Re:'Equals' Has More Than One Meaning in Math
by auggierose on Wednesday June 19, 2024 @02:09AM
(#40725332)
You ellipsed ([...]) the "can assume" part of my statement. Every mathematician knows the different ways of constructing numbers. It really is a simple YES.
Re:'Equals' Has More Than One Meaning in Math
by Someone on Wednesday June 19, 2024 @05:49AM
(#40726631)
No, it isn’t. Whether it is depends on what level the mathematician operates, and will differ on what they’re working on. That’s precisely the subject of this article: there’s an intuitive idea that, upon close scrutiny, turns out to before complex than it appears to be.
So, if you want to start from the basis in a proof assistant, you’ll have to discriminate between them and either give them different names or introduce context from which the computer can infer what you mean.
If, for example, you want to prove that two definitions of integers are equivalent, you need two different names.
Re:'Equals' Has More Than One Meaning in Math
by auggierose on Wednesday June 19, 2024 @06:22AM
(#40726821)
It helps to state carefully what you are trying to achieve. I want a theorem prover where constructing numbers such that ℕ ⊂ ℤ is easy. Doing this in a theorem prover is not trivial, for example none of the type based theorem provers allow that (although, Isabelle/ZF might qualify; it is based on set theory on top of a weak intuitionistic type theory; but here ℕ and ℤ are not types, but sets of the same type). It is trivial though if you are a mathematician, no matter what you are working on. You know that these constructions can be done, and that as a result you can assume ℕ ⊂ ℤ (if you want to). So as long as your theorem prover doesn't support something as trivial as that, it is lacking.
And yes, that is exactly what Buzzard is talking about. For the case I am talking about here though, that is just table stakes: make sure your prover can do ℕ ⊂ ℤ ⊂ ℝ ⊂ ℂ. If you can do that, you can start worrying about more advanced issues of equality, and you might find that you have already solved a significant amount of them.
Re:'Equals' Has More Than One Meaning in Math
by robinzfc on Wednesday June 19, 2024 @12:12PM
(#40729624)
I can confirm that in Isabelle/ZF one can set up a context (locale) with the meaning of the ℕ ℤ ℝ ℂ symbols defined so that ℕ ⊂ ℤ ⊂ ℝ ⊂ ℂ. However ℕ for example will not be equal in such case to the canonical set of ZF natural numbers where 1 = {0}, 2 = {0,1} etc.
Yes and no.
NO: there already are multiple ways to construct “1 the natural number”, even if you restrict that to set-theoretic ones. See https://en.wikipedia.org/wiki/Set-theoretic_definition_of_na.... Von Neumann defined the natural number 1 as {∅} (the set with as only element the empty set), Frege and Russell as the equivalence class of all sets with a single element (and that’s not a circular definition)
Integers typically are constructed from those (See for example https://mathesis-online.com/integers, which defines an integer i as the equivalence class of pairs (a,b) of natural numbers such that a = b + i)
In this construction, “the integer 1” is the infinite set {(1,0},{2,1},{3,2},…}. That obviously (in the mathematical sense) is different from the one element set {∅}
Rationals and reals are constructed on top of that (see for example https://www.quora.com/How-is-the-set-of-real-numbers-constru... for a way to construct those)
YES: when you decide to ignore the low-level stuff and just use the intuitive definition of reals and integers.
(Alternatives:
- don’t ignore it, but start at the basis, calling the integers something else, then prove equivalence between a well-defined subset of the reals and the integers and then give that subset the name “integers”.
- directly construct the reals without first constructing the integers and then define the integers as a subset of them. I’m not aware of any way to define the reals without first defining integers, though. )