Talk:Von Neumann–Bernays–Gödel set theory

From Wikipedia, the free encyclopedia

I accidently started a duplicate: http://en.wikipedia.org/wiki/Neumann-Bernays-G%f6del_Axiomatic_Set_Theory

I was planning a full exposition of the system... should I move stuff over to here?

Dustinmulcahey

Yes, there should be just one article, at least at first. Later there might be a need for seperate articles. Paul August ☎ 20:10, Mar 9, 2005 (UTC)

Contents

[edit] Proposed new name: "Von Neumann-Bernays-Gödel set theory"

I think that a better name for this article would be: "Von Neumann-Bernays-Gödel set theory". I will move it there, and fix all the links unless someone objects. Paul August ☎ 16:45, Mar 11, 2005 (UTC)

This was my intent also, I will do it. MarSch 14:51, 19 Apr 2005 (UTC)

Is there any particular reason why this article, along with Zermelo–Fraenkel set theory and Shimura–Taniyama theorem, all use the "–" character in the article name, which has to be percent-escaped in the URL as "%E2%80%93", rather than a simple ASCII "-" which would not need escaping? Is this some rogue standard of the mathematical literature of which I am unaware, or perhaps just an artifact of the character set being used by the person(s) who last moved them? Finally, would it not make more sense to move them all back to the forms that are perhaps more likely to be typed directly without special character sets. Presently those three, at least, exist as redirects to the article versions with the Unicode dashes. I'd appreciate it if someone could clue me in. --KGF0 ( T | C ) 07:52, 25 October 2006 (UTC)

For some reason (which I do not understand), some people are on a campaign to replace all hyphens by ndashes. I concluded that "resistance is useless". Sorry. JRSpriggs 10:09, 25 October 2006 (UTC)
Ah, thanks. I found the rationale in WP:MOS: "it is used in compound adjectives referring to multiple people, so as to clarify that for example the name of the Poincaré–Birkhoff–Witt theorem refers to three people while the Birch–Swinnerton-Dyer conjecture refers to two (one of them called Swinnerton-Dyer)." Obnoxious to some degree, but sensible in a pedantic sort of way. --KGF0 ( T | C ) 12:25, 25 October 2006 (UTC)
Yes, I see "The hyphen (-) is used to form compound words. The en-dash (–) is used to specify numeric ranges. The em-dash (—) can be used to link clauses of a sentence. Other dashes, notably the double-hyphen (--), should be avoided.". Frankly, I think that that is nit-picking. As a work around for your problem, in some cases I can copy the text or title with cut-and-paste (control-c followed by control-v) rather than having to type in that mess. I also keep a list of frequently used links on my talk page. JRSpriggs 06:10, 26 October 2006 (UTC)
Why should you ever need to type it? There's a redirect from the hyphens version. If there's not a redirect from the version you like to type, well, add it. Redirects are almost free; as long as the target is unambiguous, and the redirect isn't tendentious in some way, no one should mind. --Trovatore 07:37, 26 October 2006 (UTC)
Also I often use the category system to navigate among articles. This only requires one to point and click rather than type in the name of the article. JRSpriggs 07:46, 27 October 2006 (UTC)

[edit] bug?

I got this message after editing:

Warning: gzuncompress(): data error in /usr/local/apache/common-local/php-1.4/includes/memcached-client.php on line 868

am reporting it. MarSch 15:53, 19 Apr 2005 (UTC)


[edit] axiomatization

I added axiomatization in the spirit of the article on Morse-Kelley set theory Randall Holmes 19:40, 22 December 2005 (UTC)

[edit] alternative axioms for NBGo aka vNBGU

I can't find my copy of Set Theory for the Mathematician (Jean E. Rubin), but I found Equivalents of the Axiom of Choice II (H. Rubin and J.E. Rubin)

I'll try to copy the axioms used there, but representing it as a two-sorted theory is difficult....

Primitive terms are Cl (or Class) and/or Atom and the element relation, ∈.


A1: Characterization of Atoms
  1. Atom(X) \leftrightarrow \neg Cl(X)
  2. Atom(X) \rightarrow (\forall Y.Y \not \in X)
D1: Definition of Set
Set(X) \iff Class(X) \wedge (\exists Y.Class(Y) \wedge X \in Y)
A2 = Axiom of Extensionality
A3 = Empty Set Axiom
A4 = Class Construction Schema
A5 = Power Set Axiom
A6 = Union Axiom
A7 = Pairing Axiom
A8 = Axiom of Replacement
Func(F) \wedge Set(domain(F)) \rightarrow Set(range(F))
A9 = Axiom of Infinity
A'10 = Axiom of Foundation
If X is a non-empty class, all of whose members are sets (not urelemnts), then there is a element u of X such that u \cap X = \emptyset

(Choice is not assumed in that book, for obvious reasons, but corresponding axioms including)

Set Choice: If x is a set of pairwise-disjoint non-empty sets, there is a choice set c such that if y ∈ x then \left| c \cap y \right|=1
Class Choice 1: If X is a class of disjoint non-empty sets, there is a choice class C such that if y ∈ x then \left| C \cap y \right| = 1
Class Choice 1a: There is a function F on the class of all non-empty sets such that F(x) &isin x;
Class Choice 2: For any (binary) relation R there is a function F ⊆ R with the same domain.

and the strongest form

Class Trichotomy: Any two proper classes are equipollent. (Or any proper class is equipollent to V.)
or the equivalent V (the universe) is equipollent to On.

Arthur Rubin | (talk) 23:27, 24 January 2006 (UTC)

Also, Separation (If y is a Class, and y ⊆ x, and x is a set, then y is a set) follows:

Proof: if y is empty, then y is the empty set. Otherwise, let z be an element of y, and construct

F = \left\{ (u,v) \mid u \in x \wedge ((u \in y \wedge v = u) \or (u \not \in y \wedge v = z)) \right\}

"Clearly", domain(F)=x and range(F)=y. Arthur Rubin | (talk) 23:28, 24 January 2006 (UTC)

[edit] comparative axiomatization

Thanks to Arthur Rubin for posting his mother's axiom set; I wrote mine more or less off the top of my head (though it is equivalent to standard formulations!)

Here are the obvious correspondences:

  • A1 has no analogue in my set (because I assume full extensionality; I'm not allergic to atoms, though).
  • A2 is my Extensionality and extensionality as modified by the absence of atoms (of course, Rubin's theory is one-sorted, something I'm also not allergic to).
  • A3 is redundant in my set of axioms; I'm not sure if it is in Rubin's.
  • A4 is the same as my Class Comprehension.
  • A5 is the same as my power set.
  • A6 is the same as my union.
  • A7 is the same as my pairing.
  • A8 is a consequence of my Limitation of Size; a weaker form of Limitation of Size asserting that a class function with a set domain also has a set range would be equivalent to A8 (my current version says "injection", which is not appropriate in the absence of choice; I'll change that).
  • A9 is probably the same as my infinity (I haven't looked at Rubin's axiom in the book yet).
  • A10 is again probably the same as my Foundation (I need to look).

Limitation of Size also entails global choice, which is not an axiom in Rubin, but which she probably did consider (I'll look).

I have no objection to Rubin's axioms, as long as von Neumann's very powerful axiom (with its strong historical claim to connection with this theory) is discussed. Also, a full analysis of the class comprehension axiom into a finite subscheme (as I present here) is nice, and is not found in Rubin (she lists a finite subscheme but does not motivate it).

Do note that some people seem to care about having a two-sorted theory (or at least "two-sorted language" (I'm not one of them, but I wrote it with them in mind). Randall Holmes 00:59, 25 January 2006 (UTC)

Rubin's notation strikes me as somewhat old-fashioned; I would write some of it differently. Randall Holmes 00:48, 25 January 2006 (UTC)

[edit] latest modifications

I changed the weak form of Limitation of Size so that it agrees with Rubin's A8. This is better in the absence of Choice than the form I gave originally. Randall Holmes 00:56, 25 January 2006 (UTC)

[edit] I do have a copy...

I do have a copy of Rubin, Set Theory for the Mathematician, at my fingertips. Do you want anything in particular? Randall Holmes 07:13, 25 January 2006 (UTC)

I finally found it. It's behind a picture of my mother (!) Arthur Rubin | (talk) 14:54, 25 January 2006 (UTC)

[edit] comment in edit summary is wrong

the fix I just made to the weaker form of Limitation of Size is not needed; Infinity by itself implies that the empty set is a set (since it is an element of the provided set). Randall Holmes 17:02, 25 January 2006 (UTC)

[edit] Sci.math and edits

There's a thread on this article on sci.math/sci.logic which complains about it; that seems to have led to an edit and reversion. Maybe it would help if people explained their reasons here. Gene Ward Smith 22:50, 7 May 2006 (UTC)

[edit] = NBG or NGB?

I think, among mathematicians this theory is called "NGB" rather than "NBG". But it is not quite important, I admit. 89.133.5.184 21:14, 16 August 2007 (UTC)

My late mother used vNBG or NBG. Do you have a particular mathematician in mind? — Arthur Rubin | (talk) 23:30, 16 August 2007 (UTC)

Oh no. It was a fatal error, sorry. I looked after it and rather NBG, really. Consider it as I haven't said anything. 89.133.5.184 15:15, 17 August 2007 (UTC)

No problem. I've done worse. — Arthur Rubin | (talk) 15:46, 17 August 2007 (UTC)

[edit] Axiom of the diagonal?

I am not sure that the axiom of the diagonal is strictly necessary, it follows from the other axioms. This is because:

x \in [=] \leftrightarrow \exists y : (y \in x) \wedge (\forall u)(u \in x \implies u = y) \wedge (\exists z : (z \in y) \wedge (\forall v)(v \in y \implies v = z))

Then we can break down the formula on the right into a ppf (we can replace = using extensionality and the truth table for equivalence, "for all" by "not exists not", and implies by also using its own truth table). We can then prove that this class exists by using the axioms of ranges, intersection, complement and membership. Kidburla2002 20:40, 9 September 2007 (UTC)

I think you need to go into more detail to break that out, but it doesn't really matter. We're not looking for a minimal axiomitization here. — Arthur Rubin | (talk) 23:05, 9 September 2007 (UTC)

[edit] conservative extension?

  • NBG (as written) is a conservative extension of ZFC (at least I think it is think; strong global choice, which follows from "limitation of size", may have a consequences of a choice schema which don't follow from set-choice.)
  • NBG (+ separation (which follows from "empty set" and replacement) - "limitation of size") is a conservative extension of ZF.
  • NBGU (modified to include urelements, as possibly a 3-typed theory) may be a conservative extension of ZFU+AC
  • NBGU (with the appropriate separation modifications ) may be a conservative extension of ZFU.

If we can find sources for the above statements, perhaps they should all be in the article? — Arthur Rubin | (talk) 18:43, 12 October 2007 (UTC)

[edit] reverse assoc2?

Isn't the reverse of assoc2 required, rather than the version given in the article, in order to return the argument list to standard form? Example: (1,(2,(3,(4,(5,(6,7)))))) A1-> ((1,2),(3,(4,(5,(6,7))))) A1-> (((1,2),3),(4,(5,(6,7)))) C2-> (4,( ((1,2),3) , (5,(6,7)) )). Now, using A2, we get: A2-> (4,( ( ((1,2),3) , 5 ),(6,7)) A2-> (4,(((((1,2),3),5),6),7) whereas using reverse A2, or A2*, we get: A2*-> (4,( (1,2) , (3,(5,(6,7))) )) A2*-> (4,(1,(2,(3,(5,(6,7)))))).

89.0.153.121 (talk) 22:39, 28 January 2008 (UTC)

Receiving no comment, I edited said axiom, and my edit was undone. Any explanation? 89.0.47.251 (talk) 20:25, 23 February 2008 (UTC)
I did not want to take the time to examine your question until you made an issue of it by actually editing the article. After reading and understanding the explanatory paragraph after axiom Assoc2, I realized that the original version was correct and your "correction" was wrong. So I changed it back. Your example above has A2 and A2* reversed as far as what their effects are. Assoc2 has to move the middle term back from being associated with the term on the left to being associated (as it originally was) with the term on the right, and that is what it does. JRSpriggs (talk) 07:37, 24 February 2008 (UTC)
I re-checked, and you are correct. My bad. 89.0.225.73 (talk) 19:56, 24 February 2008 (UTC)