Representation theory. Please help by finding the error leading to absurd conclusion

82 Views Asked by At

Claim: all irreducible representations of a given group, $G$, are pairwise isomorphic.

Proof: Let $[r(g), V]$ and and $[p(g), W]$ be irreducible representations of group $G$. Select a fixed vector $v_0$ in $V$, which is NOT an eigenvalue of $r(g)$ for any $g$, except $**1**$. Define the transformation $T$, that maps vector $r(g)v_0$ (in $V$) to vector $p(g)w_0$ in $W$. Extend this mapping to include any $g$ in $G$.

The map $T$ intertwines the two representations. Consider any group element $h$

$T r(h)r(g)v_0 = T r(hg)v_0 = p(hg)v_0 = p(h)p(g)v_0 = p(h) T r(g)v_0$

Because the representations are irreducible, the map is defined for all of $V$ (i.e. $r(g)v_0$ includes all of $V$, for different $g$ in $G$), and is surjective onto all of $W$. By Schur's lemma, a surjective map between irreducible representations must be an isomorphism.

1

There are 1 best solutions below

1
On

The best thing to do when confronted with something that seems absurd is to look at an example and see what happens.
Consider $G=S_3$, consider $V$ = sign representation, and $W$ the trivial representation (so they are both one-dimensional and irreducible).
You can immediately see that your first requirement fails here: any vector is an eigenvector for $r(g)$ (which is either the identity or multiplication by $-1$), so you cannot select your $v$ (this means that you'd have to take out the "all irreducible representations" above, maybe you meant "all irreducible representation of dimension more than one"?).

Moreover, I assume that you meant "is not an eigenvector for any $g \neq 1$", since any nonzero vector is an eigenvector for the identity matrix $r(1)$.

Now consider $G=S_4$. Pick $V$ as the degree two irreducible representation, the one in which the normal $V_4$ acts trivially, and the quotient (isomorphic to $S_3$) acts as the unique $2$-dimensional irreducible representation of $S_3$. Here we have a kernel! So there are other elements that act as the identity, therefore, again, there is no such $v_0$.

We could now go search for "all irreducible faithful representations of dimension more than one"... and we would get a problem when we notice that $T$ cannot be extended to all $g \in G$. (what is $w_0$, by the way? Same as $v_0$ but for $W$?) but before elaborating on that, are you sure this is what you wanted?