I have read that Dedekind cuts allow you to define the real numbers from the rationals. For example, $\sqrt{2}$ can be defined in the following way:

  • Partition the rational numbers into two sets $A$ and $B$, such that all of the elements of $A$ are less than the elements of $B$
  • $A=\{a \in \mathbb{Q}:x<0 \text{ or } x^2 < 2$}
  • $B=\{b \in \mathbb{Q}: x > 0 \text{ and } x^2 \geq 2$}
  • Because $B$ has no lower bound, there is a 'gap' in the number line
  • We define $\sqrt{2}$ to fill this gap: $\sqrt{2}$ is the unique number such that $x^2=2, x>0$. $\sqrt{2}$ sits on the partition line that we previously drew (hence why we say $\sqrt{2}$ is that cut)

Hopefully, I understand Dedekind cuts well enough to ask this question. It seems to me that because $\pi$ is transcendental, it cannot fill a gap in the same way that $\sqrt{2}$ does. There is no polynomial equation that $\pi$ helps solve. By contrast, $\sqrt{2}$ solves the equation $x^2=2, x \geq 0$. Nevertheless, $\pi$ is a real number, and Dedekind cuts define the real numbers. So, can $\pi$ be defined using Dedekind cuts, or is more work needed? Moreover, does the usual geometric definiton of $\pi$ have a well-defined mathematical meaning if you have not constructed the real numbers from the rationals?


We know that $$\pi=4\int_0^1\frac{dx}{1+x^2}.$$ A lower Riemann sum of this integral is $$L_n=\frac4n\sum_{k=1}^{n}\frac1{1+k^2/n^2}.$$ An upper Riemann sum of this integral is $$U_n=\frac4n\sum_{k=0}^{n-1}\frac1{1+k^2/n^2}.$$ We can define $A=\{a\in\Bbb Q:a<L_n\text{ for some }n\in\Bbb N\}$ and $B=\{b\in\Bbb Q:b>U_n\text{ for some }n\in\Bbb N\}$.

| cite | improve this answer | |

Any sequence $c_n$ that converges to $\pi$ can be used to define such a cut, with

$$ A=\{a\in\mathbb Q\mid a\lt c_n\text{ infinitely often}\}\hphantom{\;.} $$


$$ B=\{b\in\mathbb Q\mid b\gt c_n\text{ infinitely often}\}\;. $$

| cite | improve this answer | |
  • $\begingroup$ Thank you for responding. What does 'infinitely often' mean in this context? $\endgroup$ – Joe May 4 at 8:28
  • $\begingroup$ @Joe: That there are infinitely many $n\in\mathbb N$ for which $a\lt c_n$ (or $b\gt c_n$). $\endgroup$ – joriki May 4 at 8:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.