Let $\frak g$ be a semi-simple finite dimensional Lie algebra over the complex numbers $\mathbb C$. Then every non irreducible representation of $\frak g$ is completely reducible.

Q1: Is category f.g. left $U(\mathfrak{g})$-Mod is equivalent to the category finite dimensional $\operatorname{Rep}(\mathfrak{g})$?

Q2: If every f.g $U(\mathfrak{g})$ module is completely reducible (in other words, they can be written as direct sums of simples) and above category equivalence holds, then from completely reducibility, I deduce for semi-simple Lie algebra, $U(\mathfrak{g})$ is semisimple as a ring. Hence, I would expect all $U(\mathfrak{g})$ modules to be projective and injective at the same time. Is this correct?

Q3: If Q1, Q2 are wrong, what is the relationship between semi-simple Lie algebras and semi-simple rings? (I would not expect for infinite dimensional representation, above categorical equivalence holds. The "semi-simple" ring should be interpreted as every f.g. module is decomposed into simple modules only.)

  • 1
    $\begingroup$ Then every finite-dimensional representation is completely reducible. So we don't need "non irreducible". $\endgroup$ Apr 13 '19 at 18:38
  • 2
    $\begingroup$ $U(\mathfrak g)$ is not finite dimensional ! $\endgroup$ Apr 13 '19 at 18:38
  • $\begingroup$ @NicolasHemelsoet Ah, my dumbness. I mistaken the product with lie bracket which is not associative. Thanks for the correction. So $U(g)$ is never finite dimensional unless $dim_C(g)=0$. $\endgroup$
    – user45765
    Apr 13 '19 at 18:49
  • $\begingroup$ @DietrichBurde Is semi-simpleness in representation same as semi-simpleness as ring sense? I mean semi simple here for f.g. objects only. The ring semi-simpleness demands much stronger statement for any modules, they decompose into simples. $\endgroup$
    – user45765
    Apr 13 '19 at 18:53
  • 2
    $\begingroup$ @user45765 “Every fg module is semisimple” is equivalent to “every module is semisimple” for rings with identity. $\endgroup$
    – rschwieb
    Apr 13 '19 at 22:00

Just to give a definite answer :

Q1 : No. For example, any simple modules can be expressed as some quotient of $U(\mathfrak g)$ (look for "Verma modules"), and there is so such module in $Rep(\mathfrak g)$.

Q2 : No. For example, the natural sequence $$ 0 \to M(-2) \to M(0) \to L(0) \to 0 $$ does not split (here $M(-2), M(0)$ are two Verma modules for $\mathfrak{sl}_2$ and $L(0)$ is the trivial representation). Note that $M(0), M(-2)$ are infinite-dimensional.

Q3 : The only relation is that $Rep(\mathfrak g)$ is a semisimple category when $\mathfrak g$ is semisimple. So your last sentence is the correct interpretation.

Remark : $U(\mathfrak g)$ is a rather complicated ring. For more information you can look at Dixmier's book on Universal Envelopping algebra. Also you can look at Humphrey's book about BGG category $\mathcal O$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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