Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am not quite sure what this question is asking for:

Given $f(\vec{x})=x^2+xy+y^2+yz+z^2+xz$, find a basis for the corresponding inner product on $\mathbb R^3$.

(I was told that there is an obvious orthogonal basis -- obtainable by inspection)

I don't even know what "corresponding inner product" means...


share|cite|improve this question
A positive-definite quadratic form is essentially an inner product. To get an explicit expression for the symmetric bilinear form, look up the polarisation formula. – Zhen Lin Dec 2 '11 at 0:19
Thanks, @ZhenLin, but I still don't quite understand this, would you mind elaborating? – Owen H. Dec 2 '11 at 0:31
@ZhenLin: What does it mean to find a basis for an inner product? – Owen H. Dec 2 '11 at 0:39
The problem is a bit vague, but note $f(\vec{x}) = \vec{x}^T A \vec{x}$ where $A = \left[\begin{matrix} 1 & 1/2 & 1/2\\1/2 & 1 & 1/2\\1/2 & 1/2 & 1\end{matrix}\right]$. Given positive semidefinite matrix $B$, one can define an inner product: $\langle \vec{x}, \vec{y} \rangle_B = \vec{x}^T B \vec{y}$. See the connection? – p.s. Dec 2 '11 at 0:51
I think "find a basis for an inner product" means "find a basis B for the vector space such that B is orthonormal with respect to the inner product." – Gerry Myerson Dec 2 '11 at 3:26

I'm not sure how much to give away. The matrix being introduced is $$ M \; = \; \left( \begin{array}{rrr} 1 & \frac{1}{2} & \frac{1}{2} \\ \frac{1}{2} & 1 & \frac{1}{2} \\ \frac{1}{2} & \frac{1}{2} & 1 \end{array} \right) , $$

which is symmetric, and an orthogonal basis is indeed pretty easy to find. If required, one may then normalize to get an orthonormal basis. If we take a column vector $x$ and its transpose $x^T,$ then the value of the quadratic form at the point $x$ is just $$ f(x) = x^T \; M \; x. $$ The inner product of two vectors $x,y$ is given by $$ \langle x,y \rangle = \frac{1}{2} \left(f(x+y) - f(x) - f(y)\right) = x^T M y = y^T M x$$ which is one way to write polarization.

Here is something not mentioned. If you take all three entries in $\vec{x}$ to be integers, the result $f(\vec{x})$ is a nonnegative integer. Not all positive integers show up, though. A number $n \geq 0$ has a "representation" $n=f(x)$ if and only if $n$ is not of the form $4^k (16 m + 14).$ Go figure.

share|cite|improve this answer
Thanks, Will. Is the "basis" simply the columns of $M$? I don't know what the definition of a basis of an inner product is... Also, very interesting last point! – Owen H. Dec 2 '11 at 0:45
In any case, the columns of M are not the basis of anything. From what you asked, find an orthogonal basis of eigenvectors of M... What course is this for? The type of answer for a linear algebra course (everything over the real numbers) is rather different from the best reply for a number theory course. – Will Jagy Dec 2 '11 at 0:53
It is for linear algebra. – Owen H. Dec 2 '11 at 1:08
Though the $x,y,z$ may or may not be real. – Owen H. Dec 2 '11 at 1:16
What is the textbook? – Will Jagy Dec 2 '11 at 2:12

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.