Let us work over $\mathbb C$. Suppose that $G$ is a semisimple algebraic group and let $H \subset G$ be a maximal torus. Consider a dominant weight $\omega$, then one can associate a unique irreducible $G$-representation $V(\omega)$.

Consider the following example. Let $G=\mathrm{SL}_3(\mathbb C)$ and let $W=\mathbb C^3$ be the standard $G$-representation. We define $$ V=\mathrm{Sym}^2W, $$ it is a $6$-dimensional irreducible representation for $G$. Moreover, we can also see $V$ as the space of symmetric $3 \times 3$ matrices. In particular, it inherits the structure of unital associative algebra using the matrix product. This is an example of Jordan algebra. On the other hand, if we define $$ \tilde V=W \otimes W^\vee, $$ it contains an irreducible $G$-representation that corresponds to the Lie algebra $\mathfrak g$.

In fact, these are all examples of $G$-representations that have also an algebra structure. My question is: given a group $G$ and a dominant weight $\omega$, can we know when $V(\omega)$ has a unital algebra structure?

  • 2
    $\begingroup$ Any algebra $V$ can be viewed as a map (the multiplication) $\mu: V \otimes V \to V$. If we don't require associativity OR any relation to $G$, then of course we can define an algebra structure on every irrep $V$ of $G$. Of course you do want some relation to $G$, I expect you want $\mu$ to be a $G$-map? Then the first step is to find out (using Weyl character formula or such) for which $V$ we have that $V$ appears as a summand in the decomposition of $V \otimes V$ into irreps. This is already a restriction. $\endgroup$
    – Vincent
    Commented Sep 7, 2021 at 9:24
  • $\begingroup$ The second issue of the resulting algebra being unital or not is trickier (that is why my two comments are comments rather than answers) but I am also not sure if we agree on what that word means. I thought it means there must be an element $e \in V$ such that $ex = xe = x$ for all $x \in V$. But this doesn't hold in the case of Lie algebras, which seems to be one of your examples $\endgroup$
    – Vincent
    Commented Sep 7, 2021 at 9:26
  • 1
    $\begingroup$ @Vincent: Yes, I want that $\mu$ is $G$-equivariant. I correct the misprint: I want the algebra to be unital in the sense that you wrote $\endgroup$
    – Bobech
    Commented Sep 7, 2021 at 9:35
  • 3
    $\begingroup$ For symmetric matrices the product is not an operation (the product of two symmetric matrices is not always symmetric). You need the symmetrized product $AB+BA$, which is not associative (but indeed Jordan). $\endgroup$
    – YCor
    Commented Sep 7, 2021 at 10:34
  • 1
    $\begingroup$ Correct, the operation on $Sym^2$ is $A.X = AXA^T$, not $A.X = AXA^{-1}$ like in the case of $W\otimes W^\ast$. The latter does commute with the Jordan algebra structure, the former does not. $\endgroup$ Commented Sep 7, 2021 at 13:26

1 Answer 1


The unit poses a problem here. If you require the multiplication map $\cdot: V\otimes V\to V$ to be $G$-equivariant, then this means $gv\cdot gw = g(v\cdot w)$. Letting $w=1_V$ and using that $g:V\to V$ is surjective, we see that $g(1_V) = 1_V$. In particular: $\mathbb{C} 1_V$ is a subrepresentation of $V$ isomorphic to the trivial representation. If $V$ is irreducible, this forces $V$ itself to be the trivial representation.


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.