_1,​A_2\}$,​ denoted $mgu(A_1,​A_2)$
  
-Most general unifier.  ​To compute it we can use the standard [[Unification]] algorithm.+To compute it we can use the standard [[Unification]] algorithm.