0
$\begingroup$

Let $A$ be an algebra over a field $k$. Suppose that we have a non degenerate bilinear pairing $\beta:A \otimes A \to k$.

Let $\{a_i\}$ be a basis of $A$.

I would like to show that there exist a basis $\{a'_i\}$ of $A$ such that we have $\beta(a_i, a'_j)=\delta_{ij}$, where $\delta_{ij}$ is the Kronecker delta.

I thought I could use Gram-Schmidt process. But the problem is not to construct an orthonormal basis (the basis $\{a_i\}$ is fixed).

Also is it always true that $\beta(a_i,a_i)\neq0$?

I appreciate any help.

$\endgroup$
1
$\begingroup$

This has nothing to do with algebras in particular and is just a construction on finite-dimensional vector spaces. If $V$ is a finite-dimensional vector space over a field $k$, equipped with a nondegenerate bilinear pairing $\beta$, then the natural map $\iota: V \to V^*$ sending $w \mapsto (v \mapsto \beta(v,w))$ is injective (by nondegeneracy of $\beta$) and therefore surjective as well (since $\dim V^* = \dim V$ and both are finite-dimensional). So $\iota$ gives an isomorphism between $V$ and $V^*$.

Hence, given a basis $\{a_i\}$, we may, for each $j$ ($1 \le j \le n$), construct an element of $V^*$ which sends $a_i \to \delta_{ij}$ (i.e. $a_j$ to 1 and all other $a$'s to 0). The inverse image of this element under $\iota$ is the $a'_j$ you are looking for.

It is not true, in general, that $\beta(a_i, a_i) \ne 0$ for $a_i \ne 0$. For example, over any field $k$, we could take the nondegenerate inner product on the vector space $k^n$ defined by $\beta(x,y) = \sum x_i y_i$. Then $\beta(x,x) = \sum x_i^2$ and the right side could be 0 even if $x \ne 0$, for many fields $k$.

$\endgroup$

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.