3
$\begingroup$

A Lie algebra $\mathfrak{g}$ generates its universal enveloping algebra $\mathrm{U}\mathfrak{g}$, which has the structure of a Hopf algebra. Modules of $\mathrm{U}\mathfrak{g}$ are exactly the of modules of $\mathfrak{g}$ itself.

In the case $\mathfrak{g} = \mathfrak{su}_2$, the simple modules are labeled uniquely (up to isomorphism) by their dimension, and we label them $V_d$ for $d = 1, 2,$ etc. Note that $\mathrm{U}\mathfrak{su}_2$ is itself a $\mathfrak{su}_2$ module and decomposes into simple modules as $\bigoplus_{j = 0}^\infty V_{2j+1}$, i.e. only odd-dimensional modules occur. We can consider truncations of this sum too: $$\mathrm{U}_J\mathfrak{su}_2 \equiv \mathrm{End}\,V_{2J+1} \simeq V_{2J+1}^\ast \otimes V_{2J+1} \simeq \bigoplus_{j = 0}^{2J} V_{2j+1}$$ In particular, $\dim \mathrm{U}_J\mathfrak{su}_2 = 1 + 3 + \cdots + (4J+1) = (2J+1)^2$, which is $\dim \mathrm{End}\,V_{2J+1}$. An explicit relationship between the basis adapted to the decomposition of $\mathrm{U}_J \mathfrak{su}_2$ given above and the basis $\{E_{ij}\}$ (with components 1 at $(i,j)$ and 0 elsewhere) on $\mathrm{End}\,V_{2J}$ can be given in terms of the $3jm$ symbols. The result shows that $\mathrm{U}_J \mathfrak{su}_2$ is isomorphism to $\mathrm{End}\,V_{2J+1}$ as associative algebras.

W. Smoke's paper Invariant Differential Operators uses the relationship between $\mathrm{U}\mathfrak{su}_2$ and differential operators on the $\mathrm{SU}(2)$ group manifold. (He actually works for a generic Lie algebra $\mathfrak{g}$ but my focus is only on $\mathfrak{su}_2$ for now.) The filtration of $\mathrm{U}\mathfrak{su}_2$ is the filtration of differential operators by degree, where $\mathrm{U}_J\mathfrak{su}_2$ gives differential operators with degree at most $J$. At the top of his eighth page (p. 467 in the journal), it is proved that $\mathrm{U}_J \mathfrak{su}_2$ is a subcoalgebra of $\mathrm{U}\mathfrak{su}_2$.

My understanding is that there is no Hopf algebra structure on matrix algebras like $\mathrm{End}\,V_{2J+1} \simeq \mathrm{End}\,\mathbb{C}^{2J+1}$ because they are simple. There is an algebra structure and a coalgebra structure, but these must not be compatible in the sense of bialgebras. But in the limit $J \to \infty$ the compatibility seems to be restored, since we know $\mathrm{U}\mathfrak{su}_2$ is a Hopf algebra. How does this work?

$\endgroup$
9
  • 2
    $\begingroup$ If two algebras are isomorphic and one of them has an additional Hopf algebra structure then you can transport that structure along the isomorphism; this is called, as you might expect, "transport of structure." $\endgroup$ Jun 23, 2015 at 16:31
  • $\begingroup$ Excellent. The construction seems like it should go through to me. But I'm a physicist and algebra is something I learn on the side meaning I'm not always confident of my hunches.In Hazewinkel et al. "Lie Algebras and Hopf Algebra" I read that the kernel of the counit is a nontrivial ideal in the algebra, but a simple matrix algebra would have no such thing. $\endgroup$
    – josh
    Jun 23, 2015 at 16:40
  • $\begingroup$ Oh, I misread your question. I think your truncation is not a Hopf algebra. $\endgroup$ Jun 23, 2015 at 16:41
  • $\begingroup$ Oh okay. What breaks down is the coalgebra structure of $\mathrm{U}_J\mathfrak{su}_2$, or is it the needed compatibility of the algebra and coalgebra structures? $\endgroup$
    – josh
    Jun 23, 2015 at 16:46
  • 1
    $\begingroup$ In what way is your $U_J\mathfrak{su}_2$ an algebra? $\endgroup$ Jun 23, 2015 at 16:51

1 Answer 1

4
$\begingroup$

The problem is that $U_J\frak{su}_2$ as you define it, that is $\bigoplus\limits_{j=0}^{2J}V_{2j+1}$, is NOT the space of all differential operators of order at most $J$. For the latter, Formula (2) in the paper suggests that it is the sum $\bigoplus\limits_{j=0}^J S^j(\mathfrak{su}_2)$, so its dimension is equal to the sum $\sum\limits_{j=0}^{J}\frac{(j+1)(j+2)}{2}=\frac{(J+1)(J+2)(J+3)}{6}\ne(2J+1)^2$.

In particular, a natural way to view $U_J\frak{su}_2$ as you define it is as a quotient space of $U\frak{su}_2$, not a subspace, so there is a natural algebra structure but no natural coalgebra structure.

$\endgroup$

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.