2
$\begingroup$

Show that given $A=\left( \begin{array}{cc}\alpha &\beta\\\beta&\delta\\\end{array}\right) \in M_2(\mathbb{R}^2)$, $\langle x,y \rangle = x^TAy, (x,y \in \mathbb{R}^2)$, defines an inner product on $\mathbb{R}^2$ if and only if $\alpha >0$ and det$(A)>0$.

I tried to prove the $\Leftarrow$ direction first:

($\Leftarrow$) Assume that $\alpha > 0, det(A) > 0 \Rightarrow \alpha \delta > \beta^2 \Rightarrow \delta > 0$.

Now I need to show that $x^TAy$ defines an inner product on $\mathbb{R}^2$, so I need to show that $\langle x,y \rangle$ is bilinear, symmetric and positive definite $\forall x,y \in \mathbb{R}^2$

Bilinearity: Let $x_1,x_2,y_1,y_2,x,y \in \mathbb{R}^2, \lambda, \mu \in \mathbb{R}$, then $$\langle \lambda_1 x_1 + \lambda_2 x_2, y \rangle = (\lambda_1 x_1 + \lambda_2 x_2)^TAy= \lambda_1 x_1^TAy + \lambda_2 x_2^TAy = \lambda_1 \langle x_1,y \rangle + \lambda_2 \langle x_2,y \rangle$$

$$ \langle x, \mu_1y_1 + \mu_2y_2 \rangle = x^TA( \mu_1y_1 + \mu_2y_2) = x^TA\mu_1y_1+xA\mu_2y_2= \mu_1 \langle x,y_1 \rangle + \mu_2\langle x, y_2 \rangle$$

Symmetry follows just as easily and is also not affected by $\alpha > 0, det(A)>0$

But I have encountered some problems when trying to prove positive definiteness.

If $ x = \left( \begin{array}{c} \xi_1 \\ \xi_2 \\ \end{array} \right)$, then $$x^TAx = \alpha \xi_1^2 + 2\beta \xi_1 \xi_2 + \delta \xi_2^2$$

Now I'm not sure why this has to be greater than 0 if $\alpha > 0, det(A)>0$, if I plug in numbers it works every time, but I can't do the math to prove this. Can anybody help me please?

I also don't really know how to prove the other direction, so would appreciate some help on that as well.

$\endgroup$
  • 3
    $\begingroup$ Write $xAx$ in the form $$xAx = \alpha\left(\xi_1+\frac{\beta}{\alpha}\xi_2\right)^2 + \frac{\alpha\delta-\beta^2}{\alpha}\xi_2^2.$$ $\endgroup$ – ivanpenev Jun 1 '14 at 17:49
  • $\begingroup$ Should that be $\langle x,y\rangle:=x^TAy$? $\endgroup$ – Cameron Buie Jun 1 '14 at 17:52
  • $\begingroup$ @CameronBuie Yes it's supposed to be $x^Ty$, I'll edit my post. $\endgroup$ – eager2learn Jun 1 '14 at 17:53
  • $\begingroup$ @ivanpenev Thanks a lot, I didn't see that at all. $\endgroup$ – eager2learn Jun 1 '14 at 17:54
  • $\begingroup$ @eager2learn: You're welcome. $\endgroup$ – ivanpenev Jun 1 '14 at 17:57
2
$\begingroup$

This is a general theorem: a real symmetric square matrix defines an inner product iff it is positive definite, which means $\;\alpha\,,\,\det A >0\;$ , as given:

$\implies:\;\;$ taking $\;x=y=\binom10\;$ , we have that

$$x^tAx=(1\;0)\begin{pmatrix}\alpha&\beta\\\beta&\delta\end{pmatrix}\binom10=(1\;0)\binom\alpha\beta=\alpha$$

Since $\;x\neq 0\;$ the above must be positive and we have thus that $\;\alpha>0\;$

Taking $\;x=(0,1)\;$ and doing something similar as above, we get $\;\delta>0\;$ .

Finally, taking $\;x=\binom\delta{-\beta}\neq0\;$ ,we get

$$x^tAx=(\delta\;-\beta)\binom{\alpha\delta-\beta^2}{0}=\delta(\alpha\delta-\beta^2)$$

Again, the above must be positive, and since $\;\delta>0\;$ we get $\;\det A:=\alpha\delta-\beta^2>0\;$

$\endgroup$

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.