5
$\begingroup$

Is there a way to define modulo division (or functions of modular arithmetic in general) as superposition of (elementary?) functions?

For example, the multiplication is first introduced as summation, the exponentiation - as multiplication, etc. The trigonometric functions can be expressed through other elementary functions. Moreover, a wide range of "continuous" functions can be expressed as infinite series of different nature, whether we are talking about complete set of functions in $L_p$ space, or about derivatives of infinitely differentiable function combined in Taylor series.

I wonder if it is possible to define functions from modulo arithmetics as closed-form expressions of functions which belong to classes listed in the table below. I have no doubts that it is possible to do so via infinite series, although I do not know how. However, I am particularly interested in closed-form expressions from the table

enter image description here

I find this table from this wikipedia article extremely interesting, although it is devoted to the closed-form expressions only and missing modulo arithmetic.

$\endgroup$
4
$\begingroup$

I believe what you are looking for is this:

$$ mod(x,M) = [\frac{1}{2} + \frac{i}{2\pi}ln(-e^{-i2\pi x/M})]\times M $$

You can check it out graphed here:

https://www.wolframalpha.com/input/?i=plot+(1%2F2+%2B+(i%2F(2*pi))*log(-1*exp(-i*2*pi*x%2FM)))*M+where+M+%3D+10

I stumbled upon this post because I was playing around with the function $(-1)^x$ which has a real and imaginary part that are periodic and out of phase. Using Euler's formula I pulled out the part of the function that was cyclical and graphed it and saw that it was a saw wave. I generalized it by scaling it in the x and y direction by M.

Here is the Euler formula transform of the same function:

$$ mod(x,M) = [\pi + \frac{ln(cos(\frac{2\pi (x+M/2)}{M}+M+log(2))+i*sin(\frac{2\pi (x+M/2)}{M}+M+log(2)))}{i}]\times \frac{M}{2\pi} $$

You can also derive this by taking the closed form solution that drops out of the infinite Fourier series:

http://mathworld.wolfram.com/FourierSeriesSawtoothWave.html

Note that the Gibbs phenomenon that you get with a finite number of terms when approximating the saw wave disappears with the closed form solution.

$\endgroup$
  • $\begingroup$ This website uses MathJax for math formulas and symbols. $\endgroup$ – Silvia Ghinassi Jun 27 '16 at 18:18
  • $\begingroup$ I converted it to MathJax. $\endgroup$ – Nathan Janos Jun 28 '16 at 18:29
2
$\begingroup$

Though it isn't perfect, you can create a "saw-tooth" function which is pretty much a periodic linear function. For example, if you want to take $\mod 3$, you have the function $f$ which has period $3$ and looks like $f(x) = x$ for $x \in [0, 3)$.

You can approximate this saw-tooth function using a Fourier series which involves harmonics of the form $e^{2 \pi i n x}$.

$\endgroup$

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.