Another identity implied by the Jacobi identity for semidirect product of Lie algebra

54 Views Asked by At

I'm working through Introduction to Lie Algebras by Erdmann and Wildon. In Chapter 3, Exercise 3.9 asks the reader to develop the semidirect product of Lie algebras. I got very stuck on the verification of the Jacobi identity, but the link here suggested the hint/trick of breaking the identity down so there are zeros in each pair, which gets you the required verification. However...

Recalling that the Lie bracket is defined as: $$ [(s_1,x_1),(s_2,x_2)]=([s_1,s_2],[x_1,x_2]+\theta(s_1)(x_2)-\theta(s_2)(x_1)), $$ where $\theta:S\rightarrow \text{Der}(X)$, I did the obvious thing and try to push the identity through as far as I could, simplifying along the way, and the most simplification I could achieve was to get: $$ (0,\theta(s_1)[x_2,x_3]+\theta(s_2)[x_3,x_1]+\theta(s_3)[x_1,x_2]), $$ which, if the Jacobi holds, must equal 0, which implies that in any semidirect product, this is true: $$ \theta(s_1)[x_2,x_3]+\theta(s_2)[x_3,x_1]+\theta(s_3)[x_1,x_2]=0. $$

I couldn't figure out how to prove this, but is this an actual known identity in semidirect products? That seems like a very restrictive condition.

For reference, here are the steps of my calculation. Start here:

$$ [(s_1,x_1),[(s_2,x_2),(s_3,x_3)]]+[(s_2,x_2),[(s_3,x_3),(s_1,x_1)]]+[(s_3,x_3),[(s_1,x_1),(s_2,x_2)]]. $$

Then we have: $$ [(s_2,x_2),(s_3,x_3)]=([s_2,s_3],[x_2,x_3]+\theta(s_2)(x_3)-\theta(s_3)(x_2)), $$

$$ [(s_3,x_3),(s_1,x_1)]=([s_3,s_1],[x_3,x_1]+\theta(s_3)(x_1)-\theta(s_1)(x_3)), $$

$$ [(s_1,x_1),(s_2,x_2)]=([s_1,s_2],[x_1,x_2]+\theta(s_1)(x_2)-\theta(s_2)(x_1)). $$

Then,

$$ [(s_1,x_1),([s_2,s_3],[x_2,x_3]+\theta(s_2)(x_3)-\theta(s_3)(x_2))]= ([s_1,[s_2,s_3]],[x_1,[x_2,x_3]]+\theta(s_1)([x_2,x_3]+\theta(s_2)(x_3)-\theta(s_3)(x_2))-\theta([s_2,s_3])(x_1)), $$

$$ [(s_2,x_2),([s_3,s_1],[x_3,x_1]+\theta(s_3)(x_1)-\theta(s_1)(x_3))]= ([s_2,[s_3,s_1]],[x_2,[x_3,x_1]]+\theta(s_2)([x_3,x_1]+\theta(s_3)(x_1)-\theta(s_1)(x_3))-\theta([s_3,s_1])(x_2)), $$

$$ [(s_3,x_3),([s_1,s_2],[x_1,x_2]+\theta(s_1)(x_2)-\theta(s_2)(x_1))]= ([s_3,[s_1,s_2]],[x_3,[x_1,x_2]]+\theta(s_3)([x_1,x_2]+\theta(s_1)(x_2)-\theta(s_2)(x_1))-\theta([s_1,s_2])(x_3)). $$

Things start to go our way, where you recognize the Jacobi identity for the underlying Lie algebras. Adding across components, we get that $[s_1,[s_2,s_3]]+[s_2,[s_3,s_1]]+[s_3,[s_1,s_2]]=0$ and $[x_1,[x_2,x_3]]+[x_2,[x_3,x_1]]+[x_3,[x_1,x_2]]=0$. In the second component, here's what's "left:"

$$ \theta(s_1)([x_2,x_3]) + \theta(s_1)\circ\theta(s_2)(x_3) -\theta(s_1)\circ\theta(s_3)(x_2) - \theta([s_2,s_3])(x_1) + \theta(s_2)([x_3,x_1]) + \theta(s_2)\circ\theta(s_3)(x_1) - \theta(s_2)\circ\theta(s_1)(x_3) - \theta([s_3,s_1])(x_2) + \theta(s_3)([x_1,x_2]) + \theta(s_3)\circ\theta(s_2)(x_2) - \theta(s_3)\circ\theta(s_2)(x_1) - \theta([s_1,s_2])(x_3). $$

Rearranging:

$$ \theta(s_1)([x_2,x_3])+\theta(s_2)([x_3,x_1])+\theta(s_3)([x_1,x_2]) + \theta(s_1)\circ\theta(s_2)(x_3) - \theta(s_2)\circ\theta(s_1)(x_3) + \theta(s_2)\circ\theta(s_3)(x_1) - \theta(s_3)\circ\theta(s_2)(x_1) + \theta(s_3)\circ\theta(s_1)(x_2) - \theta(s_1)\circ\theta(s_3)(x_2) - \theta([s_2,s_3])(x_1) - \theta([s_3,s_1])(x_2) - \theta([s_1,s_2])(x_3). $$

Things cancel, for example:

$$ \theta(s_1)\circ\theta(s_2)(x_3) - \theta(s_2)\circ\theta(s_1)(x_3) - \theta([s_1,s_2])(x_3) = \theta([s_1,s_2])(x_3)-\theta([s_1,s_2])(x_3)=0. $$