3
$\begingroup$

I saw somewhere (I appreciate if anyone has any references to proof of this fact) that if $A$ is a finite dimensional associative algebra such that $\textrm{dim}(A)<n$, then $A$ satisfies the standard identity of degree $n$: $$s_n(x_1,\dots, x_n)=\sum\limits_{\sigma\in S_n}\textrm{sgn}(\sigma)x_{\sigma(1)}\cdots x_{\sigma(n)},$$ where $S_n$ is the symmetric group of degree $n$.

Is there a non-associative version of this result? I mean, if $A$ is a finite-dimensional (not necessarily associative) algebra, then $A$ satisfies some kind of "non-associative" standard polynomials? For example, maybe: $$P_1(x_1)=x_1=s_1(x_1),$$ $$P_2(x_1,x_2)=x_1x_2-x_2x_1=s_2(x_1,x_2),$$ $$P_3(x_1,x_2,x_3)=x_1(x_2x_3)-(x_1x_2)x_3-x_1(x_3x_2)+(x_1x_3)x_2-x_2(x_1x_3)+(x_2x_1)x_3+x_2(x_3x_1)-(x_2x_3)x_1+x_3(x_1x_2)-(x_3x_1)x_2-x_3(x_2x_1)+(x_3x_2)x_1=x_1(x_2x_3-x_3x_2)+x_2(x_3x_1-x_1x_3)+x_3(x_1x_2-x_2x_1)+(x_3x_2-x_2x_3)x_1+(x_1x_3-x_3x_1)x_2+(x_2x_1-x_1x_2)x_3$$ and etc... Thanks in advance for answers!

$\endgroup$
  • 3
    $\begingroup$ This is the Amitsur-Levitzki identity. Some basic searching around this name yields the paper Minimal Identities of Octonian Algebras, although this seems to be a particular case of your question. $\endgroup$ – Mark Nov 24 '19 at 6:03
  • 2
    $\begingroup$ The Amitsur-Levistski is quite different. It says that $M_n(k)$ satisfies the identity with $s_{2n}$, although its dimension is $n^2$. $\endgroup$ – Denis Serre Nov 24 '19 at 18:39
  • 1
    $\begingroup$ @LSpice You're right. Thanks! $\endgroup$ – cl4y70n____ Nov 25 '19 at 3:06
4
$\begingroup$

The first question, why the standard identity $s_n$ holds in any $< n$-dim associative algebra is easy: This is a multilinear identity, so in order to check it for arbitrary $x_1,...,x_n$, it is enough to check it for basic elements $x_1,...,x_n$. Since the dim $<n$, two of the $x_1,...,x_n$ are equal, which makes $s_n=0$.

In the non-associative case the situation is slightly more complicated. For every elements $x_1,....,x_n$ of the algebra denote $x_1\cdot ...\cdot x_n=(...(x_1x_2)x_3)...)x_n$ (one can choose any other arrangement of parentheses. Then $s_n$ is defined as usual, and the proof that $s_n=0$ is an identity is as above.

$\endgroup$

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.