MathOverflow is a question and answer site for professional mathematicians. 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 $\frak{g}$ be a finite-dimensional complex nilpotent Lie algebra. Given $\xi\in\frak{g}$, what is known about the intersection of $im(ad_{\xi})$ (the image of $ad_{\xi}:\frak{g}\rightarrow\frak{g}$) and $z(\frak{g})$ (the centre of $\frak{g}$)? In particular, under what conditions is it true that the intersection is non-trivial?

Alternatively, under what conditions will $im(ad_{\xi})$ contain a 1-dimensional ideal?


share|cite|improve this question
up vote 2 down vote accepted

Whenever $ad_{\xi}$ is an endomorphism of $\mathfrak{g}$ whose corresponding partition $\pi: 1^{s_{1}}2^{s_{2}} \cdots \;$ of $\dim \mathfrak{g}$ is such that $s_{1} =0$, then we have $im\; ad_{\xi} \cap Z(\mathfrak{g}) \neq \{0\}$.

As $\mathfrak{g}$ is nilpotent, $ad_{\xi}$ acts as a nilpotent endomomorphism of $\mathfrak{g}$, for every $\xi \in \mathfrak{g}$, and we can associate to $ad_{\xi}$ a partition $\pi(\xi)$ of $n=\dim \mathfrak{g}$ as follows: the Jordan form of $ad_{\xi}$ is

$$J=\begin{bmatrix} J_{1}\\\ & \ddots \\\ && J_{r}\end{bmatrix}$$

where $J_{i}$ is an $n_{i}\times n_{i}$ $0$-Jordan block and $n_{1}\geq \ldots \geq n_{r} >0$. Then, $n_{1} +\ldots +n_{r}=n$ is a partition of $n$, which we will denote $\pi(\xi): 1^{s_{1}}2^{s_{2}} \cdots $, so that $$1s_{1}+2s_{2}+\ldots =n$$ is the partition of $n$ determined by $J$.

Let $\xi \in \mathfrak{g}$ and let $\pi(\xi) : 1^{s_{1}}2^{s_{2}} \cdots $ be the partition associated to $ad_{\xi}$ and suppose that $(x_{1},\ldots,x_{n})$ is a Jordan basis of $ad_{\xi}$, so that the matrix of $ad_{\xi}$ is in Jordan form. Thus, $x_{1},x_{n_{1}+1},\ldots, x_{n_{r-1}+1}$ is a basis of $$Z_{\mathfrak{g}}(\xi)=\ker ad_{\xi}.$$

As $\mathfrak{g}$ is nilpotent we must have $Z(\mathfrak{g}) \neq \{0\}$. Let $w \in Z(\mathfrak{g})$ be nonzero. Then, $w \in Z_{\mathfrak{g}}(\xi)$ and so $w= c_{1}x_{1}+\ldots + c_{r} x_{n_{r-1}+1}$.

Assume that $s_{1}=0$ in the partition $\pi$ (so that $\pi$ contains no parts equal to $1$). Thus, $x_{2},\ldots, x_{n_{r-1}+2} \notin Z_{\mathfrak{g}}(\xi)$ and $x=c_{1}x_{2}+\ldots +c_{r}x_{n_{r-1}+2}$ is such that $[\xi,x] =w\in Z(\mathfrak{g})$.

While this condition is sufficient it is not necessary: for example, consider the nilpotent Lie algebra $\mathfrak{n}_{3}$ of strictly upper triangular $3\times 3$ complex matrices. Then,

$$\xi = \begin{bmatrix} 0&1&0\\\0&0&1\\\0&0&0\end{bmatrix}$$

is such that $im \; ad_{\xi}=Z(\mathfrak{n}_{3})$, while $\pi(\xi): 12$.

share|cite|improve this answer

Of course, the easiest case is when $\mathfrak{g}$ has nilpotence class 2 (that is, $[[\mathfrak{g},\mathfrak{g}],\mathfrak{g}]=0$). Under this assumption one has trivially that $im(ad_{\xi})⊆Z(\mathfrak{g})$, and $im(ad_{\xi}$) contains a 1-dimensional ideal whenever $\xi$ is not in $Z(\mathfrak{g})$.

share|cite|improve this answer

For generic $\xi$, the answer is yes.

Denote the lower central series as $$ \mathfrak{g}_0 = \mathfrak{g} \supset \mathfrak{g}_1 = [\mathfrak{g},\mathfrak{g}_0] \supset \mathfrak{g}_2 = [\mathfrak{g},\mathfrak{g}_1]\supset\ldots $$

Let $r$ be minimal such that $\mathfrak{g}_r\neq \{0\}$, $\mathfrak{g}_{r+1} = \{0\}$, and $s$ minimal such that $\textrm{ad}_\xi^s \neq 0$, and $\textrm{ad}_\xi^{s+1} = 0$. Suppose $\xi$ is such that $\textrm{Im}(\textrm{ad}_\xi)\cap Z(\mathfrak{g}) = \{0\}$, and also that it has maximal $s$ with respect to this property. We must have $s<r$, since $\mathfrak{g}_r \subset Z(\mathfrak{g})$.

Now for any other $x\in\mathfrak{g}$, view $\textrm{ad}_x^{s+1} = 0 $ as a system of polynomials on $\mathfrak{g}$. It's nontrivial since $s<r$, and since $\xi$ is a solution.

share|cite|improve this answer

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.