Let $K$ be a commutative ring. Is it true that $T(V \oplus W) \cong T(V) \otimes_K T(W)$ as $K$-algebras for any $K$-modules $V$ and $W$?

The reason I ask is that I have heard it mentioned (I don't remember where) that the canonical comultiplication on the tensor algebra (which gives it a Hopf algebra structure) is the image of the diagonal $\Delta: V \to V \oplus V$ by the tensor algebra functor. In order for this to be true, we would need $T(\Delta): T(V) \to T(V \oplus V) \cong T(V) \otimes_K T(V)$. Also, for Lie $K$-algebras I know that $U(\mathfrak{g} \times \mathfrak{h}) \cong U(\mathfrak{g}) \otimes_K U(\mathfrak{h})$ so I would hope that this isomorphism is inherited from a similar isomorphism of Tensor algebras.

But I have some doubts that this is true. The tensor algebra functor $T: K \text{Mod} \to K\text{Alg}$ is left adjoint to the forgetful functor, and thus preserves coproducts, so $T(V \oplus W) \cong T(V) + T(W)$ where the right hand side is the coproduct of algebras. But then we would need $T(V) \otimes_K T(W) \cong T(V) + T(W)$ as $K$-algebras in order for this to work. But the coproduct of $K$-algebras is the tensor product if we are working in the category of commutative $K$-algebras; otherwise the coproduct is more complicated.


This is indeed not true. As you observe, $T(V\oplus W)$ is the coproduct of $T(V)$ and $T(W)$ in the category of $K$-algebras, which is typically larger than $T(V)\otimes_K T(W)$.

For a very simple example, take $V=W=K$. Then $T(V)$ and $T(W)$ are both polynomial rings in one variable over $K$; write $T(V)=K[x]$ and $T(W)=K[y]$. Then $T(V)\otimes_K T(W)\cong K[x,y]$, but $T(V\oplus W)\cong K\langle x,y\rangle$, the ring of noncommutative polynomials in two variables.

However, there still is a canonical ring-homomorphism $T(V\oplus W)\to T(V)\otimes_K T(W)$ given by the universal property of the coproduct. So the diagonal map $\Delta:V\to V\oplus V$ induces a map $T(V)\to T(V\oplus V)$ which can then be composed with this map to give a map $T(V)\to T(V)\otimes T(V)$. This is the comultiplication of the tensor algebra.

  • 1
    $\begingroup$ The "obvious" homomorphism comes rather from the universal property of the free algebra, no? $\endgroup$ – Mariano Suárez-Álvarez Jul 16 '16 at 4:47
  • $\begingroup$ Sure, that's a more direct way to get it. $\endgroup$ – Eric Wofsey Jul 16 '16 at 4:59

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.