4

I have to solve the following equation: $x=2^{-18} \mod 143$.

The problem is that I can't use Fermat's little theorem as $\varphi(143)=120$ which doesn't help at all. The other method I know is to find the inverse of $2^{18} \mod 143$ using Euclid's extended algorithm but that would mean to find the inverse of $362144 \mod 143$ which doesn't seem like a good method to me...Any other ideas how I can solve this? Thank you!

Functional analysis
  • 185
KeykoYume
  • 492
  • Why do you think Euclid's method is not good? It is THE method for this problem. Alternatively, you can first find the remainders modulo 11 and 13 separately and then combine them. – Aravind Apr 18 '15 at 18:59

3 Answers3

4

You can work modulo $11$ and $13$ and stitch the results together using the Chinese Remainder Theorem. For example:

$$2^{18}\cdot 2^2=(2^{10})^2\equiv 1 \bmod 11$$ by Fermat, so that $2^{-18}\equiv 4 \bmod 11$ as a start.

For CRT note that $6\cdot 11-5\cdot 13=1$ so that $-65a+66b=c \equiv a \bmod 11, \equiv b \bmod 13 $, and the same is true of $c$ reduced modulo $143$.

Mark Bennet
  • 100,194
3

Here are two approaches. First by exponentiation and Gauss's algorithm, second by CRT.

${\rm mod}\ 143\!:\,\ \color{#c00}{2^{18}}\equiv (2^2\!\cdot\!\!\overbrace{ 2^7}^{\large -15}\!)^2\equiv 60^2\equiv 20\overbrace{(37)}^{\large 3(60)}\equiv 5\!\overbrace{\!(5)}^{\large 4(37)}\!\!\equiv \color{#c00}{25}$

Therefore $\,\ \color{#c00}{2^{-18}\equiv \dfrac{1}{25}}\equiv \dfrac{6}{150}\equiv \dfrac{6}7\equiv \dfrac{120}{140}\equiv \dfrac{120}{-3}\equiv -40\ $ by Gauss's Algorithm

Or, we can use little Fermat and CRT (Chinese Remainder) and, again, all mental arithmetic.

${\rm mod}\ 11\!:\,\ 2^{10}\equiv 1\,\Rightarrow\, 2^{-18}\equiv (2^{10})^2 2^{-18}\equiv 2^2\equiv 4$

${\rm mod}\ 13\!:\,\ 2^{12}\equiv 1\,\Rightarrow\, 2^{-18}\equiv (2^{12})^2 2^{-18}\equiv 2^6\equiv -1$

${\rm mod}\ 11\!:\,\ 4\equiv x\equiv -1+13 \color{#a0f}k\equiv -1+2k\iff 2k\equiv 5\equiv 16\iff \color{#a0f}{k\equiv 8}$

Substituting: $\,\ x\equiv -1+13( \color{#a0f}{8\!+\!11n})\equiv 103+143n$

Community
  • 1
Bill Dubuque
  • 272,048
  • Nicely done. I needed a pen and paper for mine (just - I like to think I could have done it without, but ...), and wanted to illustrate a general method. I wish I'd seen this, though. I did get the same answer, which is a comfort. – Mark Bennet Apr 18 '15 at 19:47
  • 1
    @Mark Actually CRT isn't too painful, see the way I'd do it above. – Bill Dubuque Apr 18 '15 at 20:06
  • It wasn't too bad the way I did it, which was similar. I made the mistake of using $12$ rather than $-1$ modulo $13$ (even having calculated $-1$). There were some easy reductions, but ... That's what being tired and distracted does. – Mark Bennet Apr 18 '15 at 20:23
  • 1
    @Mark Generally using least magnitude remainders is simpler for hand calculations (so I should've used $,\color{#a0f}{k\equiv 8\equiv -3},$ above, yielding $,x \equiv -40)\ \ $ – Bill Dubuque Apr 18 '15 at 20:29
1

Finding the inverse is made a bit easier using this method: $$2^2 \equiv 4\bmod 143$$ $$2^4 \equiv 4^2 \equiv 16 \bmod 143$$ $$2^8 \equiv 16^2 \equiv 256 \equiv 113 \bmod 143$$ (it's getting tricky here) $$2^{16} \equiv 113^2 \equiv 12769 \equiv 42 \bmod 143$$ $$2^{18} \equiv 2^{16}2^2 \equiv 42*4\equiv168\equiv25 \bmod 143$$ And since 25 and 143 are coprime, by Euclid's algorithm we have: $$ \begin{matrix} - & - & 5 &1&2&1&1\\ 0 & 1 & -5 & 6 & -17 & 23 & -40 \\ 143 & 25 & 18 & 7 & 4 & 3 & 1 \\ \end{matrix} $$ So -40 is the inverse of $2^{18}\bmod143$, thus $x \equiv 103\bmod143$. Takes some more time than other solutions, but it requires only Euclid's extended algorithm and some long division to do 12369 mod 143.

Joald
  • 625