2
$\begingroup$

Prove: $$ \det\left[ \begin{array}{cccc} 1+x_1y_1 & x_1y_2 & \cdots & x_1y_n\\ x_2y_1 & 1+x_2y_2 & \cdots & x_2y_n \\ \vdots & \vdots & \ddots & \vdots \\ x_ny_1 & x_ny_2 & \cdots & 1+x_ny_n \\ \end{array} \right]=1+\sum_{i=1}^{n}x_iy_i$$

I tried to do some elementary operations, and develop by first row, but couldn't get further.

Need to be proven without using eigenvalues.

Any help appreciated.

$\endgroup$
  • 1
    $\begingroup$ I am fairly sure you can modify math.stackexchange.com/a/2112473 to solve the more general question of computing $\det \left( \begin{array}{cccc} a_1 & x_1 y_2 & \cdots & x_1 y_n \\ x_2 y_1 & a_2 & \cdots & x_2 y_n \\ \vdots & \vdots & \ddots & \vdots \\ x_n y_1 & x_n y_2 & \cdots & a_n \end{array} \right)$. $\endgroup$ – darij grinberg Feb 26 '17 at 15:34
4
$\begingroup$

The determinant can be written under the form:

$\det(I_n+UV^T)$ where $U=\left(\begin{array}{c}x_1\\x_2\\ \vdots \\x_n\end{array}\right)$ and $V=\left(\begin{array}{c}y_1\\y_2\\ \vdots \\y_n\end{array}\right).$

Let us recall that the matrix determinant lemma (https://en.wikipedia.org/wiki/Matrix_determinant_lemma) says that

$$\det(A+UV^T)=(1+V^TA^{-1}U) \det(A)$$

Taking $A=I$ gives:

$$\det(I_n+UV^T)=1+V^TU$$

which is the desired result.

$\endgroup$
1
$\begingroup$

Use $n$-linearity with respect to the first column to put the first $1$ alone on the first column.

$\endgroup$
  • $\begingroup$ Could you please explain? $\endgroup$ – Itay4 Feb 26 '17 at 15:54
0
$\begingroup$

The matrix $A=(x_1,\dots,x_n)^T(y_1,\dots,y_n)$ is rank 1, so its characteristic polynomial is $X^n-\text{tr}(A)X^{n-1}$. Put $X=-1.$

Question has changed since this answer!

$\endgroup$
  • $\begingroup$ You should delete this answer as you have given another one. $\endgroup$ – Jean Marie Feb 27 '17 at 8:11
0
$\begingroup$

If you just take a deep breath and calculate it will come out!

For after all, the answer is clearly going to be a sum of terms each of which is a product of $k$ $x$'s and $k$ $y$'s, for $k=0,1,\dots,n$. Every such term will arise as a product of $n-k$ $1$'s from the diagonal, multiplied by the top degree term of its cofactor, that is, multiplied by $\det\hat{x}^T\hat{y}$, where $\hat{x}$ is just the $k$-vector got from some $k$ of the indices. But now this little matrix is rank $1$ and so singular for all $k\not=0,1$. The only non-zero terms are then $1$ and $\sum x_iy_i$.

$\endgroup$

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.