News for hackers. Stuff that matters.
Re:'Equals' Has More Than One Meaning in Math by gizmo686 on Tuesday June 18, 2024 @02:11AM (#40714551)
All of the ways is quite the ask.

There are 2 main main approaches: explicit constructions and axiomatic constructions.

Dedikend cuts and Von Neuman numbers are both examples of explicit constructions. You can point to exactly what a given number "is", and then prove properties about you construction.

The other approach is the axiomatic construction, where you define what properties you want your numbers to have. You then (hopefully) prove that such a system exists and is, in some sense unique.

Taking the integers, the most common explicit construction is the set-theoretic Von Neumon construction, where 0={}, 1={0}, 2={0,1}, 3={0,1,2}...

You can also imagine defing 0={}, 1={0}, 2={1}, 3={2}

Note that while we say these are equivalent definitions, they produce different results in the sense that, for example, 0∈3 in the former, but not in the latter. [0].

You can also define natural numbers as a subset of the cardinals. In this construction, 3 represents the collection of all sets that can be put in a 1-to-1 correspondance to the set { {}, {{}}, {{{}}} }. Here, the exlusion of various infinities is a rather arbitrary part of the definition.

You can also define integers as a subset of combinatorical games, given by 0={|}, 1={0|}, 2={1|}. This naturally leads to negative numbers, -1={|0}, and a zoo of "numbers" not seen outside of combinatorical game theory, such as ↑ = {0|{0|0}}. Still, if you restrict yourself to the integer looking numbers and follow the game theory rules of addition, you get something equivalent to the natural numbers.

You can also define them as strings constructed from the alphabet of a single character, so 0="", 1="a", 2="aa". (This is the free monoid over a single set)

Lambda Calculus gives us the church encoding, where:

  * 0 = λf.λx.x
  * 1 = λf.λx.f x
  * 2 = λf.λx.f (f x)
Under the axiomatic definition of numbers, the classical definition is given by the 9 Peano axioms, which amount to assuming the existence of 0 and a "succesor" function S(n) informally corresponding to S(n) = n+1.

All of the above examples are things I have actually seen and that arise naturally in their respective branch of mathametics.

[0] This contradiction is not actually an issue, because no one ever actually says 0∈3, even though it is technically true in some constructions.

Re:'Equals' Has More Than One Meaning in Math by anon291 on Wednesday June 19, 2024 @12:34AM (#40724846)
These are all constructions though. They don't say what the natural numbers are.

The natural numbers is a set with a starting element and that has, for each element, a thing immediately following.

Things like 0 in 3 make no sense since 0 refers to the first thing in whatever representation you have, not a particular one.

I love formal theorem proving and provers, but this shouldnt lead us to believe that the natural numbers are fundamentally any particular representation

N is defined by its properties not its representation.

Re:'Equals' Has More Than One Meaning in Math by lmm on Wednesday June 19, 2024 @12:59AM (#40725011)
> These are all constructions though. They don't say what the natural numbers are.

A construction is exactly what would tell you what they are.

> The natural numbers is a set with a starting element and that has, for each element, a thing immediately following.

OK, but, like, which one? There are lots of sets that this describes.

> Things like 0 in 3 make no sense since 0 refers to the first thing in whatever representation you have, not a particular one.

If 0 and 3 are sets then we can legitimately ask whether 0 is a member of 3 or not. If 0 and 3 aren't sets then what are they? If the natural numbers are a set then 0 and 3 should also be sets, since set theory only talks about sets.

> N is defined by its properties not its representation.

Then what is it? If you say it's a set with those properties, then it should be a set and we can ask set questions about it. If you say it's somehow all of the sets with those properties, or some other concept that somehow encompasses them, then you still haven't explained what it actually is.

Re:'Equals' Has More Than One Meaning in Math by BalinKing on Wednesday June 19, 2024 @02:20AM (#40725386)
Not a direct reply to your question, but it seems to me that GP is describing a "structuralist" view (cf. https://en.wikipedia.org/wiki/Structuralism_(philosophy_of_m..., on which I'd guess much has been written over the years (both for and against).
Re:'Equals' Has More Than One Meaning in Math by anon291 on Wednesday June 19, 2024 @12:51PM (#40730017)
I think so yes.

The article here espouses a constructivist view of mathematics

While I like that view personally, I think it's wrong to act as if it's the only one like this article does.

Many people have different views.

Re:'Equals' Has More Than One Meaning in Math by lupire on Wednesday June 19, 2024 @07:59AM (#40727413)
If I choose to wear a fedora, is a fedora an essential part of who I am? Or an arbitrary choice that be changed?

This question goes back to Plato.

You use Von Neuman naturals. I use Peano naturals. Which are the real naturals? Can they both be the naturals, even though they are clearly different? The natural numbers are not any one construction. They are the abstract idea of whose properties are what all the constructiona have in common.

Math is like God. We use feeble human language to approximately talk about something that is beyond our ability.

Re:'Equals' Has More Than One Meaning in Math by anon291 on Wednesday June 19, 2024 @12:50PM (#40730002)
Right and this is my point the view espoused in the article is not universal
Re:'Equals' Has More Than One Meaning in Math by lmm on Wednesday June 19, 2024 @06:31PM (#40733101)
> Math is like God. We use feeble human language to approximately talk about something that is beyond our ability.

That sounds like giving up. The approach the article is describing is a lot more satisfying - forming this concept of equivalence, and then using it to give genuine, rigorous answers to questions like:

> You use Von Neuman naturals. I use Peano naturals. Which are the real naturals? Can they both be the naturals, even though they are clearly different?

Re:'Equals' Has More Than One Meaning in Math by anon291 on Wednesday June 19, 2024 @11:52AM (#40729443)
> OK, but, like, which one? There are lots of sets that this describes.

Any, and all. One can do the thinking without specifying what. I believe this is the notion of a 'universal property' in category theory.

> If 0 and 3 aren't sets then what are they?

I mean, if you want to give me a construct I would use, I'd say the following

0 = \s -> \z -> z 3 = \s -> \z -> s s s z

If you say that these are akin to church numerals, that is correct if you look at it as a lambda calc term. However I'm saying something much stronger than just a lambda calculus term. That's just my notation.

I'm saying that zero is the concept of taking two things (one an initial thing and the other a method to take one thing and get the next thing) and then returning the initial thing. While three is the concept of taking the same two things and then doing the method three times on the first thing. I used lambda notation because that is an easy way to convey the idea, not because that is what three is.

> Then what is it? If you say it's a set with those properties, then it should be a set and we can ask set questions about it. If you say it's somehow all of the sets with those properties, or some other concept that somehow encompasses them, then you still haven't explained what it actually is.

It's a set in the colloquial sense, not in the ZFC sense.

There's an idea of a thing, and then the definition of the thing. Both are valid in my mind.

For example, heat is the movement of atoms. It's also the energy in a thing. But it's also the feeling of hot that I cannot describe. It's also an idea in its own right. Things can be many things. There's an infinite hierarchy of description that one may apply to concepts, and, in my philosophy, at some level the description becomes unable to be expressed.

Re:'Equals' Has More Than One Meaning in Math by lmm on Wednesday June 19, 2024 @06:29PM (#40733088)
> One can do the thinking without specifying what.

After one has figured out a rigorous approach, one can generally ignore it in day-to-day work. But you need to have that foundation there or you can go horribly wrong. (Analogy: we don't generally think about limits when doing day-to-day calculus. But when people tried to do calculus pre-Weierstrass, unless they were geniuses they "proved" all sorts of nonsense and got crazy results).

> I believe this is the notion of a 'universal property' in category theory.

That's one of the good answers to this problem. But you still need the constructions (if only to prove that an object with the required properties exists at all) and you need a lot of the kind of work the article is talking about, where you establish that these different constructions are equivalent and you can lift all your theorems along those equivalences.

> I'm saying that zero is the concept of taking two things (one an initial thing and the other a method to take one thing and get the next thing) and then returning the initial thing. While three is the concept of taking the same two things and then doing the method three times on the first thing. I used lambda notation because that is an easy way to convey the idea, not because that is what three is.

OK but then what kind of things are these "concepts"? What kind of questions can and can't we ask about them? If you care about "what the natural numbers are" then these kind of questions are worth asking.

Re:'Equals' Has More Than One Meaning in Math by simpaticoder on Wednesday June 19, 2024 @01:36AM (#40725167)
We must remember that these constructions retcon the actual human experience with numbers, which is firmly rooted in counting and measurement, that always come after this experiential basis. Formal definitions appear driven by an aesthetic consensus, mostly about minimalism and axiom flavor. Individual human experience with numbers reflects human history and at both scales 'counting and measurement' is more fundamental in the sense that it always comes first, and therefore all constructions derive from it, must maintain consistency with it, and conflicts must be resolved within it, and it is safe to ignore inconsistencies that can't be expressed within it.
Re:'Equals' Has More Than One Meaning in Math by staunton on Wednesday June 19, 2024 @03:03AM (#40725617)
In those terms, the foundations (of any topic or field) are never fundamental. I think people in this thread are using a different meaning of "fundamental".
Re:'Equals' Has More Than One Meaning in Math by lupire on Wednesday June 19, 2024 @07:51AM (#40727369)
Without formal constructions, humans have a nasty habit of proving things that wrong or not even wrong. Our ancestors dealt with this problem in the 19th Century by creating tools to help distinguish truth from balderdash.
Re:'Equals' Has More Than One Meaning in Math by simpaticoder on Wednesday June 19, 2024 @08:41AM (#40727718)
Your statement is consistent with my point which I mean to point out the psycho-social origin of such constructions and therefore the limits of their utility. I do not say they are not useful at all.