1
$\begingroup$

Here is a determinant of a $(k+m) \times (k+m)$ Block matrix.

\begin{align} D=\begin{vmatrix} a_{11} &a_{12} & \cdots & a_{1k} &0 &\cdots &0 \\ a_{21}& a_{22}& \cdots & a_{2k} & 0 &\cdots &0 \\ \vdots& \vdots & & \vdots & \vdots & &\vdots\\ a_{k1} & a_{k2} & \cdots & a_{kk} & 0 &\cdots & 0\\ c_{11}& c_{12} & \cdots& c_{1k} & b_{11} & \cdots & b_{1m}\\ \vdots& \vdots & & \vdots & b_{21}&\cdots & b_{2m}\\ c_{m1}& c_{m2} & \cdots & c_{mk} & b_{m1}& \cdots & b_{mm} \end{vmatrix} \end{align}

If I have got a determinant $$D_1= \begin{vmatrix} 0 &\cdots &0&a_{11} &a_{12} & \cdots & a_{1k} \\ 0 &\cdots &0 &a_{21}& a_{22}& \cdots & a_{2k} \\ \vdots& & \vdots & \vdots & \vdots & &\vdots\\ 0 &\cdots & 0&a_{k1} & a_{k2} & \cdots & a_{kk} \\ b_{11} & \cdots & b_{1m}& c_{11}& c_{12} & \cdots& c_{1k}\\ \vdots& & \vdots & \vdots & \vdots& & \vdots\\ b_{m1}& \cdots & b_{mm}&c_{m1}& c_{m2} & \cdots & c_{mk} \end{vmatrix} $$ Then $D_1$ is equal to $(-1)^{k \times m}$$D$.

I know that the existence of the factor -1 is due to the interchange of 2 row, but i have a question on that $k \times m$.In my book,it said that i have to do $k\times m$ times row operations to transform $D_1$ into $D$.However,i thought only k times is needed for $D_1$ transform into $D$.

If i have done 1 times row operation for $D_1$ $$D_1= \begin{vmatrix} 0 &\cdots &0&a_{11} &a_{12} & \cdots & a_{1k} \\ 0 &\cdots &0 &a_{21}& a_{22}& \cdots & a_{2k} \\ \vdots& & \vdots & \vdots & \vdots & &\vdots\\ 0 &\cdots & 0&a_{k1} & a_{k2} & \cdots & a_{kk} \\ b_{11} & \cdots & b_{1m}& c_{11}& c_{12} & \cdots& c_{1k}\\ \vdots& & \vdots & \vdots & \vdots& & \vdots\\ b_{m1}& \cdots & b_{mm}&c_{m1}& c_{m2} & \cdots & c_{mk} \end{vmatrix} =\begin{vmatrix} a_{11} &\cdots &0&0 &a_{12} & \cdots & a_{1k} \\ a_{21} &\cdots &0 &0& a_{22}& \cdots & a_{2k} \\ \vdots& & \vdots & \vdots & \vdots & &\vdots\\ a_{k1} &\cdots & 0& 0& a_{k2} & \cdots & a_{kk} \\ c_{11} & \cdots & b_{1m}&b_{11} & c_{12} & \cdots& c_{1k}\\ \vdots& & \vdots & \vdots & \vdots& & \vdots\\ c_{m1}& \cdots & b_{mm}&b_{m1}& c_{m2} & \cdots & c_{mk} \end{vmatrix} $$ Correct me if i have made any mistakes

$\endgroup$
1
$\begingroup$

You can think of it in this way: take column $k+1$ (the first one containing $b$s) and swap it with column $k$. Then swap column $k$ with column $k-1$, etc, until you have "bubbled" the original column $k+1$ all the way to the left, after a total of $k$ swaps.

Repeating this process for all $m$ columns on the right requires a total of $km$ swaps.

$\endgroup$
1
$\begingroup$

The discrepancy is that you have to interchange two adjacent rows. From this it is quite clear you have to perform $k\cdot m$ such operations.

$\endgroup$
  • $\begingroup$ Is that means i can't interchange the 1st row with the kth row directly?I have to interchange 1st with 2nd and then with 3rd....until kth.Is this a special case for row operation in block matrices $\endgroup$ – Vulcan Feb 7 '13 at 16:06
  • $\begingroup$ If you swap the 1st and $k+1$st row directly, that leaves the $k+1$st row in the right place, but the 1st row in the wrong place (unless $m=k$.) $\endgroup$ – user7530 Feb 7 '13 at 16:09
  • $\begingroup$ I have to think about it.Thank you $\endgroup$ – Vulcan Feb 7 '13 at 16:26
  • $\begingroup$ When i think of the situation of m<k,i got what you mean finally.Thx again $\endgroup$ – Vulcan Feb 7 '13 at 16:53
0
$\begingroup$

Apply Laplace expansion http://en.wikipedia.org/wiki/Laplace_expansion. you can directly calculate $D_1$

$\endgroup$

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.