0

Why is the number of solutions in a linear congruence $ax\equiv b(mod m)$ equal to the $gcd(a,m)$ such that $gcd(a,m)$ divides $b$?

TheLast Cipher
  • 1,718
  • 19
  • 28

1 Answers1

1

Let $\ d = \gcd(A,M)\,$ and $\, a,b,m = [A,B,M]/d = A/d,\,B/d,\,M/d.\ $ Then

$$\begin{align} A x \equiv&\ B\pmod{\!M}\\ \iff\ Ax\ -&\ B\ =\ j M,\quad\, {\rm for\ some}\ \ j\in\Bbb Z\\ \iff\ ax\,\ -&\ b\,\ =\,\ j\, m,\quad {\rm for\ some}\ \ j\in\Bbb Z,\, \text{ by scaling by $d$ or $d^{-1}$} \\ \iff\,\ ax\equiv&\ b\,\ \pmod{\!m}\\ \iff\,\ \ \ x\equiv&\: b/a\!\!\!\!\pmod{\!m}\quad\text{[note $a^{-1}$ exists by $\,\gcd(a,m)=1$]} \end{align}$$

The root $\,x\equiv b/a\pmod{\!m}\,$ lifts to $\,\color{#c00}d\,$ roots $\bmod \color{#c00}{d}m = M\,$ since

$$\begin{align} \bmod{dm}\!:\ \ b/a + jm\equiv&\ b/a + km\\ \iff\qquad\ \ (j-k)m\equiv&\ 0\\ \iff\ md\mid (j-k)m\quad &\\ \iff\ d\ \,\mid\,\ j-k\qquad\ \ \ &\\ \iff\ \ j\equiv k\pmod{\!d} \end{align}\qquad\qquad\qquad\qquad$$

Thus choosing normal reps $\,j=0,1,2,\ldots\,d\!-\!1\,$ yields precisely $\color{#c00}d$ roots $\bmod M,\,$ namely

$$ \{ b/a + j\, m\ : \ j = 0,1,2,\ldots,d\!-\!1\}\qquad\qquad\qquad\quad $$

Bill Dubuque
  • 265,059
  • 37
  • 279
  • 908
  • what is the rule that states what you did from $ax\equiv b(\mod m)$ to $x\equiv b/a (\mod m)$? I've seen it before but I forgot. thanks – TheLast Cipher Sep 20 '17 at 14:13
  • @The We multiplied both sides of the congruence by $\,a^{-1},\,$ which is valid by the [Congruence Product Rule.](https://math.stackexchange.com/a/879262/242) It helps to think of congruences as generalized equalities, since they enjoy many of the same properties of equations (e.g. the linked Congruence Laws). Using a symbol visually similar to an equal sign helps to emphasize that analogy. – Bill Dubuque Sep 20 '17 at 14:21
  • alright I shall study those first, thanks! can you teach me how to think about this more "fluidly"? I don't feel a strong grasp of this modular arithmetic although I do understand intuitively why $ax\equiv b(\mod m)$ means ax-my=b. I still don't feel comfortable with it. – TheLast Cipher Sep 20 '17 at 14:29