MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This might be inappropriate for the MO-level. If so I'll delete it...

Suppose $V$ is a $\mathbb{Z}$-graded vector space and $\overline{T}(V):=V \oplus V\otimes V \oplus \otimes^3 V \ldots$ is the 'reduced tensor power' of $V$ in the graded sense.i.e:

$V\otimes V:=\oplus_{z\in\mathbb{Z}}\oplus_{p+q=z}V_p\otimes V_q$

We don't see $T(V)$ as a (co)algebra, just as a plain graded vector space.

Then on a first "empirical observation" I would say, that if we apply the reduced tensor functor again, i.e make $\overline{T}(\overline{T}(V))$ then this is naturally isomorphic to $\overline{T}(V)$ because "all tensor powers are already there".

So this should be a consequence of the 'freeness' of $\overline{T}(V)$, but I can't see how this can be proofen.

share|cite|improve this question
Assume that each homogeneous component of $V$ is free of finite rank over $\mathbb Z$. Assume also that the $0$-th homogeneous component of $V$ is $0$. Then, if $H\left(W\right)$ denotes the Hilbert series of a graded $\mathbb Z$-module $W$ whose homogeneous components are free, then we have $H\left(\overline T\left(V\right)\right) = \frac{H\left(V\right)}{1-H\left(V\right)}$. From this, it should be pretty easy to see that in general, $H\left(\overline T\left(V\right)\right) \neq H\left(\overline T\left(\overline T\left(V\right)\right)\right)$. – darij grinberg Jul 22 '13 at 15:51
I'm not sure I understand this.You mean if the underlying field of the vector space is finite? In that case ok, I have to rewrite the question, because I'm only interested in $\mathbb{R}$-vector spaces. – Nevermind Jul 22 '13 at 16:11
To make darij's comment more explicit: Let $V_1 = \mathbb{R}$ and all other $V_0=0$. Then the graded pieces of $\bar{T}(V)$ have dimensions $(0,1,1,1,1,\dots)$ and the graded pieces of $\bar{T}(\bar{T}(V))$ have dimensions $(0,1,2,4,8,16, \dots)$. This simply isn't true. – David Speyer Jul 22 '13 at 17:39
ok. I see now . – Nevermind Jul 22 '13 at 18:15

If $V$ is $\mathbb Z$-graded, then $\overline{T}(V)$ is $\mathbb Z_{\ge1}\times \mathbb Z$-graded with $\overline{T}(V)_{p,q} = \bigoplus_{z_1+z_2+\dots+z_p = q}V_{z_1}\otimes V_{z_2}\otimes\dots\otimes V_{z_p}$.

$\overline{T}$ is a functor from $\mathbb Z$-graded vector spaces to $\mathbb Z_{\ge1}\times \mathbb Z$-graded algebras which is left adjoint to the forgetful functor, where one has to pay attention to the degrees ...

share|cite|improve this answer
And by the way: $\overline{T}$ is in addition a functor into the category of locally nilpotent graded coassociative coalgebras and there it is right adoint to the forgetful functor – Nevermind Jul 22 '13 at 16:06
But as I said, the graded tensor power makes $\overline{T}(V)$ into a $\mathbb{Z}$-graded vector space by $\overline{T}(V)_j:=\oplus_{k}\oplus_{p_1,\ldots,p_k=j}V_{p_1}\otimes \cdots\otimes V_{p_k}$, where each $p_h\geq 1$. – Nevermind Jul 22 '13 at 16:16
You decided to ignore the grade $k$ which exactly encodes your feeling that "all the tensor powers are already there". But in $\overline{T}\overline{T}(V)$ they are there multiple times; you then have 3 degrees and cannot reduce them to 2 correctly. – Peter Michor Jul 24 '13 at 5:17

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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