occurs check

In theorem proving, unification without the occurs check can lead to unsound inference. For example, the Prolog goal

$X\; =\; f(X)$

will succeed, binding X to a cyclic structure which has no counterpart in the Herbrand universe.

As another example,{{cite book| author=David A. Duffy| title=Principles of Automated Theorem Proving| year=1991| publisher=Wiley}}; here: p.143

without occurs-check, a resolution proof can be found for the non-theoremInformally, and taking $p(x,y)$ to mean e.g. "x loves y", the formula reads "If everybody loves somebody, then a single person must exist that is loved by everyone."

$(\backslash forall\; x\; \backslash exists\; y.\; p(x,y))\; \backslash rightarrow\; (\backslash exists\; y\; \backslash forall\; x.\; p(x,y))$: the negation of that formula has the conjunctive normal form $p(X,f(X))\; \backslash land\; \backslash lnot\; p(g(Y),Y)$, with $f$ and $g$ denoting the Skolem function for the first and second existential quantifier, respectively. Without occurs check, the literals $p(X,f(X))$ and $p(g(Y),Y)$ are unifiable, producing the refuting empty clause.

File:Example for syntactic unification without occurs check leading to infinite tree svg.svg

Prolog implementations usually omit the occurs check for reasons of efficiency, which can lead to circular data structures and looping.

By not performing the occurs check, the worst case complexity of unifying a term

$t\_1$ with term $t\_2$

is reduced in many cases from

$O(\backslash text\{size\}(t\_1)+\backslash text\{size\}(t\_2))$

to

$O(\backslash text\{min\}(\backslash text\{size\}(t\_1),\backslash text\{size\}(t\_2)))$;

in the frequent case of variable-term unification, runtime shrinks to

$O(1)$.

{{refn|group=nb|Some Prolog manuals state that the complexity of unification without occurs check is

$O(\backslash text\{min\}(\backslash text\{size\}(t\_1),\backslash text\{size\}(t\_2)))$ (in all cases).{{cite tech report|author1=F. Pereira |author2=D. Warren |author3=D. Bowen |author4=L. Byrd |author5=L. Pereira | title=C-Prolog's User's Manual Version 1.2| year=1983| institution=SRI International|URL=http://www.cs.duke.edu/csl/docs/cprolog.html| access-date=21 June 2013}}

This is incorrect, as it would imply comparing arbitrary ground terms in constant time (by unifying $eq(t\_1,t\_2)$ with $eq(X,X)$).}}

Modern implementations, based on Colmerauer's Prolog II,

{{cite book| author=A. Colmerauer| author-link=Alain Colmerauer| title=Prolog and Infinite Trees| year=1982| publisher=Academic Press|editor1=K.L. Clark |editor2=S.-A. Tarnlund }}

{{cite journal|author1=M.H. van Emden |author2=J.W. Lloyd | title=A Logical Reconstruction of Prolog II| journal=Journal of Logic Programming| year=1984| volume=2| pages=143–149}}

{{cite journal|author1=Joxan Jaffar |author2=Peter J. Stuckey | title=Semantics of Infinite Tree Logic Programming| journal=Theoretical Computer Science| year=1986| volume=46| pages=141–158| doi=10.1016/0304-3975(86)90027-7| doi-access=free}}

{{cite journal| author=B. Courcelle| author-link=Bruno Courcelle| title=Fundamental Properties of Infinite Trees| journal=Theoretical Computer Science| year=1983| volume=25| issue=2| pages=95–169| doi=10.1016/0304-3975(83)90059-2| doi-access=free}}

use rational tree unification to avoid looping. However it is difficult to keep the complexity time linear in the presence of cyclic terms. Examples where Colmerauers algorithm becomes quadratic {{cite conference|author1=Albertro Martelli | author2=Gianfranco Rossi | title=Efficient Unification with Infinite Terms in Logic Programming| conference=The International Conference oj Fifth Generation Computer Systems| year=1984| url=https://www.ueda.info.waseda.ac.jp/AITEC_ICOT_ARCHIVES/ICOT/Museum/FGCS/FGCS84en-proc/84eFLP2-2.pdf}} can be readily constructed, but refinement proposals exist.

See image for an example run of the unification algorithm given in Unification (computer science)#A unification algorithm, trying to solve the goal $cons(x,y)\; \backslash stackrel\{?\}\{=\}\; cons(1,cons(x,cons(2,y)))$, however without the occurs check rule (named "check" there); applying rule "eliminate" instead leads to a cyclic graph (i.e. an infinite term) in the last step.

ISO Prolog implementations have the built-in predicate unify_with_occurs_check/2 for sound unification but are free to use unsound or even looping algorithms when unification is invoked otherwise, provided the algorithm works correctly for all cases that are "not subject to occurs-check" (NSTO).7.3.4 Normal unification in Prolog of ISO/IEC 13211-1:1995. The built-in acyclic_term/1 serves to check the finiteness of terms.

Implementations offering sound unification for all unifications are Qu-Prolog and Strawberry Prolog and (optionally, via a runtime flag): XSB, SWI-Prolog, [http://ctp.di.fct.unl.pt/~amd/cxprolog/ CxProlog], [http://tau-prolog.org/ Tau Prolog], [https://github.com/trealla-prolog/trealla Trealla Prolog] and [https://github.com/mthom/scryer-prolog/ Scryer Prolog]. A variety

{{cite journal|author1=Ritu Chadha |author2=David A. Plaisted | title=Correctness of unification without occur check in prolog| journal=The Journal of Logic Programming| year=1994| volume=18| issue=2| pages=99–122| doi=10.1016/0743-1066(94)90048-5| doi-access=free}}{{cite conference|author1=Thomas Prokosch |author2=François Bry| title=Unification on the Run| conference=The 34th International Workshop on Unification| year=2020| pages=13:1–13:5| url=http://www3.risc.jku.at/publications/download/risc_6129/proceedings-UNIF2020.pdf}} of optimizations can render sound unification feasible for common cases.

{{cite journal| author=W.P. Weijland| title=Semantics for Logic Programs without Occur Check| journal=Theoretical Computer Science| year=1990| volume=71| pages=155–174| doi=10.1016/0304-3975(90)90194-m| doi-access=free}}

{{reflist|group=nb}}

{{reflist}}

Category:Automated theorem proving

Category:Logic programming

Category:Programming constructs

Category:Unification (computer science)