Can someone help me to solve the following congruent equation. I have tried to use Fermat's little theorem but failed to solve this: $x^{12}=87(\mod 101)$
-
$\begingroup$ What's wrong with brute force? It only takes a few seconds with a machine (including the programming time. The actual computation is effectively instantaneous). $\endgroup$– luluMar 3, 2019 at 15:13
-
$\begingroup$ General note: these problems tend to be really hard, computationally. Brute force is an excellent approach, for numbers as small as these. You could, of course, find a primitive root and work from there, but that's a lot more computation. $\endgroup$– luluMar 3, 2019 at 15:14
-
1$\begingroup$ @lulu "Only a few seconds including the time for programming" - Somewhat exaggerated, but I agree to both comments. $\endgroup$– PeterMar 3, 2019 at 15:57
-
$\begingroup$ @Peter "Solve x^12=87 mod 101" is all it take in WA $\endgroup$– luluMar 3, 2019 at 16:06
-
1$\begingroup$ @Peter there's Tonelli-Shanks for primes. Not even close to worth it here. For composite moduli the problem is equivalent to factorization. $\endgroup$– luluMar 3, 2019 at 16:25
3 Answers
Let us try to use Fermat's little theorem
We know $$x^{108}=(x^{12})^9\equiv 87^9\equiv 36\mod 101$$ implying $$x^8\equiv 36\mod 101$$
This gives us $$x^{104}=(x^8)^{13}\equiv 36^{13}\equiv 95\mod 101$$
implying $$x^4\equiv 95\mod 101$$ Not sure whether we can simplify even further.
The integer square roots of $95$ modulo $101$ are $14$ and $87$, the square roots of $14$ modulo $101$ are $32$ and $69$ (two solutions) and the square roots of $87$ modulo $101$ are $17$ and $84$ (the two other solutions).
Algorithmically let's use Shanks' baby giant step. The Remark below shows $2$ is a primitive root $({\rm ord}\,2 = 100)$ so $\,87 \equiv 2^{\large n}\,$ is solvable, e.g. by baby-giant step, i.e. we solve $\,87\cdot 2^{\large 10j}\equiv 2^{\large k}\,$ for $\,0\le j,k < 10 = \lceil\sqrt{100}\rceil,\,$ by repeatedly mutliplying $\,87\,$ by $\,2^{\large 10}\! \equiv 1010\!+\!14\equiv 14\,$ until we reach some $\,2^{\large k}.\,$ It requires $\,\color{#c00}4\,$ multiplications, i.e.
$$87\equiv-14 \overset{\large \times\color{#c00}{14}}\to 6\overset{\large\times\color{#c00}{14}}\to -17\overset{\large\times\color{#c00}{14}}\to -36\overset{\large\times\color{#c00}{14}}\to 1 $$
so $\ \smash[t]{87(\overbrace{2^{\large 10}}^{\large \color{#c00}{14}})^{\Large\color{#c00}4}}\equiv 1\overset{\large \times\, 2^{\LARGE 60}\!}\Longrightarrow 87\equiv 2^{\large 60}\!\equiv 2^{\large 12x}\!\!\!\iff$ $\! 12x\equiv 60\pmod{\!100}\!\iff\! n\equiv 5\pmod{\!25}$
Hence the solutions are $\,x\equiv 2^{\large 5}\!\equiv 32,\,\ x\equiv 2^{\large 30}\!\equiv 14^{\large 3}\!\equiv 17\,$ and their negatives $(\times\, 2^{\large 50}\!\equiv -1)$
Remark $ $ We skipped an optimization in order to better show the general method. Namely, at the start we already have $\,87\equiv -14\equiv -2^{\large 10}\equiv 2^{\large 60}\,$ by $\,-1\equiv 2^{\large 50},\,$ by $\,{\rm ord}\,2 = 100.\,$
To prove $\,{\rm ord}\,2 = 100,\,$ by the Order Test it suffices to show that $\,2^{\large 100/p}\!\not\equiv 1$ for all primes $\,p\mid 100,\,$ i.e. $\,2^{\large 20}\!\not\equiv 1,\, $ $2^{\large 50}\!\not\equiv 1.\,$ That's easy: $\,2^{\large 20}\!\equiv 14^{\large 2}\!\equiv 196\equiv -6,\,$ so $\,2^{\large 50}\!\equiv 14(-6)^{\large 2}\!\equiv 17(-6)\equiv -1$
Numbers congruent to $87 \pmod {101}$ are $87,\; 87+101=188,\; 87+2\times 101=\color{red}{289}, ...$
and $87-101=-14,\;87-2\times 101=-115,\;87-3\times 101=\color{blue}{-216},...$
Note that therefore $17^2=289 \equiv87\pmod{101}$ and $(-6)^3=-216 \equiv 87 \pmod{101}.$
Furthermore $17\times6=102 \equiv 1 \pmod{101}.$
Therefore $17^5\equiv17^2/6^3\equiv -1 \pmod{101}.$
Thus $17^{12}=17^{10}\times17^2=(17^5)^2\times17^2\equiv87 \pmod {101},$
so $x=17$ is a solution to $x^{12}\equiv87\pmod{101}$.
-
$\begingroup$ I realize I found a solution but not all solutions. $\endgroup$ Mar 3, 2019 at 16:15
-
$\begingroup$ The solutions are $\pm17,\,\pm 32$, easily computable by baby-giant step - see my answer. $\endgroup$ Mar 3, 2019 at 18:41