Let $L$ be a central extension of a simple Lie algebra $\mathfrak{g}$ such that $L=[L,L]$. It is not difficult to see that if $H^1(\mathfrak{g}, \mathfrak{g})=0$ then $H^1(L,L)=0$. In other words, if all derivations of $\mathfrak{g}$ are inner, then all derivations of $L$ are inner.

Is the converse true?

For instance, this is indeed the case when $L$ is the universal central extension of $\mathfrak{g}$. (See Theorem 2.2 of [G.M. Benkart - R.V. Moody: Derivations, central extensions, and affine Lie algebras. Algebras Groups Geom. ${\bf 3}$ (1986), no. 4, 456--492.])

  • $\begingroup$ Maybe useful to recall that $H^1(\mathfrak{h},\mathfrak{h})=0$ means that every self-derivation of $\mathfrak{h}$ is inner. $\endgroup$
    – YCor
    May 27, 2019 at 13:39
  • $\begingroup$ @Yves, I agree. The problem is just that a non-inner derivation of $\mathfrak{g}$ is not necessarily induced by a derivation of $L$. I would expect that the answer to my question is negative, however I have not a counterexample. $\endgroup$ May 29, 2019 at 9:01
  • $\begingroup$ Yes, I got this. My idea was to have a simple Lie algebra with a sufficiently big $H_2$, and a small but nontrivial Out, and mod out its universal central extension by a "random enough" subspace. Unfortunately my knowledge of simple Lie algebras of infinite dimension or in finite characteristic is too narrow to think of such examples. $\endgroup$
    – YCor
    May 29, 2019 at 9:06

1 Answer 1


I don't known if I interprete correctly your question. First, if $g$ is simple, then always $H^1(g,g)=0$. ALSO, $H^1(L,L)=0$ do not implies that $L$ is a central extensión of a simple algebra.For example, take $L$=the non abelian 2-dimensional Lie algebra, you can compute by hand Der(L,L) and check that they are all inner. (Same happens for borel subalgebras of semisimple ones).

  • 1
    $\begingroup$ Ciao Marco, in fact you are interpreting uncorrectly my question...First of all, I am not assuming that the simple Lie algebra $\mathfrak{g}$ is finite-dimensional and defined over a field of characteristic zero, so it is not true in general that $H^1(\mathfrak{g},\mathfrak{g} )=0$. Second, $L$ is by hypothesis a central extension of a simple Lie algebra because this is just the specific situation I am interested in. $\endgroup$ May 30, 2019 at 12:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.