Solvable Lie algebra

See @olver86 pag. 151.
Throughout, let g⊆gl(n,K) be a finite‑dimensional matrix Lie algebra over a field K of characteristic zero (typically K=R or C).

Define the derived series,by using the commutator subalgebra, by

g(0)=g,g(k+1)=[g(k),g(k)],k≥0.

We say g is solvable if

∃rs.t.g(r)={0}.

In the matrix setting one has complete splitting of the adjoint action. The following are equivalent:

  1. g is solvable.
  2. There exists a chain of ideals
{0}=I0⊲I1⊲I2⊲⋯⊲In=g,dim⁡Ik=k.
  1. There exists a sequence of subalgebras
{0}=a0⊂a1⊂…⊂ak=g

such that ai is an ideal of ai+1 and the quotient ai+1/ai is abelian.
4. Equivalently, there exists a basis {v1,…,vn} of g such that

[vi,vj]∈span{v1,…,vj−1},∀1≤i<j≤n.
  1. Equivalently, in that same basis each adjoint adv is represented by an upper‑triangular n×n matrix.

It can be shown that a connected Lie group G⊂GL(n,K) is solvable if and only if its Lie algebra is solvable.

On the other hand, if g is a finite dimensional Lie algebra, there exists a unique maximal solvable ideal called the radical of a Lie algebra.