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

Does a quadratic form always come from symmetric bilinear form ? We know when $q(x)=b(x,x)$ where $q$ is a quadratic form and $b$ is a symmetric bilinear form. But when we just take a bilinear form and $b(x,y)$ and write $x$ instead of $y$,does it give us a quadratic form ?

share|cite|improve this question

Yes, every quadratic form (over a finite-dimensional $\mathbb{R}$-space) can be expressed in terms of a symmetric bilinear form, because if your quadratic form $Q(x)$ (for $x \in \mathbb{R}^n$) is written as $$ Q(x) = \sum_{i\le j} c_{ij} x_i x_j $$ then $Q(x) = x^T Ax$, where $A$ is a symmetric matrix given by $A = (a_{ij})$ with $$a_{ij} = a_{ji} = \frac{c_{ij}}{2}, i < j, \quad a_{ii} = c_{ii}$$

share|cite|improve this answer
yes but my question is a little general.i mean if we have any quadratic form does it have to be come from symmetric bilinear form ? – Turku Kirli Nov 18 '12 at 21:50
I just constructed a symmetric matrix $A$ such that $Q(x) = x^T Ax$, so then the map $B(x,y) = x^T A y$ is a symmetric bilinear form. – Christopher A. Wong Nov 18 '12 at 22:08
but when we think as a function,if we have a quadratic form then do we have to have a bilinear form ? – Turku Kirli Nov 18 '12 at 22:17
The construction I gave shows that for every quadratic form (on a finite-dim space) $Q(x)$, we can write $Q(x) = B(x,x)$ for a symmetric form. – Christopher A. Wong Nov 18 '12 at 22:22
ok but what about non-symmetrics ? :) – Turku Kirli Nov 18 '12 at 22:25

Sure. In finite dimension over the real numbers you do not even need the bilinear form symmetric, because any matrix is the sum of a symmetric and a skew-symmetric matrix, and these are easy to find.

Meanwhile, as long as characteristic of the underlying field is not $2,$ you can go back with $$ c(x,y) = \frac{1}{2} \left( q(x+y) - q(x) - q(y) \right). $$ Since this recipe gives a symmetric bilinear form, those are preferred.

Even in infinite dimension, I suppose you can symmetrize with $$ f(x,y) = \frac{1}{2} \left( h(x,y) + h(y,x) \right). $$ There won't be any matrices.

share|cite|improve this answer
I see but there is something missing i guess.because if we have just bilinear form (not symmetric) and write x instead of y then we get quadratic i wrong ? – Turku Kirli Nov 18 '12 at 21:46
@TurkuKirli, if you start with non-symmetric you still get a quadratic form. It may be the constant $0$ form, for example. If your original form is not symmetric you cannot recover it from the quadratic form. – Will Jagy Nov 18 '12 at 21:54
But i still can not get it :/ i mean when we have bilinear form yes we may get a quadratic form.but what about the inverse ? when we have a quadratic form ( yes we can get a symmetric bilinear form) but can not we get any bilinear form ? – Turku Kirli Nov 18 '12 at 21:59
@TurkuKirli, I sent you an email with some preliminary observations. You should probably delete the message with your email address now. – Will Jagy Nov 18 '12 at 23:33
okey thank you :) – Turku Kirli Nov 19 '12 at 5:32

If we have b symmetric bilinear form we can get q quadratic form
$q\colon V \to \mathbb{R}$

conversely if q is a quadratic form
$q\colon V \to \mathbb{R}$
we can define
$\frac 12$(q(v+w)-q(v)-q(w)):=b(v,w)

the vital answer is you just get a bilinear form not always a symmetric bilinear form. because the definition $\frac 12$(q(v+w)-q(v)-q(w)) leads us to $\frac 12$(b(v,w)+b(w,v))

share|cite|improve this answer
For your converse part, you need to mention that the characteristic of the field needs to be different from 2, other wise your 1/2 will not exist. – Old John Jan 5 '13 at 14:33
Sorry I didn't get the point.Don't i do my works on the field $\mathbb{R}$? So do I have to say anything else? – Serkan Yaray Jan 5 '13 at 14:41
Well, the question does not mention that the underlying field is $\mathbb{R}$, so, as Will Jagy mentions in his answer, we have make sure that the characteristic is not 2. – Old John Jan 5 '13 at 14:46
@SerkanHafızYaray Why do you say "not always a symmetric bilinear form"? The bilinear form defined by b(v,w)=(q(v+w)-q(v)-q(w))/2 is always symmetric, by the commutativity of addition. – rschwieb Mar 6 '13 at 19:42

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.