4
$\begingroup$

It is well known that the equation $$(*)\;\;\;\;ab-ba=1$$ is unsolvable in a Banach algebra.

I search for some reasonable generalization of this equation in higher variable for investigation of solvability of such generalized equations. The above equation can be read as $$(**)\;\;\;\;\;\sum_{\sigma \in S_2} s(\sigma)\prod a_{\sigma_i}=1$$ provided we put $a_1=a,\;a_2=b.$

So our first question is the following:

Is there a Banach algebra with three elements $a,b,c$ which satisfy the following equation?$$(***)\;\;\;\;\;abc+bca+cab-bac-cba-acb=1$$

Our next question:

Apart from $(***)$, What would be some other generalization of $(*)$ whose solvability in a Banach algebra would be an interesting and non trivial question?

$\endgroup$
0

1 Answer 1

8
$\begingroup$

In the algebra of real $2\times 2$ matrices, take $$ a=\left[ \begin{array}{cc} 0 & 1\\ 0 & 0\end{array} \right],\ b=\left[ \begin{array}{cc} 0 & 0\\ 1 & 0\end{array}\right],\ c=\left[ \begin{array}{rr} 1/3 & 0\\ 0 & -1/3\end{array}\right]. $$ In the other direction, is it true that if for all $n$ an equation has no solutions in $n\times n$ real matrices, then it has no solutions in any Banach algebra? What are the equations (or finite sets of equations) that for all $n$ have no solutions in $n\times n$ real matrix algebras?

$\endgroup$
6
  • 9
    $\begingroup$ If you start out with a non-trivial finitely presented group $G$ that has no finite-dimensional representations, then you get relations that are no satisfiable in matrices but are satisfied in $\ell^1(G)$. On the other side, in a suitable form there is no such example known where one would be able to show that relations cannot be "almost" satisfied in matrices (given a suitable norm like the operator norm). $\endgroup$ May 8, 2018 at 15:27
  • 2
    $\begingroup$ Can this be put in equational form: $xy = 1 \neq yx$? No matrix algebra has such elements, but there are plenty of Banach algebras that do. More generally, there are traces on any subalgebra of a product of matrix algebras, but not for Banach algebras in general. $\endgroup$ May 8, 2018 at 19:33
  • $\begingroup$ @RichardStanley Dear Professor Stanley thank you very much for this interesting example. $\endgroup$ May 9, 2018 at 7:49
  • $\begingroup$ @DavidHandelman Dear Prof. Handelman, according to your comment it would be interesting to find an equation which does not have a solution in finite dimensional $C^*$ algebras but it has a solution in a $C^*$ algebra which admit a faithful trace.Is there an equation with this property? $\endgroup$ May 9, 2018 at 8:19
  • 1
    $\begingroup$ @DavidHandelman: You can express stronger conditions like the Cuntz algebra relations: $xy+wz=1$ and $yx=zw=1$. $\endgroup$ May 9, 2018 at 10:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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