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

We have the following symmetric bilinear form on $\\Q^4$ (vector space over rational numbers) with respect to standard basis $\{e_1,e_2,e_3,e_4\}$ $$g(v,w)=v^tAw$$ $$A=\begin{bmatrix}1 & 2 &3&4\\2 & 3 & 4 & 5\\3&4&5&6\\4&5&6&7\end{bmatrix}$$

How to find a basis on which g is diagonal. I suppose using Gram-Schmidt algorithm we find orthogonal basis, but could someone explain this explicitly, please?

share|cite|improve this question
Do you know about eigenvectors? – Joe Johnson 126 Mar 13 '13 at 11:04
May be helpful: – user38268 Mar 13 '13 at 11:23
You don't really need Gram-Schmidt (since you are not looking for an orthonormal basis, and you are working on $Q^4$). If you do the naive thing you'd find that with $v_1 = (1,2,3,4)$ and $v_2 = (0,1,2,3)$. Your $A$ is equal to $v_1 v_1^T - v_2 v_2^T$. It remains to complete this to a basis and finding its dual. – Willie Wong Mar 13 '13 at 11:32
Sorry I need to find orthogonal basis on which g is diagonal. P.S there is some problem in the system and thats why I cant login and posting it as a guest. – user66661 Mar 13 '13 at 23:18
@user66661: you need to register an account and merge your existing accounts. – robjohn Mar 14 '13 at 6:50

There is no need to look for an orthogonal basis. Note that our goal is not to diagonalize $A$ using similarity transform, but to diagonalize $A$ using congruence. That is, what we are looking for is not an invertible matrix $P$ such that $P^{-1}AP$ is diagonal, but an invertible $P$ such that $P^TAP$ is diagonal.

(When $P$ is a real orthogonal matrix, $P^{-1}AP$ and $P^TAP$ coincide, but it is not always possible to find such an orthogonal $P$ over the rationals and this is not our goal anyway.)

You may apply simultaneous elementary row and column reductions to obtain $P$. More specificically, let $$ E_1=\begin{pmatrix}1&0&0&0\\-2&1&0&0\\-3&0&1&0\\-4&0&0&1\end{pmatrix}. $$ Then $$ B=E_1AE_1^T=\begin{pmatrix}1&0&0&0\\0&-1&-2&-3\\0&-2&-4&-6\\0&-3&-6&-9\end{pmatrix}. $$ Now let $$ E_2=\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&-2&1&0\\0&-3&0&1\end{pmatrix}. $$ Then $$ D=E_2BE_2^T=\begin{pmatrix}1&0&0&0\\0&-1&0&0\\0&0&0&0\\0&0&0&0\end{pmatrix}. $$ Now $D=(E_2E_1)A(E_1^TE_2^T)$. So you may take the basis vectors as the column vectors of $P=E_1^TE_2^T$.

share|cite|improve this answer
How did you come to choose$\;E_1,E_2\;$ the way you did? – DonAntonio Oct 24 '13 at 14:07
It's just Gaussian elimination, isn't it? – user1551 Oct 30 '13 at 19:02
Very nice. I have been seeing that people do not differentiate well between orthogonal diagonalization and, well, just diagonalization. There is a published algorithm; please see……… – Will Jagy Aug 22 '15 at 19:21
I think there is a good chance you will be able to see el.Salvador's selection from the Lipschutz and Lipson Schaum's outline by clicking on this image link from his answer: I will check after saving the comment... Yes, worked just fine. – Will Jagy Aug 22 '15 at 20:31
@WillJagy That's really a surprise. I used to think that Schaum's outlines are written for exam preps. – user1551 Aug 23 '15 at 14:08

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.