As I was graphing functions in Desmos graphing calculator, I typed in the function $$\lceil{x-\lfloor{x}\rfloor}\rceil$$

which, after some reasoning, unsurprisingly generates the values $0$ or $1$. My question is, can you - with any given amount of floor and ceiling functions - get a the values $0,1$ and $2$? In general, can you prove that with any amount of floor and ceiling functions you can obtain only whole numbers from $0$ to $n$.


Modular/remainder functions aren't allowed. I'm still in high school so I would appreciate an informal way of either proving the existence of such a function or actually showing a function that can cycle from $0$ to $n$.

  • 9
    $\begingroup$ "Modular/remainder functions aren't allowed." — $\ $ $x-\lfloor x\rfloor$ is exactly a remainder function: It's the remainder for division by $1$. $\endgroup$ – celtschk Aug 30 '16 at 7:38

The function $f(x)$ that cycles through $0,1,...,n-1$ is $f(x) = x \bmod n$. So the question is basically how to write it without using $\operatorname{mod}$.

The largest multiple of $n$ lower than or equal to $x$ is $n \left\lfloor \frac{x}{n} \right\rfloor$ and $\operatorname{mod}$ is the difference between $x$ and this largest multiple of $n$. So, in the end:

$$ f(x) = x - n \left\lfloor \frac{x}{n} \right\rfloor $$

is equivalent to the $\operatorname{mod}$ function and will loop through $0,1,...,n-1$ for sequential $x \in \mathbb{N}$.

[ EDIT ] Also, for an example of a function that loops through the same values, but takes only integer values between them:

$$ \lfloor f(x) \rfloor = \left\lfloor{x}\right\rfloor - n \left\lfloor\frac{x}{n}\right\rfloor $$

  • 1
    $\begingroup$ You can write a congruency with parentheses using the LaTeX \pmod operator: f(x) \equiv x \pmod n renders as: $f(x) \equiv x \pmod n$. For a single $mod$ to appear in upright font declare it explicitly as an operator name: \operatorname{mod} → $\operatorname{mod}$. $\endgroup$ – CiaPan Aug 30 '16 at 15:59
  • $\begingroup$ @CiaPan Thanks for the tips. Edited to fix the typesetting. $\endgroup$ – dxiv Aug 30 '16 at 16:27
  • $\begingroup$ @CiaPan: is it correct to write $f(x) = x \pmod n$ ? With $n>0$, I use $a\equiv b\pmod n$ to mean that $n$ divides $b-a$; and $a=b\bmod n$ to additionally mean that $0\le a<n$. $\endgroup$ – fgrieu Aug 30 '16 at 16:39
  • 1
    $\begingroup$ @dxiv: my problem is with $x\pmod n$; unless I err, my math teacher (France, 1978-1982) thought that the remainder of the division of $x$ by $n$ (as meant here) is noted $x\bmod n$, not $x\pmod n$; that $y\equiv x\pmod n$ means $y-x$ is divisible by $n$; and that $x\pmod n$, if used at all, would be the infinite set of $y$ with $y\equiv x\pmod n$. $\endgroup$ – fgrieu Aug 30 '16 at 19:25
  • 1
    $\begingroup$ @fgrieu is correct here. It is a common mistake to use "$x \pmod{n}$" instead of the intended "$x \bmod n$". The accepted answer to the very question you linked uses the right notation. Don't assume that the asker was using the correct notation. $\endgroup$ – user21820 Aug 31 '16 at 2:38

Sure. $x-\lfloor x \rfloor$ is the fractional part of $x$, so it can take any value in the range $[0,1)$.

So $2(x-\lfloor x \rfloor)$ can take any value in the range $[0,2)$, which means $\lceil 2(x-\lfloor x \rfloor) \rceil$ can be any of $0$, $1$, or $2$.

Similarly $n(x-\lfloor x \rfloor)$ can take any value in the range $[0,n)$, which means $\lceil n(x-\lfloor x \rfloor) \rceil$ can take on any integer value between $0$ and $n$ inclusive.

  • $\begingroup$ This doesn't necessarily give $x \in \mathbb{N}$ which (is not entirely clear in the question, but) seems to be required. $\endgroup$ – dxiv Aug 30 '16 at 3:25
  • $\begingroup$ @dxiv When does this occur? $\endgroup$ – Ian Limarta Aug 30 '16 at 3:29
  • $\begingroup$ @IanLimarta For example $\lceil 2(x - \lfloor x \rfloor) \rceil = 0$ for all $x \in \mathbb{N}$. Please edit the original question and clarify what you mean by "achieve the set of values $\{x\space|\space\space{x\in{\mathbb{Z_{\geq0}\}}}}$". $\endgroup$ – dxiv Aug 30 '16 at 3:34
  • $\begingroup$ @dxiv Edited. I'm sorry but I meant whole numbers. I should have just put that instead. $\endgroup$ – Ian Limarta Aug 30 '16 at 3:45

$\def\floor#1{\lfloor#1\rfloor}$Take any positive integer $n$ and real $x$.

$\floor{x-n\floor{\frac{x}{n}}} = \floor{x} \bmod n$. [This cycles through $0$ to $n-1$ with one unit for each value.]

$\floor{n(x-\floor{x})} = \floor{nx} \bmod n$. [This cycles through $0$ to $n-1$ with period length $1$.]

The easiest way to see why is to notice that $n \floor{\frac{x}{n}}$ is the nearest multiple of $n$ no greater than $x$, so subtracting that from $x$ yields the remainder of $x$ modulo $n$. If you then perform a further floor, you will get the integer part of the remainder, as desired. You can first scale $x$ by a constant before the whole process, to change the period length. This is how we can get the second expression above.


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.