Where do I foul up?
In (e.g.) $SU(3)=A_2$ the tensor product $(1,0)\otimes (1,0)=(2,0)+(0,1)$. No $(0,0)$. But on the other hand still $(1,0)\otimes (0,0)=(1,0)$. This immediately begs the question
- what happened to Schur's lemma and
- imagine some 6j symbol $(ABC|DEF)$ where $A=B=(1,0)$, $C=(0,0)$. Is $ABC$ no admissible triangle and the 6j vanishes?
My only idea is that the Dynkin diagram is symmetric and you first must symmetrize/antisymmetrize over the reps to get the correct ones, i.e. the proper irreps in this case would be $((1,0)+(0,1))/V2$ and $((1,0)-(0,1))/V2$. (And what does this do to the dimensions? Same as $(1,0)$ and $(0,1)$, I think.)