Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

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

Let $A$ be a smooth Hopf Algebra over a field $k$ with comultiplication map $\Delta$ and Augmentation Ideal $I$.

Then I can regard the composition of $\Delta$ with the natural projection of $A \otimes A$ to the square of $I$, i.e. the map

$A \rightarrow A \otimes_k A \rightarrow A/I^2 \otimes_k A/I^2$.

Now my questions:

(1) Can one say what this map does with an element $x$ of $A$ explicitly? In particular, what is it's kernel? I'm not quite sure if one can really say something in this general setting...

(2) Is $A/I^2 \otimes_k A/I^2$ naturally isomorphic as $k-$algebra to something more familiar? Can it be that it is $(A\otimes_kA)/I\otimes_kI$? Or what else? And how does the above map then look like?

Thanks a lot for your effort!

share|cite|improve this question
2) I think the formula you're looking for is $A/I\otimes B/J = (A\otimes B)/(I\otimes B + A \otimes J)$. I would advise to look at a simple example like $A=k[x]$. – AFK Jan 12 '12 at 10:44
In that example you get as kernel $I^3$, but I don't know if one can generalize this... – Cyril Jan 14 '12 at 11:57
Elementary questions are not necessarily easy, I guess. – HbCwiRoJDp Jan 14 '12 at 13:19
But easily formulated...;) – Cyril Jan 14 '12 at 19:33
up vote 3 down vote accepted

(1) The kernel is $I^3$ if $\mathrm{char}(k)\ne 2$, and is $I^2$ otherwise.

Denote by $\delta: A\to A\otimes A \to A/I^2\otimes A/I^2$. The first observation is $A/I^2\otimes_k A/I^2=(A\otimes_k A)/(I^2\otimes A+ A\otimes I^2)$. See YBL's comments. So the kernel of $\delta$ is $\Delta^{-1}(I^2\otimes A+ A\otimes I^2)$.

Let's describe $\Delta(x)$ for $x\in A$. Note that $A=k+I$. So $$A\otimes_k A=k+k\otimes_k I+ I\otimes_k k+ I\otimes_k I.$$
Decompose $\Delta(x)=\lambda+1\otimes a+ b\otimes 1+ \sum_i u_i\otimes v_i$ as above. By the definition of $I$, the composition $A\to A\otimes A\to A/I \otimes A =A$, $x\mapsto \Delta(x)\mapsto \lambda+ a$ is identity. Hence $a=x-\lambda=b$ (by symmetry). From now on, suppose $x\in I$. Then $\lambda=0$ and $$\Delta(x)=1\otimes x + x\otimes 1 + \sum_i u_i\otimes v_i.$$ This implies that $$\Delta(I^2)\subseteq I\otimes I + I^2\otimes A + A\otimes I^2$$ and $$\Delta(I^3)\subseteq (k\otimes I + I\otimes k + I\otimes I)^3 \subseteq A\otimes I^2 + I^2\otimes A.$$ On the other hand, again considering $A\to A\otimes A \to A/I \otimes A$, we see that if $x\in\ker \delta$, then $x\in I^2$. So $$ I^3\subseteq \ker \delta \subseteq I^2.$$

The above inclusions imply that $\Delta$ induces a map $$ \tilde{\Delta}: I^2/I^3 \to I/I^2\otimes I/I^2 $$ and we have $\ker\delta/I^3=\ker\tilde{\Delta}$. Now consider $\phi : I/I^2\otimes I/I^2 \to I^2/I^3$ induced by $A\otimes A\to A$, $u\otimes v\mapsto uv$. Let's show that $\phi\circ \tilde{\Delta}=4.\mathrm{Id}$. Let $y\in I$ and $\Delta(y)=1\otimes y + y\otimes 1 + \sum_jw_j\otimes z_j$ avec $w_j, z_j\in I$. A direct computation shows that $$\Delta(xy) = x\otimes y + y\otimes x + 1\otimes (xy)+ (xy)\otimes 1+ t, \quad t\in I^2\otimes I+ I\otimes I^2.$$ This implies that Then $\phi\circ \tilde{\Delta}(\overline{xy})=4\overline{xy}$ and the same equality for all $z\in I^2$.

End of the proof. If $\mathrm{char}(k)\ne 2$ and $z\in \ker\delta$. Then $4\bar{z}=\phi\circ\tilde{\Delta}(\bar{z})/4=0$ in $I^2/I^3$. So $z\in I^3$ and $\ker\delta =I^3$.

Suppose $\mathrm{char}(k)=2$. Let $x\in I$. Then $$\Delta(x^2)=\Delta(x)^2\in 2(x\otimes x) + I^2\otimes A + A \otimes I^2=I^2\otimes A + A\otimes I^2.$$ As $I^2$ is generated by the $x^2$, $x\in I$ (in characteristic $2$ !), we have $I^2\subseteq \ker\delta\subseteq I^2$ and we are done.

This is a little complicated, there should be a more direct proof using Hopf algebra properties...

(2) I don't have much to say. The tensor product $A/I^2\otimes A/I^2$ is a local Artinian $k$-algebra whose vector dimension is $(1+d)^2$ where $d=\dim I/I^2$ is the dimension of the tangent space of $\mathrm{Spec} A$ at the origin. Except when $A$ is Artinian, $A/I^2\otimes A/I^2$ is not isomorphic to $(A\otimes A)/(I\otimes I)$ because the latter is not Artinian.

share|cite|improve this answer
Applause for this! I didn't think one could answer it in such general form. You helped me a lot, thanks! – Cyril Jan 15 '12 at 18:21

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.