![]() | ![]() |
Suppose r is a primitive root of n, and a is a positive integer relatively prime to n.
We have already seen that all of the powers r1,r2,r3,…,rφ(n) produce distinct least positive residues \pmod{n}, and are all relatively prime to n.
Thus, if 1 \le x \le \varphi(n), then r^x \equiv a \pmod{n} must have a unique solution.
Let us call this unique solution the the index, base r of a, modulo n, and denote it by
x = \textrm{ind}_r aNotice this definition has a certain similarity to that of the logarithms one normally encounters in pre-calculus courses -- namely, that the real-value x which solves r^x = a is given by x = \log_{r} a.
Of course, when one thinks of logarithms, one immediately remembers the rules for how they can be manipulated -- rules like the following:
One might wonder if an index base r behaves similarly. Our curiosity is well-founded -- we can indeed prove an index-based analog to each of the items shown above:
\text{ind}_r (xy) \equiv \text{ind}_r x + \text{ind}_r y \pmod{\varphi(n)}
Proof:
Note that r^{\text{ind}_r (xy)} \equiv xy \equiv r^{\text{ind}_r x} \cdot r^{\text{ind}_r y} \equiv r^{\text{ind}_r x + \text{ind}_r y} \pmod{n} Recall, however, that for primitive roots r, having r^i \equiv r^j \pmod{n}, requires i \equiv j \pmod{\varphi(n)}, so it must be the case that \text{ind}_r (xy) \equiv \text{ind}_r x + \text{ind}_r y \pmod{\varphi(n)}
QED.
\text{ind}_r b^k \equiv k \cdot \text{ind}_r b \pmod{\varphi(n)} assuming k is a positive integer
Proof:
Note, directly from the definition of index, we can argue that r^{k \cdot \text{ind}_r b} \equiv (r^{\text{ind}_r b})^k \equiv b^k \equiv r^{\text{ind}_r b^k} Which again, recalling that for primitive roots r, having r^i \equiv r^j \pmod{n}, requires i \equiv j \pmod{\varphi(n)} yields \text{ind}_r b^k \equiv k \cdot \text{ind}_r b \pmod{\varphi(n)}
QED.
\textrm{ind}_r 1 \equiv 0 \pmod{\varphi(n)}
Proof:
To see this, note that if x = \textrm{ind}_r 1, then r^x \equiv 1 \pmod{n}, with 1 \le x \le \varphi(n).
However, r is a primitive root, so x = \varphi(n). Consequently, x \equiv 0 \pmod{\varphi(n)}
QED.