31
$\begingroup$

Let $V_N$ denote the $N$-dimensional representation of the quantum group $U_q(\mathfrak s\mathfrak l_2)$. I am told that in the limit $N\to\infty$ with $q=e^{2\pi i/n}$ and $N/n\to\alpha\in(0,1)$, the representation $V_N$ of $U_q(\mathfrak s\mathfrak l_2)$ "converges" in some precise sense to an infinite-dimensional representation of the complex Lie group $\operatorname{SL}_2(\mathbb C)$.

What is the precise formulation of this result, and where is it discussed in the literature? [My searches so far have not found anything substantial.]

$\endgroup$
  • 1
    $\begingroup$ I think arxiv.org/abs/hep-th/0312282v2 contains something like what you want, but for $N=n.$ Unfortunately I wasn't able to find anything more general/precise by looking through papers that reference it. $\endgroup$ – dhy Oct 1 '18 at 3:03
9
$\begingroup$

This is a very interesting question. I have also made some search but i have not found this result explicitly mentioned somewhere in the literature. However, i remember i have heard such a claim in the past. So i will try to record my thoughts on the problem. Here is the way i understand it:

Since you are considering the limits $N\to \infty$ and $N/n\to\alpha\in(0,1)$, this implies that $n\to\infty$ thus $q\to 1$. In this limit: $U_q(\mathfrak{sl}_2)\stackrel{q\to 1}{\longrightarrow} U(\mathfrak{sl}_2)$ and so you get a representation of the UEA and thus of the lie algebra $\mathfrak{sl}_2$. Since the corresponding lie group $SL(2,\mathbb{C})$, is simply connected, its category of representations is equivalent to the category of representations of the lie algebra $\mathfrak{sl}_2$ (in the sense that there is a bijection -up to isomorphism- between the lie group and the lie algebra representations). Thus, we finally arrive at a representation (up to isomorphism) of the lie group $SL(2,\mathbb{C})$. And since $N\to\infty$, the limit -if it exists- will be an infinite dim representation.

Let us try to examine the situation a little closer and attempt to establish whether -and when- the limit is well defined and exists :

The "convergence" of the quantum group to the lie algebra: A non-trivial point in the preceding argument is how to obtain a rigorous formulation of the limit:

$U_q(\mathfrak{sl}_2)\stackrel{q\to 1}{\longrightarrow} U(\mathfrak{sl}_2)$

This is a subtle point, which -imo- does not have a unique answer; it depends on the description one uses for the quantum group $U_q(\mathfrak{sl}_2)$. Generally, we can either view $U_q(\mathfrak{sl}_2)$ as a $\mathbb{C}[[h]]$-algebra and write $q=e^{h/2}$ and $K=q^H$ (in which case $q\to 1\Leftrightarrow h\to 0$ and we use Del' Hospital for the $E$-action limit and power series expansion for the $K$-action limit), or we can view $U_q(\mathfrak{sl}_2)$ as a suitable quotient of a larger $\mathbb{C}$-algebra, which is however well defined at $q=1$. Both methods are well-known. See for example:
- How $U_{q}(\mathfrak{sl}_{2})$ becomes the universal enveloping algebra $U(\mathfrak{sl}_{2})$ of $\mathfrak{sl}_{2}$,
- Quantized Enveloping Algebras at $q=1$,
- Exponentiations over the quantum algebra $U_{q}(\mathfrak{sl}_{2})$ (see the discussion and the computations of section 10, p. 65)

The "convergence" of the representation: Let us study the limit on the irreducible, f.d. representations of $U_{q}(\mathfrak{sl}_{2})$ with $q=e^{2\pi i/n}$:
(First recall that if $q$ is not a root of unity, then the f.d., irred, are highest weight representations. They are parameterized by $\varepsilon=\pm 1$ and the positive integers, i.e. we will denote such $N$-dim, irred, modules as $V_{\varepsilon, N-1}$. The notion of the $N\to\infty$ limit here is well defined in the following sense: The $V_{1,N-1}$ modules here have matrix elements which are not generally continuous functions of $q$ but they can be handled with methods similar to those mentioned above to show that

$V_{1,N-1}\stackrel{q\to 1}{\longrightarrow}V_{N-1}$

where $V_{N-1}$ are the usual $N$-dim, irred, highest weight reps of $\mathfrak{sl}_2$. The $V_{-1,N-1}$ modules "vanish" when $q\to 1$. One can be more precise at that point but it would be irrelevant to the rest. We can equally well consider that we are taking the $N\to\infty$ limit considering the $V_{1,N-1}$ family only).
Now, let us come back to the root of unity case: $q=e^{2\pi i/n}$. It is well known that, the irreducible modules now have an upper bound in their dimension: ${\small e=\begin{cases} n, & \text{$n$: odd} \\ n/2, & \text{$n$: even.} \end{cases}}$. It is also well known that the $N$-dim irred, representations, now fall into two classes: First, representations for which $N<e$. These are of the form $V_{\varepsilon,N-1}$, with $\varepsilon=\pm 1$, i.e. they are exactly the same modules as in the non-root of unity case; i.e. the highest weight modules. However, the second class includes irred. reps with dimension $N=e$. These look quite different than the highest weight modules; they generally fall into distinct subclasses including cyclic, semi-cyclic, etc modules. So it would be wise to exclude these $\dim V_{N-1}=N=e$ modules from the limitting procedure (if we want some reasonable sense of convergence).
Thus, while $N\to\infty$, it is important to keep in mind that $N<e$. This is -imo- the meaning of the $\alpha$ parameter: $N/n\to\alpha\in(0,1)$, which in some sense expresses the "relative rate of divergence" between the dimension $N$ of the rep and the order $n$ of the root of unity $q$.

Let us now attempt to compute these limits; i will not include details on the handling of the order of the limits $N/n\to\alpha$, $n\to\infty$, since i admit i have not -yet- been able to obtain a rigorous formulation of the following but here is what i have got:

  • First consider the limit $N/n\to\alpha$. We get the result:

    $V_{1,N-1}\stackrel{N/n\to \alpha}{\longrightarrow}V\big(q^{n\alpha}\big)=V\big(e^{2\pi i\alpha}\big)$

where $V_{1,N-1}$ is the $N$-dimensional, $U_q\big(\mathfrak{sl}(2)\big)$-module of highest weight $\lambda=q^{N-1}$ and the $V\big(q^{n\alpha}\big)$ is the $U_q\big(\mathfrak{sl}(2)\big)$-Verma module of highest weight $\lambda=q^{n\alpha}=e^{2\pi i\alpha}$.
The $U_q\big(\mathfrak{sl}(2)\big)$-action for the $V_{1,N-1}$ module is given by: $$ K\cdot v_p=q^{N-2p-1}v_p, \ \ E\cdot v_p=[N-p]_qv_{p-1}, \ \ F\cdot v_p=[p+1]_qv_{p+1} $$ whereas its $N/n\to\alpha$ limit, gives the $U_q\big(\mathfrak{sl}(2)\big)$-action for the $V\big(q^{n\alpha}\big)$-Verma module: $$ K\cdot v_p=q^{n\alpha-2p-1}v_p, \ \ E\cdot v_p=[n\alpha-p]_qv_{p-1}, \ \ F\cdot v_p=[p+1]_qv_{p+1} $$ where $[x]_q=\frac{q^x-q^{-x}}{q-q^{-1}}$, as usual.

  • The next step will be to compute the Verma module $V\big(q^{n\alpha}\big)$ limit at $q\to 1\Leftrightarrow n\to\infty$;
    Conjecture: This produces a $U\big(\mathfrak{sl}(2)\big)$-Verma module $V({\small 4\pi i \alpha-1})$:

    $V\big(q^{n\alpha}\big)\stackrel{n\to\infty}{\longrightarrow}V({\small 4\pi i \alpha-1})$

with $U\big(\mathfrak{sl}(2)\big)$-action given by: $$ H\cdot v_p=-(4\pi i\alpha-2p-1)v_p, \ \ E\cdot v_p=-(4\pi i\alpha-p)v_{p-1}, \ \ F\cdot v_p=(p+1)v_{p+1} $$ This is an infinite dimensional, Verma module, of highest weight $4\pi i \alpha-1$, for the (undeformed) universal enveloping algebra $U\big(\mathfrak{sl}(2)\big)$.

In any case, if there is not some silly mistake in my conjecture (the result comes from a mixture of computations and intuition), we finally arrive at an infinite dimensional $U\big(\mathfrak{sl}(2)\big)$-Verma module and from there we can get the corresponding infinite dimensional, $SL(2,\mathbb{C})$ representation. I will try to come back if -and when- i will be able to obtain a somewhat more rigorous formulation of the last limit.

$\endgroup$
  • 2
    $\begingroup$ A couple of comments: (1) in order for the limit to "converge", it is important to fix the parity of $N$ as it tends to $\infty$. (2) In order for the limit to be a representation as opposed to a representation-up-to-isomorphism one can fix, throughout the limiting process, a trivialization of one of the (one-dimensional) weight spaces -- e.g. the zero-weight space (assuming $N$ has been chosen to always be odd). $\endgroup$ – André Henriques Apr 22 at 11:13

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.