1
$\begingroup$

The goal is to prove that $\lceil x+m\rceil=\lceil x\rceil +m$, where $x$ is a real number and $m$ is an integer. The book outlines the following proof:

Write $x=n-\epsilon$, where $n$ is an integer and $0\leq\epsilon<1$; thus, $\lceil x\rceil=n$. Then $\lceil x+m\rceil=\lceil n-\epsilon+m\rceil=n+m=\lceil x\rceil+m$.

If we "read between the lines," it really seems like the following is being communicated:

\begin{align} \lceil x+m\rceil&=\lceil n-\epsilon+m\rceil\\ &= \lceil-\epsilon\rceil+(n+m)\tag{circular reasoning?}\\ &= 0+(n+m)\tag{since $\epsilon\in[0,1)$}\\ &= \lceil x\rceil+m \end{align}

Is the book's proof fine and I'm just not seeing something clearly or is there a subtle error somewhere (if so, what could be done to fix the proof?)?

$\endgroup$
  • $\begingroup$ how do you define $\lceil x \rceil$ if it's not with $\lceil n - \epsilon \rceil = n$ for any integer $n$ and $\epsilon \in [0;1[$ ??? $\endgroup$ – reuns Jan 11 '16 at 4:29
  • $\begingroup$ It's not quite circular since we know that $n+m$ is an integer. Since $\epsilon < 1$, we can evaluate $\lceil n + m - \epsilon \rceil$. $\endgroup$ – amcerbu Jan 11 '16 at 4:33
  • $\begingroup$ I don't think it's circlular. It is presumed that if $0 \le e < 1$ then ceil(n - e) = n for integer n. (either this is the definition or a very basic proposition). Thus there is no circularity at all ceil(x + m) = ceil(n - e + m) = ceil((n+m) - e) = n + m = ceil(n) + m. (because (n+m) is an integer so ceil((n+m) -e) = n + m). $\endgroup$ – fleablood Jan 11 '16 at 4:34
  • $\begingroup$ @user1952009 Yeah, I think that makes sense. The restriction on $\epsilon$ makes the argument non-circular; I guess I was just seeing the explication of the argument in a somewhat circular fashion when that's really not the case $\endgroup$ – interrogative Jan 11 '16 at 4:37
0
$\begingroup$

As is pointed out in the comments, there is no circular reasoning here. To clear things up substitute $u=n+m$ (which will be an integer) and then use the theorem you already used and accepted, that $\lceil n-\epsilon \rceil = n$ where $n$ is any integer (and replace $n$ by $u$)

| cite | improve this answer | |
$\endgroup$

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.