1
$\begingroup$

Let $u_1,...,u_k\in\mathbb{R}^n$ such that there is a non-zero quadratic form $Q$ satisfying $Q(u_i)=1$ for all $i=1,...,k$. Is there a positive definite quadratic form satisfying the same equations?

I know that the set of positive definite forms is open (and convex?) in the space of quadratic forms (considered as $\mathbb{R}^{\frac{1}{2}n(n+1)}$). Also, I tried considering the above equations as $k$ equations in $\frac{1}{2}n(n+1)$ variables, but not really sure how this helps, since these aren't homogenous equations.

Any help would be appreciated.

$\endgroup$
1
$\begingroup$

It seems the following.

A negative answer is suggested because a quadratic form $Q$ is positively definite iff a set $$\{u\in\Bbb R^n:Q(u)=1\}$$ is bounded, which can be seen considering a canonical form $$Q(x_1,\dots,x_n)=\lambda_1x_1^2+\dots +\lambda_nx_n^2.$$ Also we can present an explicit example. Let $n=2$ and $Q(x,y)=x^2-2y^2$. Then $$Q(1,0)=Q(3,2)=Q(17,12)=1.$$ Let $P(x,y)=ax^2+bxy+cy^2$ be a quadratic form such that $$P(1,0)=P(3,2)=P(17,12)=1.$$ Then $a=1$ and

$\cases{ 9+6b+4c=1\\ 289+204b+144c=1}$

$\cases{3b+2c=-4\\ 17b+12c=-24}$

$\cases{ 18b+12c=-24\\ 17b+12c=-24}$

so $b=0$ and $c=-2$. Hence $P=Q$.

$\endgroup$

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.