-1

I was reading through the proof of this and I'm having trouble grasping what the contradiction is. Since $a\leq b+\varepsilon$ this is equivalent to $a-\varepsilon\leq b$. Suppose $b<a$ and taking $\varepsilon=\frac{1}{2}(a-b)$ we have $a-\varepsilon=a-\frac{1}{2}(a-b)=\frac{1}{2}a+\frac{1}{2}b>b$. At this point they state that this is a contradiction to the hypothesis but I don't see how it contradicts that. The contradiction or error I see is that then we have $b<a-\varepsilon\leq b\Rightarrow b<b$ which makes no sense. Although this doesn't appear to be the hypothesis. Also I was curious as to the intuition behind setting $\varepsilon=\frac{1}{2}(a-b)$.

Craig
  • 672

1 Answers1

2

You assume that $a-\varepsilon\leq b$ and derive $a-\varepsilon>b$.

The intuition is you want to to show that $a\not \leq b$. This means $a-b>0$. The $\frac12$ is just there for technical reasons. If you try with just $a-b$, it doesn't work quite as nicely.