7

Let $L \supset K$ be a finite extension of fields. The diagonal $L \otimes_K L$ we can endow with structure of $L$ algebra via $L \to L \otimes_K L,\ l \mapsto l \otimes 1_L$. Especially $L \otimes_K L$ carries structure of a $L$-module via $L$-multiplication in first factor $l \cdot (a \otimes b) := la \otimes b$.

Let $d: L \otimes_K L \to L, a \otimes b \mapsto ab $ be the diagonal map and $I$ its kernel.

I want to show that $L \otimes_K L$ is reduced (has no nilpotent elements) iff $I = I^2$, i.e. $I/I^2=0$.

Ideas: We can simplify the problem if we choose an more accessible system of generators for $L$-module $I/I^2$. I claim that $a \otimes 1_L -1_L \otimes a$ for $a \in L$ generate $I/I^2$ as $L$-module.

That's because $a \otimes b = ab \otimes 1 - a \cdot (b \otimes 1 - 1 \otimes b)$ (recall $I$ carries $L$-module structure by multiplication in first factor) and therefore for $\sum a_i \otimes b_i \in I$ we obtain by linearity

$$\sum a_i \otimes b_i = (\sum a_i b_i) \otimes 1 - \sum a_i \cdot (b_i \otimes 1 - 1 \otimes b_i)= \sum a_i \cdot (b_i \otimes 1 - 1 \otimes b_i)$$

Therefore it suffice to show that $L \otimes_K L$ is reduced iff every $a \otimes 1 - 1 \otimes a, a \in L$ is modulo $I^2$ a product of two other such $x \otimes 1 - 1 \otimes x, y \otimes 1 - 1 \otimes y, x, y \in L$.

How can I do it?
Note: If we realize that $I/I^2$ can be interpreted as module of Kahler differentials $\Omega_L$ and we use another more usual construction of $\Omega_L$ as certain quotient of some free $\bigoplus_iK[x_1,..., x_n] dx_i$ use some standard exact sequences between Kahler differentials then the claim follows immediately. But I want to know if it possible to show the claim directly(!) working only with definition of $I/I^2$ instead of making a detour over quoting results known for Kahler differntials.

hm2020
  • 1
user267839
  • 7,293
  • I'm not sure, but I think that both conditions equivalent to L/K being separable. – sss89 Aug 15 '21 at 20:26
  • "a product of two other" - no, it may be a sum of these as well. Generally speaking, I suggest to look at $I/I^2$ as $I \otimes_{L \otimes L} L$. And element calculations will not lead to anything. Try to think about formal properties of the tensor product. – Martin Brandenburg Aug 15 '21 at 21:29
  • With the approach using differentials, "then the claim follows immediately" - how? Why is $\Omega^1_L=0 \iff L \otimes_K L \text{ reduced}$ immediate? Don't we need separability as an intermediate step? – Martin Brandenburg Aug 15 '21 at 21:43
  • What definitely seems to be immedately true is as sss98 remarked the equivalence between '$L/K$ separable' and '$L \otimes_K L$ reduced': if $L/K$ separable, then by primitive element thm $L=K(a)= K[X]/f$ and therefore $L \otimes_K L = L[X]/f= \prod_i L[X]/f_i$ for irreducible decomposition $f= \prod_i f_i$ in $L[X]$. Then $f$ separable iff no factor $f_i$ has form $h^d, h \in L[X], d >2$. For the other direction assume $L \otimes_K L$ reduced and $a \in L$. – user267839 Aug 16 '21 at 13:00
  • Then $K(a)= K[X]/f_a, f_a$ MP of $a$ and $K(a) \otimes_K L = L[X]/f= \prod_i L[X]/f_i \subset L \otimes_K L$ is subring of a reduced ring and therefore reduced. Therefore as before $f_a$ separable and so $a$. – user267839 Aug 16 '21 at 13:00
  • @MartinBrandenburg: On your question why $\Omega^1_L=0 \iff L \otimes_K L \text{ reduced}$ is 'immediate'. By above we can replce the left hand side by '$L/K$ separable'. So you're right, we need separability. So by PET $L= K(a)= K[X]/f, f$ MP. Now we come to the core of my problem. As I noticed if we deal with exact sequence on differentials relating $\Omega^1_{L/K}$ to $ \Omega^1_{K[X]/K}= K[X] \cdot dx$ then we immediately see that $\Omega^1_{L/K}= K[X] \cdot dx/(f, df)$. But $df= f'(a)dx$ and $f'(a) \neq 0$ due to separability, therefore $dx=0$ in $\Omega^1_{L/K}$. – user267839 Aug 16 '21 at 13:14
  • therefre $\Omega^1_L =0$. BUT the only really nontrivial step for me which requires effort is to identify $\Omega^1_L$ with $I/I^2$. Of couse it is done in every textbook on differentials but it is a 'non-triviality' and the motivation of my question was if it is possible to show directly that $L \otimes L$ reduced (resp. that $L/K$ seprable; these two I consider by above as 'immediatelly' equivalnt to each other) is equivalent to $I =I^2$ directly on studing nilpotent elements in $I$ without doing this 'hard' piece of work to identify of $I/I^2$ and $K[X] \cdot dx/(f, df)$. – user267839 Aug 16 '21 at 13:25
  • @MartinBrandenburg: You also suggested to think of $I/I^2$ as $I \otimes_{L \otimes L} L$. I guess you use there implicitely that $L \otimes L/I \cong L$ because first of all we know only that $I/I^2 \cong I \otimes_{L \otimes L} (L \otimes_K L/I)$? – user267839 Aug 16 '21 at 13:44
  • @Isak the XI, If $L/K$ is separable then similar to your comment $L\otimes L = \prod L[X]/f_i$ and the map $L\otimes L\to L$ is theprojection on one of the factors. We get that $I^2=I$. – sss89 Aug 16 '21 at 17:24
  • @sss89: Why if $L/K$ is separable $L\otimes L\to L$ is the projection on one of the factors? Which factors? You mean $L[X]/f_i$? But $L[X]/f_i \neq L$. And why this would imply $I=I^2$? – user267839 Aug 16 '21 at 21:10
  • 1
    @Isak the XI: Since $L=K[X]/f$, $L$ contains a root of $f$, meaning that $f=(X-a)g\in L[X]$ where $g(a)\ne 0$ by separability. By the CRT, $ L\otimes L = L[X]/f = L[X]/(X-a) \times L[X]/g= L\times L[X]/g$. The map $L\otimes L\to L$ sends $X$ to $a$ which is identity on the first factor and annihilates $(0,1)$ since $gcd(g, X-a)=1$. – sss89 Aug 17 '21 at 05:02
  • I see. Then by construction ${0 } \times L[X]/g = I$. But why this implies $I=I^2$? – user267839 Aug 17 '21 at 07:15
  • 1
    @IsaktheXI: $I$ is generated by the single idempotent element $(0,1)$. – sss89 Aug 17 '21 at 10:53
  • Yes, I understand, so this proves the direction $L/K$ sepatated implies $I=I^2$. How can we show the opposite direction. I'm not such if $I=I^2$ implies immediately that $I$ is generated by idemponent elements. Do you see an argument? – user267839 Aug 17 '21 at 11:47
  • @IsaktheXI - you can also use the chinese remainder lemma - see my attached post. – hm2020 Aug 20 '21 at 06:47

1 Answers1

2

Question: "How can we show the opposite direction."

Answer: Assume $k \subseteq l:=k\{e_1,..,e_n\}$ is finite and let $K$ be the algebraic closure of $k$ with $L:=K\otimes_k l$. There is an inclusion

$$l\otimes_k l \subseteq L \otimes_k L$$

and isomorphisms $l\otimes_k l:=A \cong k\{e_i\otimes e_j\}$ and $L\otimes_K L:=B \cong K\{e_i\otimes e_j\}$. Hence $l\otimes_k l$ is a finite dimensional $k$-algebra and $L\otimes_k L$ is a finite dimensional $K$-algebra. If $I/I^2\cong \Omega^1_{l/k}=0$ it follows $\Omega^1_{B/K}=0$. If $nil(A) \neq (0)$ there is a maximal ideal $(0)\neq m \subseteq B$ with $B/m \cong K$. There is an isomorphism

$$\Omega^1_{B/K}\otimes_B B/m \cong m/m^2=0$$

hence $m^n=m$ for all $n \geq 2$. Since $dim_K(B)<\infty$ it follows $krdim(B)=0$ hence there is a direct sum decomposition

$$B\cong B_1\oplus \cdots \oplus B_d$$

with $B_i$ artinian $K$-algebra with maximal ideal $m_i \subseteq B_i$. Since $\Omega^1_{B/K}=0$ it follows $\Omega^1_{B_i/K}=0$ hence

$$\Omega^1_{B_i/K}\otimes_{B_i} B_i/m_i \cong m_i/m_i^{2}=0$$

hence $m_i^n=m_i$ for all $n\geq 2$. Since $dim_K(B_i)< \infty$ there is an $N_i$ with $(0)=m_i^{N_i}=m_i=0$. Hence $B_i$ is a field and hence $L\otimes_k L$ is a direct sum of fields, hence $L\otimes_k L$ is reduced. There is an inclusion

$$l\otimes_k l \subseteq L\otimes_K L$$

and it follows $l\otimes_k l$ is reduced - a contradiction, hence $nil(A)=(0)$.

Hint: $K\otimes_k \Omega^1_{l/k} \cong \Omega^1_{L/K}$ and $\Omega^1_{B/K} \cong L\otimes_K \Omega^1_{L/K}\oplus \Omega^1_{L/K} \otimes_K L$. Since $B:=L\otimes_K L$ is a finite dimensional non-reduced algebra, $K$-algebra there is a maximal ideal $m \subseteq B$ and since $K$ is algebraically closed it follows $B/m \cong K$.

Example: If $k:=\mathbb{Q}$ and $K:=k(\sqrt{2})$ it follows $k \subseteq K$ is separable and $\Omega^1_{K/k}=0$. You may check explicitly that

$$K\otimes_k K \cong K \oplus K$$

is a direct sum of fields, hence $K\otimes_k K$ is reduced.

Conversely: If $A$ is reduced we may argue as follows: There is a direct sum decomposition

$$A \cong A_1 \oplus \cdots \oplus A_d$$

with $A_i$ artinian, and since $A$ is reduced it follows $A_i$ is a field for all $i$. Since $I\subseteq A$ is a maximal ideal it follows $I:=A_1\oplus \cdots \oplus (0) \oplus \cdots \oplus A_d$ and hence for any element $a\in I$ it follows the element $xa=a \in I^2$ where

$$x:=(1,..,1,0,1,..,1)\in I.$$

Hence $I^2=I$. We have proved:

Lemma: Let $k \subseteq l$ be a finite extension of fields. It follows $\Omega^1_{l/k}=0$ iff $l\otimes_k l$ is reduced.

hm2020
  • 1
  • Why $\Omega^1_{l/k}=0$ implies $\Omega^1_{B/K}=0$? – user267839 Aug 17 '21 at 19:25
  • Also why if $nil(A) \neq (0)$ there is a maximal ideal $m \subseteq B$ with $B/m \cong K$? That is why the condition $nil(A) \neq (0)$ is neccessary(!) for the existience of such maximal ideal $m \subseteq B$? – user267839 Aug 17 '21 at 19:28
  • towards your Hints: The argument why $\Omega^1_{B/K}=0$ I understand now. But I still not understand why we need the assumption $nil(A) \neq (0)$ (ie that $B$ is not reduced) to assure the extence of maximal ideal $m \subseteq B$ with $B/m=K$? Or in other words why such ideal could not exist if $B$ would be reduced? – user267839 Aug 18 '21 at 14:35
  • @IsaktheXI - I added an elementary proof of the Lemma. – hm2020 Aug 20 '21 at 06:18
  • Alright, thanks. I think I have also found the answer to my second question: if $I:=nil(A) \neq (0)$, then the ideal $m \subseteq B$ with $B/m=K$ you consider in the proof is exactly that one which should additionally satisfy $I \subset m$, right? (since otherwise the assumption that $nil(A) \neq (0)$ is not neccessary for existence of such ideal $m$; we can always find an maximal ideal $m \subseteq B$ with $B/m=K$, since $K$ alg closed) – user267839 Aug 21 '21 at 11:31