3
$\begingroup$

The category $\mathcal{V}_k$ of vector spaces, over a field $k$, can be considered as a symmetric, closed monoidal category with respect to the tensor product of vector spaces, the vector space of linear maps as internal hom $[V,W]$ and the symmetry $ \tau: x \otimes y = y \otimes x$.

In addition we can see $\mathcal{V}_k$ as enriched over itself and from the tensor $\leftrightarrow$ hom adjunction we get the following natural inclusion

$[V_1,W_1]\otimes [V_2,W_2] \to [V_1 \otimes V_2, W_1 \otimes W_2]$ $\quad\quad(A)$

which corresponds under the tensor - hom ajunction to a preimage of the identity in

$hom([V_1,W_1]\otimes [V_2,W_2], [V_1 \otimes V_2, W_1\otimes W_2])\simeq\\ hom([V_1,W_1]\otimes [V_2,W_2]\otimes V_1\otimes V_2, W_1\otimes W_2)\simeq\\ hom(([V_1,W_1]\otimes V_1) \otimes ([V_2,W_2]\otimes V_2), W_1\otimes W_2)\to\\ hom(W_1\otimes W_2, W_1 \otimes W_2)$

where we used the symmetry and the natural evaluation $[V,W]\otimes V\to W$. In the most right hom set, there is of course the natural choice $id_{W_1\otimes W_2}$.

Now the question: In case we restrict to finite dimensional vector spaces. Is $(A)$ an isomorphism? If yes, how can we compute the inverse?

$\endgroup$
1
$\begingroup$

The definition of "finite dimensional" is "has a finite basis". So to answer this problem we will definitely have to pick bases. Let $(e^1_i)$, $(e^2_i)$, $(g^1_i)$ and $(g^2_i)$ be bases of $V_1$, $V_2$, $W_1$ and $W_2$, and let the dual bases be $(f^1_i)$, $(f^2_i)$, $(h^1_i)$ and $(h^2_i)$. Then $[V_1,W_1]\otimes[V_2,W_2]$ has a basis given by $(f^1_i\otimes g^1_j\otimes f^2_k\otimes g^2_l)$ and $[V_1\otimes V_2,W_1\otimes W_2]$ has a bais given by $(f^1_i\otimes f^2_j\otimes g^1_k\otimes g^2_l)$. The inverse isomorphism is the one that takes $f^1_i\otimes f^2_j\otimes g^1_k\otimes g^2_l$ to $f^1_i\otimes g^1_k\otimes f^2_j\otimes g^2_l$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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