Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $G$ be either a finite group or a Lie group. Let $r_1$ and $r_2$ be two linear representations of $G$. $r_1$ is an ordinary representation but $r_2$ may be projective. Is there a criterion to decide the following question: given $G$, $r_1$, $r_2$, does there exist an ordinary linear representation $R$ such that $r_1 \otimes R$ is isomorphic to $r_2 \otimes R$? As a slight refinement of this question: does there exist an ordinary linear representation $R$ and two ordinary linear representations $T_1$ and $T_2$ where $T_1$ and $T_2$ are each the direct sum of several copies of the trivial representation (possibly a different number of copies in $T_1$ and $T_2$) such that $r_1 \otimes T_1 \otimes R$ is isomorphic to $r_2 \otimes T_2 \otimes R$?

One obvious case when the answer is "no" is that $r_2$ is not an ordinary representation. A second case is if the group is, for example, $SU(2)$ and $r_1$ is a spin $S_1$ representation and $r_2$ is a spin $S_2$ representation with $S_1 > S_2$. Then, $r_1 \otimes R$ will contain higher spin representations than $r_2 \otimes R$ will. Is there a general answer?

share|improve this question
3  
If $r_1$ and $r_2$ are two ordinary linear representations of the same degree of a finite group $G$ and $R$ is the regular representation, then $r_1\otimes R\cong r_2\otimes R$. –  Richard Stanley Apr 25 '13 at 17:48
    
Does "degree" mean the same thing as dimension of a representation? In which case are you saying that if $r_2$ is an ordinary representation then the answer is always "yes" to the second question, as I can tensor in $T_1$ and $T_2$ to make the dimension of $r_1 \otimes T_1$ match that of $r_2 \otimes T_2$ and then pick $R$ to be the regular representation? –  Matt Hastings Apr 25 '13 at 17:56
    
Is $R$ allowed to be infinite-dimensional (in the Lie case)? If so, and we take $R$ to be left regular representation of $G$ on $L^2(G)$, then I believe Richard Stanley's observation would still work, with $r\otimes R$ being isomorphic to an amplification of $R$. (Fell's absorption principle.) –  Yemon Choi Apr 26 '13 at 1:24

1 Answer 1

up vote 3 down vote accepted

If $G$ is a compact connected Lie group, then the ring of characters is an integral domain (it is the ring of Weyl group invariants in the co-ordinate ring of a maximal torus). Let $r_1,r_2, R, T_1, T_2$ be as in the question. Assume that $r_1,r_2$ are ordinary irreducible representations

If $t_1,t_2$ are the dimensions of $T_1, T_2$, and $\chi _1$ and $\chi _2$ are the characters of $r_1,r_2$, and since the character of $R$ may be divided out in the integral domain, it follows from the assumptions that $t_1\chi _1=t_2\chi _2$. Hence $\chi _1$ occurs in the $t_2$ fold direct sum of $\chi _2$ with itself; that is, $\chi _1=\chi _2$. This is supposing that $r_2$ is an ordinary representation.

If $G$ is finite, then the comment of Richard Stanley already gives a counter-example.

share|improve this answer
    
Thanks! So it seems the question has a simple answer in both the finite and Lie cases, but the answer is very different in these two cases. –  Matt Hastings Apr 25 '13 at 18:41
    
I would say the difference comes from the connectedness or otherwise of the group. If $G$ is not conected, then the group of connected components may well be of the type given by Richard Stanley. –  Venkataramana Apr 26 '13 at 10:57
    
See my comment above. I think the difference is finiteness, but perhaps I have made an error –  Yemon Choi Apr 26 '13 at 18:29
    
I mean the following; suppose $F$ is a finite group for which Richard Stanley's example works. Suppose $G$ is a Lie group whose connected component group $G/G^0$ is $F$. Then for $G$ also, there exist two reps $r_1,r_2$ as above. –  Venkataramana Apr 27 '13 at 3:43

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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