Does there exist a surjection from the natural numbers to the reals?

  • #1
docnet
799
487
Homework Statement
1) What does it mean if two sets ##X## and ##Y## have the same cardinality?

2) Given sets ##A## and ##B## what does ##|A|\leq |B|## mean?

3) Does there exist a surjection from the natural numbers to the reals? If so, define one. If not, explain why one cannot exist.
Relevant Equations
##\mathbb{N}## and ## \mathbb{R}##
1)
Two sets have the same cardinality if there exists a bijection (one to one correspondence) from ##X## to ##Y##. Bijections are both injective and surjective. Such sets are said to be equipotent, or equinumerous. (credit to wiki)

2)
##|A|\leq |B|## means that there is an injective function from ##A## to ##B##. An injective function maps distinct elements of ##A## to distinct elements in ##B##.

3)
No.

Cantor's proof that there is no surjection ##f:\mathbb{N}\longrightarrow \mathbb{R}## seems relevant here.

I'm wondering if there is a lazier way to answer this question, although I think the following logic is flawed.

In class, we are given as a theorem that there is no bijection ##f:\mathbb{N}\longrightarrow \mathbb{R}##. Bijective functions are both injective and surjective, so there is no such ##f:\mathbb{N}\longrightarrow \mathbb{R}## that is both injective and surjective.

An example of an injection ##f:\mathbb{N}\longrightarrow \mathbb{R}##:

Let ##f:\mathbb{N}\longrightarrow \mathbb{R}## be a function defined by the rule ##f(n)=n##. ##f(n)\neq f(m)## whenever ##n\neq m##, so ##f## is injective. Since an injection ##f:\mathbb{N}\longrightarrow \mathbb{R}## was defined, it exists.

The issue is that this logic cannot deal with the possibility of an ##f## being surjective and not injective.
 
Last edited:
Physics news on Phys.org
  • #2
You are right, including your assessment of the last argument. It does not help that there is an injection, or that there is no bijection. You must take a possible surjection and deduce a contradiction. Just an idea: Assume ##f\, : \,\mathbb{N}\longrightarrow \mathbb{R}## is surjective. Then ##\emptyset \neq f^{-1}(r)\subseteq \mathbb{N}## for every ##r\in \mathbb{R}.## Hence we have a well-defined smallest element ##\mathbb{N}\ni N_r\in f^{-1}(r)## for all ##r\in \mathbb{R}.## This defines a function ##r\longmapsto N_r\,.##
What can you say about it?
 
  • Like
  • Informative
Likes docnet, sysprog and PeroK
  • #3
fresh_42 said:
You must take a possible surjection and deduce a contradiction. Just an idea: Assume ##f\, : \,\mathbb{N}\longrightarrow \mathbb{R}## is surjective. Then ##\emptyset \neq f^{-1}(r)\subseteq \mathbb{N}## for every ##r\in \mathbb{R}.## Hence we have a well-defined smallest element ##\mathbb{N}\ni N_r\in f^{-1}(r)## for all ##r\in \mathbb{R}.## This defines a function ##r\longmapsto N_r\,.##
What can you say about it?
##f## is surjective, so it is well-defined. So ##f(n)\neq f(m)\Longrightarrow n\neq m##. This leads to ##f^{-1}(f(m))\neq f^{-1}(f(n))## whenever ##m\neq n##. So the ##f^{-1}(r)## are disjoint in ##\mathbb{N}## for all ##r##. ##N_r\in f^{-1}(r)##. Hence the ##N_r## are disjoint for all ##r##. Hence ##r\longmapsto N_r## is injective.

(Is this okay?)

This means the image of ##r\longmapsto N_r## (we could call the set ##\mathbb{N}_r## for brevity) has a cardinality equal to ##|\mathbb{R}|=𝔠 ##. ##N_r\in \mathbb{N}## for all ##r##, so we have ##\mathbb{N}_r\subset\mathbb{N}##. So it must also be that ##|\mathbb{N}|\geq 𝔠##.

(Is this still okay?)

There can be no bijections from ##\mathbb{N}## to ##\mathbb{R}##. This means ##|\mathbb{N}|\neq 𝔠##. So we have ##|\mathbb{N}|> 𝔠##.

I am stumped trying to derive a contradiction. @fresh_42 could you please drop a little hint?
 
Last edited:
  • #4
docnet said:
##f## is surjective, so it is well-defined.
These are two unrelated properties. A function is well-defined if it doesn't map one element on two targets. A function is surjective if every possible target is actually hit at least once. You can only construct an implicit dependency. By saying a function is surjective, it is already assumed that it is a function, i.e. especially well-defined. If you meant this, then you are right. But it would be better to state that ##f## is well-defined by assumption, rather than by surjectivity.

docnet said:
So ##f(n)\neq f(m)\Longrightarrow n\neq m##. This leads to ##f^{-1}(f(m))\neq f^{-1}(f(n))## whenever ##m\neq n##. So the ##f^{-1}(r)## are disjoint in ##\mathbb{N}## for all ##r##. ##N_r\in f^{-1}(r)##. Hence the ##N_r## are disjoint for all ##r##. Hence ##r\longmapsto N_r## is injective.

(Is this okay?)
Think so.
docnet said:
This means the image of ##r\longmapsto N_r## (we could call the set ##\mathbb{N}_r## for brevity) has a cardinality equal to ##|\mathbb{R}|=𝔠 ##. ##N_r\in \mathbb{N}## for all ##r##, so we have ##\mathbb{N}_r\subset\mathbb{N}##. So it must also be that ##|\mathbb{N}|\geq 𝔠##.

(Is this still okay?)
Yes. You could also simply observe that ##f(\varphi (r))=r## where ##\varphi (r)=N_r## is the function we get from ##f^{-1}## by picking the least number. Now ##\mathbb{R}\stackrel{\varphi }{\longrightarrow }\mathbb{N}_r\stackrel{f}{\longrightarrow }\mathbb{R}## is the identity map, and since ##f## is surjective, we have ##\mathbb{R}=\mathbb{N}_r## and there is no such bijection.
docnet said:
There can be no bijections from ##\mathbb{N}## to ##\mathbb{R}##. This means ##|\mathbb{N}|\neq 𝔠##. So we have ##|\mathbb{N}|> 𝔠##.

I am stumped trying to derive a contradiction. @fresh_42 could you please drop a little hint?
Well, you could simply say that we already know that ##|\mathbb{N}|<|\mathbb{R}|## by Cantor's diagonal enumeration argument.
 
  • Informative
Likes docnet
  • #5
I suggest it's better to really understand the strategy of the proof before you get into the complications of writing it down formally:

1) We know that there is no bijection from ##\mathbb N## to ##\mathbb R##.

2) We also know that there is no bijection from a subset of ##\mathbb N## to ##\mathbb R##.

3) Suppose ##f## is a surjection from ##\mathbb N## onto ##\mathbb R##.

3a) The idea is to use ##f## to construct a bijection from a subset of ##\mathbb N## to ##\mathbb R##, which proves that no such surjection exists.

3b) Every real number ##r## is in the range of ##f##. Hence, there exists a non-empty set of natural numbers ##S_r## than are all mapped to ##r##. Note that ##S_r## may be a single element, finite or infinite subset of ##\mathbb N##. Note that the ##S_r## are disjoint, as ##f## is a (well-defined) function.

3c) Each ##S_r## has a smallest element ##n_r##.

3d) The set ##S = \{n_r: r \in \mathbb R \}## is a subset of (or equal to) ##\mathbb N##.

3e) Technical point: ##S## cannot be finite(!)

3f) Define ##\tilde f : S \rightarrow \mathbb R##, by ##\tilde f(n_r) = r## (or, if you prefer, ## \tilde f(n) = f(n)##.

3g) ##\tilde f## is a bijection from ##S## to ##\mathbb R##. Which is the required contradiction.
 
  • Informative
Likes docnet
Back
Top