Proof that the universal first order calculus satisfies its' universal property in the noncommutative case.

54 Views Asked by At

Recently I've been reading this paper, and in the first section on (differential calculus on associative algebras) they reference a theorem from Bourbaki's Algebra I Chapter 3. I have a problem with the proof of this theorem.

We consider a commutative ring $R$, and an associative $R$-algebra $A$.

Definition. If $\Gamma$ is an $A$-bimodule, then we say a homomorphism $g:A\rightarrow \Gamma$ is a derivation of $A$ into $\Gamma$ if $g(ab)=g(a)b+ag(b)$.

Definition. A first order calculus over $A$ is a pair $(B,d)$ where $d$ is a derivation of $A$ into $B$, and $B$ is the minimal left-$A$-module containing $\operatorname{im}d$.

Then, we define an $A$-bimodule homomorphism $$ \begin{aligned} \mu:A\otimes_R A&\rightarrow A\\ \sum_{i\in I}a_i\otimes b_i&\mapsto \sum_{i\in I}a_i b_i. \\ \end{aligned} $$ We also define an $R$-linear map $$ \begin{aligned} d:A&\longrightarrow \operatorname{ker}\mu\subset A\otimes_R A\\ a&\longmapsto 1\otimes a - a\otimes 1 \end{aligned} $$and it can be shown that $d$ is a derivation of $A$ into $\operatorname{ker}\mu$ (we will call this kernel $I$ from now on).

Then the universal first order calculus over $A$ is the pair $(I,d)$.

What I would like to prove is this:

Theorem. For every first order differential calculus $(\Gamma,g)$, there exists a unique epimorphism $p:I\rightarrow \Gamma$ such that $p\circ d=g$.

The paper references Bourbaki for this theorem, but I don't really follow the proof. The first thing I saw in Bourbaki was the claim "for any $A$-bimodule homomorphism $f:I\rightarrow \Gamma$, $f\circ d$ is a derivation of $A$ into $\Gamma$" - which I don't understand at all. We would have that $$ \begin{aligned} f\circ d(ab)&=f(d(ab))\\ &=f(d(a)b+ad(b))\\ &=f(d(a)b)+f(ad(b))\\ &=f\circ d(a)f(b)+f(a)f\circ d(b) \end{aligned} $$but I don't see the logic behind this being a derivation unless we define the $A$ action using $f$.

Any help would be appreciated, thank you!

1

There are 1 best solutions below

1
On BEST ANSWER

The last line of your calculation is incorrect: $f$ is not a map of algebras, it is a map of bimodules. This means that $$ f(ax)=af(x)\;\;\;\mbox{and}\;\;\;f(xa)=f(x)a $$ for all $x\in I$ and $a\in A$. Therefore, the last line in your calculation should be $$ (f\circ d)(ab)=(f\circ d)(a)b+a(f\circ d)(b). $$