# Decomposable Tensors over Rings

Suppose $R$ is a commutative ring and $M$ is a $R$-module. Then we can define the tensor product $M\otimes_R M$ and more generally the $k$-fold tensor powers $\otimes_R^kM$ for any $k\in\mathbb{N}$, which are again $R$-modules.

In addition the tensor power $T_R(M):=R\oplus (\oplus_{k\in\mathbb{N}}\otimes_R^{k}M)$ is an $R$-algebra with respect to the concatenation product $\otimes: T_R(M)\times T_R(M)\to T_R(M)\;;\;(x,y)\mapsto x\otimes_R y$.

Now the question is:

Under what additional assumptions on $R$ or on the module $M$ can we say that any tensor $x\in T_R(M)$ is a $R$-linear combination of decomposable tensors, that is $x=\sum_{i\in I}a_i(x_{i,1}\otimes_R\cdots \otimes_R x_{i,n_i})$ for some index set $I$ and $x_{i,j}\in M$?

-

The tensor algebra is a direct sum, so any element is a sum of 'homogeneous' tensors $x\in \otimes^n_R M$.Ok. But the second point needs more explanation. Is it a linear combination of simple tensors, i.e a finite sum? And why? A proof directly from the definition would be nice. .. If this is obvious, a proof should be easy. – Mark Neuhaus Oct 7 '13 at 7:41
I cannot edit your question for so few characters, but in you expression of the tensor power $T_R(M):=R\oplus (\oplus_{k\in\mathbb{N}}\otimes_R^{k}M)$, you have two times the "unit sector". It should be changed to $T_R(M):=(\oplus_{k\in\mathbb{N}}\otimes_R^{k}M)$ or $T_R(M):=R\oplus (\oplus_{k\geq 1}\otimes_R^{k}M)$ – Duchamp Gérard H. E. Mar 25 '15 at 5:45