lecture 7

From Scholarpedia

< An introduction to Lie algebra cohomology
This article has not been peer-reviewed or accepted for publication yet; It may be unfinished, contain inaccuracies, or unapproved changes.

Author: Dr. Jan A. Sanders, Vrije Universiteit Amsterdam
Author: Dr. Sara Lombardo, Vrije Universiteit Amsterdam

Back to the sixth lecture

On to the eigth lecture

Contents

Spectral sequences

In the examples, one uses K^{p,n}=F^pC^n(\mathfrak{g},\mathfrak{a}), but in the construction of the spectral sequence one only assumes K^{p,n}\supset K^{p+1,n} and d^n K^{p,n}\subset K^{p,n+1}.

definition

Let

Z_p^{r,n}=K^{p,n} for r\leq 0,
Z_p^{r,n}=\{a^n\in K^{p,n}| d^n a^n\in K^{p+r,n+1}\} for r>0.

remark

If p>n then Z_p^{r,n}=0.

proposition

Z_0^{1,0}=\mathfrak{a}^\mathfrak{h}=H^0(\mathfrak{g},\mathfrak{a}), where \mathfrak{a}^\mathfrak{h} denote the \mathfrak{h}-invariant elements (under d_-^{(0)}) in \mathfrak{a}.

proof

a^0\in Z_0^{1,0} implies d^0a^0 \in F^1 C^1(\mathfrak{g},\mathfrak{a}), that is, for y\in\mathfrak{h} one has

d_-^{(0)}(y)a^0=d^0a^0(y)=0.

proposition

Z_1^{1,1} \subset F^1 C^1(\mathfrak{g},\mathfrak{a}^\mathfrak{h})

proof

a^1\in Z_1^{1,1} implies that for x\in\mathfrak{g}, y\in \mathfrak{h} one has, since d^1 a^1\in F^2 C^2(\mathfrak{g},\mathfrak{a}),

0=d^1a^1(x,y)=d^{(0)}(x)a^1(y)-d^{(0)}(y)a^1(x)-a^1([x,y])=-d^{(0)}(y)a^1(x).

proposition

Z_{p+1}^{r-1,n}\subset Z_p^{r,n}

proof

Let a^n\in Z_{p+1}^{r-1,n}. Then a^n\in K^{p+1,n}\subset K^{p,n} and d^n a^n \in K^{p+r,n+1}. But this immediately implies that a^n\in Z_p^{r,n}.

proposition

d^{n-1} Z_{p-r+1}^{r-1,n-1}\subset Z_p^{r,n}

proof

Let a^{n}\in d^{n-1} Z_{p-r+1}^{r-1,n-1}. Then one can write a^n as d^{n-1}a^{n-1}, with a^{n-1}\in Z_{p-r+1}^{r-1,n-1}, that is to say, a^{n-1}\in K^{p-r+1,n-1} and a^n=d^{n-1}a^{n-1}\in K^{p,n}. Since d^na^n=d^n d^{n-1}a^{n-1}=0, it follows that a^n\in Z_p^{r,n}.

definition

E_p^{r,n}=Z_p^{r,n}/(d^{n-1} Z_{p-r+1}^{r-1,n-1}+Z_{p+1}^{r-1,n})

remark

In particular, E_p^{0,n}=Z_p^{0,n}/Z_{p+1}^{0,n}, that is, the graded version of the filtered sequence K^{p,n}.

theorem

On E_\cdot^{r,n} there is an induced coboundary operator \delta_r^1 such that

H^p(E_\cdot^{r,n},\delta_r^1)=E_p^{r+1,n}

This means that E_\cdot^{r,n} is a spectral sequence.

proof

d^n maps Z_p^{r,n} to Z_{p+r}^{r,n+1} and d^{n-1} Z_{p-r+1}^{r-1,n-1}+Z_{p+1}^{r-1,n} to d^{n} Z_{p+1}^{r-1,n}. Let [a^n]\in E_p^{r,n}. Define

\delta_r^1[a^n]=[d^n a^n]

One has [d^n a^n]\in E_{p+r}^{r,n+1}. Suppose [a^n] is a cocycle. This means that d^na^n\in d^{n} Z_{p+1}^{r-1,n}+Z_{p+r+1}^{r-1,n+1}. That is, there exist \tilde{a}^n \in Z_{p+1}^{r-1,n} and a^{n+1}\in Z_{p+r+1}^{r-1,n+1} such that

d^n a^n=d^n \tilde{a}^n+a^{n+1}.

Let \bar{a}^n=a^n-\tilde{a}^n\in Z_p^{r,n}+Z_{p+1}^{r-1,n} \subset K^{p,n}, with d^n \bar{a}^n=a^{n+1}\in Z_{p+r+1}^{r-1,n+1}\in K^{p+r+1,n+1}. Therefore \bar{a}^n \in Z_{p}^{r+1,n}. This implies that a^n=\bar{a}^n+\tilde{a}^n\in Z_{p}^{r+1,n}+Z_{p+1}^{r-1,n}.

It follows that

Z^p(E_\cdot^{r,n},\delta_r^1)=(Z_{p}^{r+1,n}+Z_{p+1}^{r-1,n})/(d^{n-1} Z_{p-r+1}^{r-1,n-1}+Z_{p+1}^{r-1,n})

The p-coboundaries consist of the elements of d^{n-1} Z_{p-r}^{r,n-1}, and one has

B^p(E_\cdot^{r,n},\delta_r^1)=(d^{n-1} Z_{p-r}^{r,n-1}+Z_{p+1}^{r-1,n})/(d^{n-1} Z_{p-r+1}^{r-1,n-1}+Z_{p+1}^{r-1,n})

Noether isomorphism

If W\subset U then

U/(W+U\cap V)\simeq (U+V)/(W+V) and (M/V)/(U/V)=M/U.

It follows that

H^p(E_\cdot^{r,n},\delta_r^1)=(Z_{p}^{r+1,n}+Z_{p+1}^{r-1,n})/(d^{n-1} Z_{p-r}^{r,n-1}+Z_{p+1}^{r-1,n})

proposition

d^{n-1} Z_{p-r}^{r,n-1}\subset Z_p^{r+1,n}.

proof

Let a^n \in  d^{n-1} Z_{p-r}^{r,n-1}. Then a^n=d^{n-1} a^{n-1} with a^{n-1}\in Z_{p-r}^{r,n-1}. Therefore a^n\in Z_p^{r+1,n}, since d^n a^n=0.\square

It follows that

H^p(E_\cdot^{r,n},\delta_r^1)=Z_{p}^{r+1,n}/(d^{n-1} Z_{p-r}^{r,n-1}+Z_{p}^{r+1,n}\cap Z_{p+1}^{r-1,n})

proposition

Z_p^{r+1,n}\cap Z_{p+1}^{r-1,n}=Z_{p+1}^{r,n}.

proof

Let a^n\in Z_p^{r+1,n}\cap Z_{p+1}^{r-1,n}. Then a^n\in K^{p+1,n} and d^n a^n \in F^{p+r+1} C^{n+1}(\mathfrak{g},\mathfrak{a}). This implies a^n\in Z_{p+1}^{r,n}. On the other hand, if a^n\in Z_{p+1}^{r,n}, we have a^n\in K^{p+1,n}\subset F^p C^n(\mathfrak{g},\mathfrak{a}) and d^na^n \in K^{p+r+1,n+1}\subset K^{p+r,n+1}. Thus, a^n \in K^{p,n} and d^na^n \in K^{p+r+1,n+1}, implying a^n\in Z_p^{r+1,n}.

Furthermore, a^n \in K^{p+1,n} and d^na^n \in K^{p+r,n+1}, implying a^n\in Z_{p+1}^{r-1,n}. The result follows. \square

corollary

H^p(E_\cdot^{r,n},\delta_r^1)=Z_{p}^{r+1,n}/(d^{n-1} Z_{p-r}^{r,n-1}+Z_{p+1}^{r,n})=E_p^{r+1,n}

This proves the theorem.



Back to the sixth lecture

On to the eigth lecture

For authors