HomeHome Metamath Proof Explorer < Previous   Next >
Related theorems
GIF version

Theorem axsep 2698
Description: Separation Scheme, which is an axiom scheme of Zermelo's original theory. Scheme Sep of [BellMachover] p. 463. As we show here, it is redundant if we assume Replacement in the form of ax-rep 2689. Some textbooks present Separation as a separate axiom scheme in order to show that much of set theory can be derived without the stronger Replacement. The Separation Scheme is a weak form of Frege's Axiom of Comprehension, conditioning it (with xz) so that it asserts the existence of a collection only if it is smaller than some other collection z that already exists. This prevents Russell's paradox ru 1934. In some texts, this scheme is called "Aussonderung" or the Subset Axiom.

The variable x can appear free in the wff φ, which in textbooks is often written φ(x). To specify this in the Metamath language, we omit the distinct variable requirement ($d) that x not appear in φ.

For a version using a class variable, see zfauscl 2701, which requires the Axiom of Extensionality as well as Replacement for its derivation.

If we omit the requirement that y not occur in φ, we can derive a contradiction, as notzfaus 2737 shows (contradicting zfauscl 2701). However, as axsep2 2700 shows, we can eliminate the restriction that z not occur in φ.

Note: the distinct variable restriction that z not occur in φ is actually redundant in this particular proof, but we keep it since its purpose is to demonstrate the derivation of the exact ax-sep 2699 from ax-rep 2689.

This theorem should not be referenced by any proof. Instead, use ax-sep 2699 below so that the uses of the Axiom of Separation can be more easily identified.

Assertion
Ref Expression
axsep yx(xy ↔ (xzφ))
Distinct variable groups:   x,y,z   φ,y,z

Proof of Theorem axsep
StepHypRef Expression
1 ax-17 969 . . . 4 ((w = xφ) → ∀y(w = xφ))
21axrep5 2694 . . 3 (∀w(wz → ∃yx((w = xφ) → x = y)) → ∃yx(xy ↔ ∃w(wz ⋀ (w = xφ))))
3 a9e 1123 . . . . 5 y y = w
4 equtr 1129 . . . . . . . . 9 (y = w → (w = xy = x))
5 equcomi 1126 . . . . . . . . 9 (y = xx = y)
64, 5syl6 22 . . . . . . . 8 (y = w → (w = xx = y))
76adantrd 391 . . . . . . 7 (y = w → ((w = xφ) → x = y))
8719.21aiv 1284 . . . . . 6 (y = w → ∀x((w = xφ) → x = y))
9819.22i 1038 . . . . 5 (∃y y = w → ∃yx((w = xφ) → x = y))
103, 9ax-mp 7 . . . 4 yx((w = xφ) → x = y)
1110a1i 8 . . 3 (wz → ∃yx((w = xφ) → x = y))
122, 11mpg 984 . 2 yx(xy ↔ ∃w(wz ⋀ (w = xφ)))
13 an12 484 . . . . . . 7 ((w = x ⋀ (wzφ)) ↔ (wz ⋀ (w = xφ)))
1413exbii 1049 . . . . . 6 (∃w(w = x ⋀ (wzφ)) ↔ ∃w(wz ⋀ (w = xφ)))
15 ax-17 969 . . . . . . 7 ((xzφ) → ∀w(xzφ))
16 elequ1 1134 . . . . . . . 8 (w = x → (wzxz))
1716anbi1d 616 . . . . . . 7 (w = x → ((wzφ) ↔ (xzφ)))
1815, 17equsex 1150 . . . . . 6 (∃w(w = x ⋀ (wzφ)) ↔ (xzφ))
1914, 18bitr3 175 . . . . 5 (∃w(wz ⋀ (w = xφ)) ↔ (xzφ))
2019bibi2i 607 . . . 4 ((xy ↔ ∃w(wz ⋀ (w = xφ))) ↔ (xy ↔ (xzφ)))
2120albii 997 . . 3 (∀x(xy ↔ ∃w(wz ⋀ (w = xφ))) ↔ ∀x(xy ↔ (xzφ)))
2221exbii 1049 . 2 (∃yx(xy ↔ ∃w(wz ⋀ (w = xφ))) ↔ ∃yx(xy ↔ (xzφ)))
2312, 22mpbi 189 1 yx(xy ↔ (xzφ))
Colors of variables: wff set class
Syntax hints:   → wi 3   ↔ wb 146   ⋀ wa 223  ∀wal 952   = wceq 954   ∈ wcel 956  ∃wex 978
This theorem was proved from axioms:  ax-1 4  ax-2 5  ax-3 6  ax-mp 7  ax-7 960  ax-gen 961  ax-8 962  ax-9 963  ax-12 966  ax-13 967  ax-14 968  ax-17 969  ax-4 971  ax-5o 973  ax-6o 976  ax-9o 1121  ax-rep 2689
This theorem depends on definitions:  df-bi 147  df-an 225  df-ex 979
Copyright terms: Public domain