...and a few definitions on the way?

When I studied Calculus using Spivak's book It was clearly shown that, in order to prove some fundamental theorems (intermediate value theorem being one of them), one had to assume an additional property of numbers (namely the least upper bound property). Other postulated properties included the field axioms and 3 more axioms (needed for the inequalities).

But then I skimmed over "Grundlagen der Analysis" by Landau (which was in the recommended reading by the way) and he developed the reals using Dedekind cuts assuming only Peano's axioms - the way I saw it. I also read on Wikipedia that the set of Dedekind cuts has the least upper bound property. So it seems possible that we can get from Peano's axioms all the way to theorems in analysis (or am I missing some additional assumptions?)

If it is true that Peano's axioms are all we need, then I want to know how far we can take them...in general (not just real analysis). Also is there an elegant way to prove the field axioms for real numbers from Peano's axioms?

  • $\begingroup$ How many definitions is 'a few'? $\endgroup$
    – Khallil
    Aug 3, 2015 at 23:46
  • $\begingroup$ you know...addition, multiplication...cuts...limits...derivatives...integrals $\endgroup$
    – user258704
    Aug 3, 2015 at 23:48
  • $\begingroup$ So you're assuming integration, something that you'd need to rigorously develop? :-/ $\endgroup$
    – Khallil
    Aug 3, 2015 at 23:49
  • $\begingroup$ No - I'm assuming I would develop everything I need to state it's definition $\endgroup$
    – user258704
    Aug 3, 2015 at 23:51
  • 2
    $\begingroup$ There is a scary and fun area of foundations called "reverse mathematics" that looks into how little or how much you need to be able to develop parts of analysis. $\endgroup$ Aug 3, 2015 at 23:52

2 Answers 2


It isn't true that "Peano's axioms are all we need". Landau's beautiful account involves some set theory (the Dedekind cuts are sets or pairs of sets of rational numbers). There are many elegant ways of getting to the real numbers from the integers, but they all need a bit more than Peano's postulates. One of my favourites is this paper by Behrend, which shows how the additive structure can be defined first and the multiplicative structure derived from it.

  • $\begingroup$ thanks - I'll look more into both behrend's paper and Landau's book (i will try to keep record of the zfc axioms he uses). I also noticed he used ordered pairs (i think, he didn't mention explicitly) to define ratios...so yeah should have seen that coming ;) $\endgroup$
    – user258704
    Aug 4, 2015 at 0:09
  • $\begingroup$ The mathematical content of this paper was paid for by the German taxpayer, and thousands of libraries have paid for the printing costs; but sixty years later we are asked $ 40.00 for the opportunity to read a copy. $\endgroup$ Aug 23, 2015 at 19:32
  • $\begingroup$ @ChristianBlatter: I understand your annoyance about paying Springer Verlag, but as regards the German taxpayer please read www-history.mcs.st-and.ac.uk/Extras/Behrend_LMS_obituary.html? Behrend was a second world war refugee. $\endgroup$
    – Rob Arthan
    Aug 23, 2015 at 20:59
  • $\begingroup$ @RobArthan: Thank you for letting me read this obituary. $\endgroup$ Aug 24, 2015 at 7:12
  • $\begingroup$ LOL. The German taxpayers are busy bailing out Greece and need all the help they can get. $\endgroup$
    – user4894
    Aug 24, 2015 at 18:41

For what it is worth, I was able to construct the set of positive real numbers as Dedekind cuts starting from Peano's axioms and using the rules of first-order logic and some set theory (my own formulation based very loosely on ZFC).

I began with Peano's axioms as follows:

  1. $1\in N$
  2. $\forall x\in N:[ S(x)\in N]$
  3. $\forall x,y\in N:[S(x)=S(y)\implies x=y] $
  4. $\forall x\in N:[S(x)\ne 1]$
  5. $\forall P\subset N: [1\in P \land \forall x\in P:[ S(x)\in P] \implies \forall x\in N:[x\in P]]$

I then constructed the usual addition ($+$) and multiplication ($\times$) functions on $N$, and constructed the set $Q^+$ of positive rational numbers such that:

$\forall x:[x\in Q^+ \iff x\subset N^2 \land \exists a,b\in N: \forall c,d:[(c,d)\in x \iff a\times d = b\times c ]]$

Note: By constructed, I do not simply mean defined. I mean that I actually prove the existence of these constructs using the axioms of set theory and the rules of logic.

I was eventually (after thousands of lines of formal proof) able to construct the usual addition ($+$) and multiplication ($\times$) functions, and the usual ordering ($<$) on $Q^+$, and the set $R^+$ of positive real numbers such that:

$\forall x:[x\in R^+\iff x\subset Q^+$

$ \land \exists y:[y\in x]$

$ \land \exists y \in Q^+:[ y\notin x]$

$\land \forall y\in x: \forall z\in Q^+:[z<y\implies z\in x]$

$ \land \forall y\in x:\exists z\in x: [y<z]] $

From set theory, I had to be able to construct subsets, power sets, Cartesian products (ordered pairs and triples only) and functions of one or two variables. I also made use of set equality (extensionality). I made no use of any other aspects of set theory.


Here are links to to key proofs for constructing the positive real numbers:

  • $\begingroup$ Very impressive. Could you provide a link to your work. $\endgroup$ Aug 13, 2015 at 16:18
  • $\begingroup$ @martycohen As promised, see Addendum. I left out a lot, e.g. proving associativity, commutativity, etc. but these proofs show how I was able to construct the positive real numbers starting from Peano's Axioms. The addition and multiplication proofs are tedious monsters, but you can get the gist of them from the introductions. $\endgroup$ Aug 24, 2015 at 18:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .