84
$\begingroup$

I stumbled across the following problem and found it cute.

Problem: We are given that $19$ divides $23028$, $31882$, $86469$, $6327$, and $61902$. Show that $19$ divides the following determinant:

$$\left| \begin{matrix} 2 & 3&0&2&8 \\ 3 & 1&8&8&2\\ 8&6&4&6&9\\ 0&6&3&2&7\\ 6&1&9&0&2 \end{matrix}\right|$$

$\endgroup$
  • 1
    $\begingroup$ This is nice! May I ask about the source of the problem - where did you see it? $\endgroup$ – Martin Sleziak Jun 8 '12 at 12:17
  • 3
    $\begingroup$ Golan's linear algebra book. $\endgroup$ – Potato Jun 8 '12 at 16:04
90
$\begingroup$

Multiply the first column by $10^4$, the second by $10^3$, third by $10^2$ and fourth by $10$ - this will scale the value of the determinant by $10^{4+3+2+1}=10^{10}$, which is coprime to $19$. Now add the last four columns to the first one - this will not change the value of the determinant. Finally notice the first column now reads $23028, 31882, 86469, 6327$, and $61902$: each is a multiple of $19$ so we can factor a nineteen cleanly out the determinant.

$\endgroup$
47
$\begingroup$

If the determinant is $0$ it is obvious that $19|0$. Now suppose that the determinant is not $0$.

$$\begin{align*} 2\cdot10^4+3\cdot10^3+0\cdot10^2+2\cdot10+8\cdot1&=23028\\ 3\cdot10^4+1\cdot10^3+8\cdot10^2+8\cdot10+2\cdot1&=31882\\ 8\cdot10^4+6\cdot10^3+4\cdot10^2+6\cdot10+9\cdot1&=86469\\ 0\cdot10^4+6\cdot10^3+3\cdot10^2+2\cdot10+7\cdot1&=06327\\ 6\cdot10^4+1\cdot10^3+9\cdot10^2+0\cdot10+2\cdot1&=61902 \end{align*}$$

By Cramer's rule

$$1=\frac{\left|\begin{matrix} 2 & 3 & 0 & 2 & 23028 \\ 3 & 1 & 8 & 8 & 31882 \\ 8 & 6 & 4 & 6 & 86469 \\ 0 & 6 & 3 & 2 & 06327 \\ 6 & 1 & 9 & 0 & 61902 \end{matrix}\right|}{\left|\begin{matrix} 2 & 3 & 0 & 2 & 8 \\ 3 & 1 & 8 & 8 & 2 \\ 8 & 6 & 4 & 6 & 9 \\ 0 & 6 & 3 & 2 & 7 \\ 6 & 1 & 9 & 0 & 2\end{matrix}\right|}$$

Then

$$\left|\begin{matrix} 2 & 3 & 0 & 2 & 8 \\ 3 & 1 & 8 & 8 & 2 \\ 8 & 6 & 4 & 6 & 9 \\ 0 & 6 & 3 & 2 & 7 \\ 6 & 1 & 9 & 0 & 2\end{matrix}\right|=\left|\begin{matrix} 2 & 3 & 0 & 2 & 23028 \\ 3 & 1 & 8 & 8 & 31882 \\ 8 & 6 & 4 & 6 & 86469 \\ 0 & 6 & 3 & 2 & 06327 \\ 6 & 1 & 9 & 0 & 61902 \end{matrix}\right|$$

But last determinant is obviously divisible by $19$.

$\endgroup$
  • 1
    $\begingroup$ I think that I must delete this answer because there is another answer which is accepted and has many up votes. $\endgroup$ – Gaston Burrull Jun 8 '12 at 5:32
  • 1
    $\begingroup$ No, yours is more detailed. You are fine. $\endgroup$ – Potato Jun 8 '12 at 5:43
  • 8
    $\begingroup$ Your answer is good. Multiple approaches can shed new light on a problem. $\endgroup$ – Potato Jun 8 '12 at 5:46
  • 7
    $\begingroup$ @GastónBurrull: Your answer is good. I would not have thought to use Cramer's rule. Please do not delete it. $\endgroup$ – MJD Jun 8 '12 at 17:55
  • 1
    $\begingroup$ Just saw this question since it was recently modified. Simpler, at least to me, is to remember that one can always add a multiple of one column to another and not change the determinant. So add $10^4$ times the first column plus $10^3$ times the second column plus $10^2$ times the third column plus $10^1$ times the fourth column to the fifth column. These column operations do not change the determinant and yield the last equation above. $\endgroup$ – robjohn Jul 12 '15 at 13:04
37
$\begingroup$

Integer proof

Perform the column operation $C_5\leftarrow 10^4C_1+10^3C_2+10^3C_3+10C_4+C_5$: the coefficient of $C_5$ is $1$ so this doesn't change the determinant.

All elements of $C_5$ ($23028$, $31882$, $86469$, $6327$, and $61902$) are now divisible by $19$, so we can factor out $19$: hence the determinant is divisible by $19$.


Modular proof

In $\mathbb Z/19\mathbb Z$, the columns $10^4C_1+10^3C_2+10^3C_3+10C_4+C_5$ sum to $0$: hence the matrix is not invertible and has determinant $0$. So in $\mathbb Z$, the determinant is a multiple of $19$.

$\endgroup$
  • 3
    $\begingroup$ I like your modular proof! $\endgroup$ – Potato Jun 8 '12 at 19:15

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.