2

I'm pretty sure the way to go is with a contradiction, letting $x = \sup E$. Then I think I need to either show that $x < x^2\in E$ (contradicting $x = \sup E$ in that an element of E is greater than x) or show that $({x^2})^2 > 2$ (contradicting the terms of E). As I wrote this I realized showing $({x^2})^2 > 2$ would be redundant since that would only mean something if $x^2\in E$ in the first place... anyway, I'm not sure how to go about this proof.

JohanLiebert
  • 1,207
  • 6
    See http://math.stackexchange.com/questions/14970/no-maximumminimum-of-rationals-whose-square-is-lessergreater-than-2, http://math.stackexchange.com/questions/91106/choice-of-xi?lq=1, http://math.stackexchange.com/questions/21076/constructing-a-larger-rational-whose-square-is-also-less-than-two?lq=1, http://math.stackexchange.com/questions/55805/set-of-rationals-less-than-sqrt2-has-no-supremum-in-mathbbq?lq=1, http://math.stackexchange.com/questions/194133/prove-that-the-set-of-positive-rational-values-that-are-less-than-sqrt2-ha?lq=1, http://math.stackexchange.com/questions/426714/ – Jonas Meyer Aug 27 '13 at 04:51
  • 1
    The question this is marked as a duplicate of is specifically asking about a method in Rudin's Principles of mathematical analysis used to prove this result. – Jonas Meyer Aug 27 '13 at 07:53

2 Answers2

4

Suppose that $x$ and $y$ are rational numbers such that $x^2<y$. For each rational $h$ with $0<h<1$, $$(x+h)^2=x^2+2xh+h^2<x^2+h(2|x|+1).$$

If $h$ is chosen such that $h\leq\dfrac{y-x^2}{2|x|+1}$ (e.g., $h$ could be taken to be equal to that quantity or $\frac12$, whichever is smaller), then $(x+h)^2<y$, and therefore $x+h$ is a rational number larger than $x$ whose square is still less than $y$.

Jonas Meyer
  • 55,715
2

For every $p \in E$, consider $$q = p - \frac{p^2 - 2}{p + 2} = \frac{2p + 2}{p + 2}$$ Note that, $q > p$. Now, $$q^2 - 2 = \frac{2(p^2 - 2)}{(p+2)^2}$$ Therefore, $q^2 < 2$. This implies $q \in E$.

Therefore, for every $p \in E$, we can find a $q > p$ such that $q \in E$. This implies that E has no largest element.

(This is a proof basically from Rudin)

TenaliRaman
  • 3,916