Axiom ax-13

Description: Axiom of Quantified Equality. One of the equality and substitution axioms of predicate calculus with equality.

An equivalent way to express this axiom that may be easier to understand is ( -. x = y -> ( -. x = z -> ( y = z -> A. x y = z ) ) ) (see ax13b ). Recall that in the intended interpretation, our variables are metavariables ranging over the variables of predicate calculus (the object language). In order for the first antecedent -. x = y to hold, x and y must have different values and thus cannot be the same object-language variable (so they are effectively "distinct variables" even though no $d is present). Similarly, x and z cannot be the same object-language variable. Therefore, x will not occur in the wff y = z when the first two antecedents hold, so analogous to ax-5 , the conclusion ( y = z -> A. x y = z ) follows. Note that ax-5 cannot prove this because its distinct variable ($d) requirement is not satisfied directly but only indirectly (outside of Metamath) by the argument above.

The original version of this axiom was ax-c9 and was replaced with this shorter ax-13 in December 2015. The old axiom is proved from this one as theorem axc9 .

The primary purpose of this axiom is to provide a way to introduce the quantifier A. x on y = z even when x and y are substituted with the same variable. In this case, the first antecedent becomes -. x = x and the axiom still holds.

This axiom is mostly used to eliminate conditions requiring set variables be distinct (cf. ax6ev and ax6e , for example) in proofs. In practice, theorems beyond elementary set theory do not really benefit from such eliminations, so direct or indirect application of this axiom is discouraged now. You need to explicitly confirm its use in case you see a sensible application in a niche.

After some assisting contributions by others over the years, it was in particular the extensive work of Gino Giotto in 2024 that helped reducing dependencies on this axiom on a large scale.

Although this version is shorter, the original version axc9 may be more practical to work with because of the "distinctor" form of its antecedents. A typical application of axc9 is in dvelimh which converts a distinct variable pair to the distinctor antecedent -. A. x x = y . In particular, it is conjectured that it is not possible to prove ax6 from ax6v without this axiom.

This axiom can be weakened if desired by adding distinct variable restrictions on pairs x , z and y , z . To show that, we add these restrictions to theorem ax13v and use only ax13v for further derivations. Thus, ax13v should be the only theorem referencing this axiom. Other theorems can reference either ax13v (preferred) or ax13 (if the stronger form is needed).

This axiom scheme is logically redundant (see ax13w ) but is used as an auxiliary axiom scheme to achieve scheme completeness (i.e. so that all possible cases of bundling can be proved; see text linked at mmtheorems.html#ax6dgen ). It is not known whether this axiom can be derived from the others. (Contributed by NM, 21-Dec-2015) (New usage is discouraged.)

	Ref	Expression
Assertion	ax-13	$$⊢\neg x=y\to \left(y=z\to \forall xy=z\right)$$

Detailed syntax breakdown

Step	Hyp	Ref	Expression
0		vx	$$setvarx$$
1	0	cv	$$setvarx$$
2		vy	$$setvary$$
3	2	cv	$$setvary$$
4	1 3	wceq	$$wffx=y$$
5	4	wn	$$wff\neg x=y$$
6		vz	$$setvarz$$
7	6	cv	$$setvarz$$
8	3 7	wceq	$$wffy=z$$
9	8 0	wal	$$wff\forall xy=z$$
10	8 9	wi	$$wff\left(y=z\to \forall xy=z\right)$$
11	5 10	wi	$$wff\left(\neg x=y\to \left(y=z\to \forall xy=z\right)\right)$$