Clarification on definition of representation of $\mathrm{Hom}(V,W)$

definitiondual-spacesfinite-groupsrepresentation-theorytensor-products

From Representation Theory by W. Fulton and J. Harris:

Let $V$ be a finite dimensional vector space, and $G$ a finite group.

Let $\rho: G \to \mathrm{GL}(V)$ be a representation of $V$. The dual of this representation on the dual space $V^*$ is $G\to \mathrm{GL}(V^*)$ given by $g \mapsto \Big( f \mapsto f\circ \rho(g^{-1})\Big)$.


Having defined the dual of a representation and the tensor product of two representations, it is likewise the case that if $V$ and $W$ are representations, then $\mathrm{Hom}(V, W)$ is also a representation, via the identification $\mathrm{Hom}(V, W) = V^* \otimes W$. Unraveling this, if we view an element of $\mathrm{Hom}(V, W)$ as a linear map $\varphi$ from $V$ to $W$, we have $(g \varphi)(v)=g\varphi(g^{-1}v)$.

The isomorphism $V^* \otimes W \to \mathrm{Hom}(V, W)$ is given by $f \otimes w \mapsto \Big(v \mapsto f(v)w\Big)$.

Taking into account the $G$ action we have, $g\cdot (f\otimes w) = g\cdot f \otimes g\cdot w = f\circ \rho(g^{-1}) \otimes \rho(g)(w)$ and
$$f\circ \rho(g^{-1}) \otimes \rho(g)w \mapsto \Big(v \mapsto f\circ \rho(g^{-1})(v)\cdot \rho(g)(w)\Big).$$

So, how do we have $(g \varphi)(v)=g\varphi(g^{-1}v)$?

Best Answer

Let us denote the given isomorphism from $V^* ⊗ W$ to $\operatorname{Hom}(V, W)$ by $Ψ$. You have already computed that the induced action of $G$ on $V^* ⊗ W$ is given by $$ g ⋅ (f ⊗ w) = ( f ∘ ρ(g^{-1}) ) ⊗ ρ(g)(w) \,, $$ and that therefore, $$ ψ(g ⋅ (f ⊗ w))(v) = (f ∘ ρ(g^{-1}))(v) ⋅ ρ(g)(w) . $$ In other words, $$ \tag{$\ast$} Ψ(g ⋅ (f ⊗ w))(v) = f(g^{-1} v) ⋅ (g w) . $$ We note that in this formula, the multiplication sign on the right-hand side refers the scalar multiplication of the vector space $W$.

The right-hand side of $(\ast)$ tells us how $(g φ)(v)$ is supposed to look like in the special case of $φ = Ψ(f ⊗ w)$. To get from this a formula that works for an arbitrary element of $\operatorname{Hom}(V, W)$, we need the express the right-hand side of $(\ast\ast)$ purely in terms of $φ$, $g$ and $v$, and without using $f$ or $w$.

The map $φ = Ψ(f ⊗ w)$ is given by the formula $$ \tag{$\ast\ast$} φ(v) = f(v) ⋅ w \,. $$ By comparing the right-hand side of $(\ast\ast)$ with the right-hand side of $(\ast)$, we see that this right-hand side of $(\ast)$ can be sligthly rewritten as $$ g (f(g^{-1} v) ⋅ w) $$ because the term $f(g^{-1} v)$ is a scalar, and therefore commutes with the action of $g$ on $W$. This term can now be rewritten as $$ g φ(g^{-1} v) $$ depending only on $φ$ and $g$.

Related Question