1

Let $\mathfrak{g}$ be a Lie algebra and $U(\mathfrak{g})$ be its universal enveloping algebra, that is $$U(\mathfrak{g}) = \bigotimes\mathfrak{g}/\mathcal{K},$$ where $\bigotimes\mathfrak g$ denotes the tensor algebra of $\mathfrak g$ and $\mathcal K$ is the ideal of $\bigotimes\mathfrak g$ generated by the elements of the form $[x,y]-x\otimes y+y\otimes x$ with $x,y\in\mathfrak g$.

As far as I understand it, if $\mathfrak g$ is the Lie algebra of a Lie group $G$, and therefore we can regard $\mathfrak g$ as the set of left-invariant vector fields in $G$, the Poincaré-Birkoff-Witt theorem tells us that $U(\mathfrak g)$ corresponds the set of left-invariant differential operators of all orders on $G$. This means that, given $h\in G$, if we denote the left and right translations on $G$ by $$\lambda(h)f(g) = f(h^{-1}g),\, \rho(h)f(g) = f(gh),\quad\forall g\in G,$$ then $$\lambda(h)\circ D = D\circ\lambda(h),\quad\forall D\in U(\mathfrak g).$$

We denote by $Z(U(\mathfrak g))$ the center of $U(\mathfrak g)$.


Now, to the main question: I have seen it mentioned that $$D\in Z(U(\mathfrak g)) \Leftrightarrow \rho(h)\circ D = D\circ \rho(h),$$ but I simply cannot understand how commutators and right translations are related. I would be grateful if someone could offer me some insight.

Vitor Borges
  • 1,305
  • But $\rho(h)$ is not a differential operator. Moreover, even if this was true, it does guarantee the "only if" part. – Vitor Borges Dec 04 '20 at 18:27
  • If $D\in Z(U(\mathfrak g))$ we only know that it commutes with all left-invariant differential operators. I really can't see how that is related to $\rho(h)$, would you mind expanding a little bit more on that? – Vitor Borges Dec 05 '20 at 01:07
  • See also Casimir element for it, in connection with this post, or perhaps other posts which fit better. – Dietrich Burde Dec 05 '20 at 09:25
  • I see that the bracket $[D,\rho(h)]$ makes sense, I just don't see what you are using to conclude that $[D,\rho(h)] = 0$, since we are only supposing that $[D,P]=0$ for $P\in U(\mathfrak g)$. – Vitor Borges Dec 05 '20 at 14:24
  • I just wanted to say that by definition of the Casimir $\rho(\Omega)$ we have $[D,\rho(\Omega)]=0$ for all $D\in U(\mathfrak{g})$. So I have read $\rho(h)=\rho(\Omega)$, but this might be a different notation (see the wikipedia link above). – Dietrich Burde Dec 05 '20 at 15:27
  • Oh, sorry. I didn't know this was a common notation for the Casimir. I am using $\rho(h)$ to denote the right translation by $h\in G$ (as defined in the question). – Vitor Borges Dec 05 '20 at 18:12

0 Answers0