Simple Affine Vertex Operator Algebras

• Simple affine vertex operator algebras are a class of VOAs constructed from affine Kac–Moody algebras, featuring a universal structure and a unique simple quotient at various levels.
• They leverage singular vector techniques and Zhu’s C2-algebra computations to uncover geometric structures like associated varieties, Dixmier sheets, and nilpotent orbit closures.
• Their construction informs module classification and quantum Hamiltonian reductions, with results applicable across Lie types including sl3 and Dℓ.

Simple affine vertex operator algebras (VOAs) are a distinguished class of vertex operator algebras constructed from affine Kac–Moody algebras. For a finite-dimensional simple Lie algebra $\mathfrak{g}$, the universal affine VOA $V^k(\mathfrak{g})$ is defined at any level $k \in \mathbb{C}$ and admits a unique simple graded quotient $L_k(\mathfrak{g})$. The structure of these algebras and their associated varieties, particularly at non-admissible or negative levels, is deeply intertwined with singular vector theory, Zhu’s $C_2$-algebras, and the geometry of nilpotent orbits and Dixmier sheets.

1. Structure of Universal and Simple Affine VOAs

Let $\mathfrak{g}$ be a finite-dimensional simple Lie algebra over $\mathbb{C}$, equipped with an invariant bilinear form. The corresponding affine Kac–Moody algebra is

$$\widehat{\mathfrak{g}} = \mathfrak{g}[t,t^{-1}] \oplus \mathbb{C}K,$$

with commutator

$$[x \otimes t^m, y \otimes t^n] = [x, y] \otimes t^{m+n} + m (x|y) \delta_{m+n,0} K, \quad [K, \widehat{\mathfrak{g}}] = 0.$$

The universal affine VOA at level $k$ is constructed as

$$V^k(\mathfrak{g}) = U(\widehat{\mathfrak{g}})/\langle \mathfrak{g}[t], K - k \rangle$$

with vacuum vector $\mathbf{1}$ and fields given by $x(z) = \sum_{n \in \mathbb{Z}} x(n) z^{-n-1}$ for $x\in\mathfrak{g}$. There is a unique simple graded quotient

$$L_k(\mathfrak{g}) = V^k(\mathfrak{g})/\mathcal{I}_{\rm max},$$

where $\mathcal{I}_{\rm max}$ is the maximal proper graded ideal. This quotient is customarily called the simple affine VOA at level $k$.

For explicit construction in type $D$ (e.g., $D_\ell$), the universal VOA can also be written as an induced module using the standard triangular decomposition of $\widehat{\mathfrak{g}}$: $$V^k(\mathfrak{g}) = U(\widehat{\mathfrak{g}}) \otimes_{U(\widehat{\mathfrak{g}}_0 \oplus \widehat{\mathfrak{g}}_+)} \mathbb{C}_k.$$ The central charge is $c = k \cdot \dim \mathfrak{g} / (k + h^\vee)$. The construction holds for all $k \ne -h^\vee$ (Jiang et al., 2024, Perse, 2012).

2. Singular Vectors and Generation of the Maximal Ideal

The structure of $\mathcal{I}_{\rm max} \subset V^k(\mathfrak{g})$ is controlled by singular vectors. For certain levels $k$, explicit singular vectors can be constructed whose $U(\widehat{\mathfrak{g}})$-span generates the maximal ideal.

$\mathfrak{g} = sl_3$ at $k = -3 + \frac{2}{2m+1}$

For the series $k = -3 + \frac{2}{q}$ with $q=2m+1$ and $m\in \mathbb{Z}_{>0}$, $V^k(sl_3)$ is not simple and its maximal ideal is generated by two independent singular vectors of conformal weight $3q$. Explicitly, up to lower-depth terms in the Li filtration, these generators take the forms

$$v_1 \equiv \sum_{i=0}^{3q-1} a_i \, e_{-\theta}(-1)^{i+1} (h_1(-1) + h_2(-1))^{3q-1-i} \mathbf{1},$$

$$v_2 \equiv \sum_{i=0}^{3q-1} b_i \, e_{-\theta}(-1)^{i+1} (h_1(-1) + h_2(-1))^{3q-1-i} \mathbf{1},$$

where $a_i, b_i \in \mathbb{C}$, $\alpha_1, \alpha_2$ are the simple roots, $\theta = \alpha_1 + \alpha_2$ is the highest root, and the weights are given explicitly with swapped root labels. No uniform closed-form expansion for arbitrary $m$ exists, but the generators are characterized by the highest weights

$$k\Lambda_0 - 3q\,\delta + \alpha_1 + 2\alpha_2, \quad k\Lambda_0 - 3q\,\delta + 2\alpha_1 + \alpha_2.$$

Generation of the maximal ideal is shown via transport from the $k=-1$ case (Adamović–Perše), use of Kashiwara–Tanisaki character formulas, and Fiebig’s category equivalences. Depth/degree analysis in the Li filtration confirms no further independent singular vectors arise (Jiang et al., 2024).

Type $D_\ell$ at $k = n-\ell+1$

For $D_\ell$ and arbitrary $n\in\mathbb{Z}_{>0}$, singular vectors

$$U_n = \left(\sum_{i=2}^\ell e_{\epsilon_1-\epsilon_i}(-1) e_{\epsilon_1+\epsilon_i}(-1)\right)^n \mathbf{1}$$

exist in $V^{k}(D_\ell)$ at $k = n-\ell+1$. For $\ell=4$, $n=1$, the automorphism group yields three independent singular vectors, fully generating $\mathcal{I}_{\rm max}$ (Perse, 2012).

3. Associated Varieties and Zhu’s $C_2$-Algebras

Given any VOA $V$, Zhu’s $C_2$-algebra is $R_V = V/C_2(V)$ with $$C_2(V) = \operatorname{span}\{a_{(-2)}b\,|\,a, b \in V\}$$. The associated variety is defined as $X_V = \operatorname{Specm}(R_V)$, a conic, $G$-invariant subvariety of $\mathfrak{g}^* \cong \mathfrak{g}$. For the universal VOA $$R_{V^k(\mathfrak{g})} \cong \mathbb{C}[\mathfrak{g}^*]$$, so $X_{V^k(\mathfrak{g})} = \mathfrak{g}^*$. For quotients by ideals generated by singular vectors, the associated variety is determined by the vanishing of the symbols of these vectors in $R_{V^k(\mathfrak{g})}$.

For $L_k(sl_3)$ at $k = -3 + \frac{2}{q}$:

The ideal $$I_k = \langle \overline{v_1}, \overline{v_2} \rangle \subset \mathbb{C}[sl_3]$$ cuts out

$$X_{L_k(sl_3)} = \{g \cdot (t(h_1 - h_2) + f_\theta) \mid g\in SL_3,\, t\in \mathbb{C}\},$$

which is the closure of the sheet $$S_{\min} = \operatorname{Ad}^* SL_3 (\mathbb{C}^*(h_1-h_2) + f_\theta)$$, a Dixmier sheet of rank 1. The nilpotent boundary is the minimal orbit $\mathcal{O}_{\min}$; $\dim X_{L_k(sl_3)} = 5$ (Jiang et al., 2024).

4. Simple Affine $W$-Algebras and Quantum Hamiltonian Reduction

For a nilpotent element $f$ of $\mathfrak{g}$, the (finite or affine) $W$-algebra $W_k(\mathfrak{g}, f)$ is constructed via quantum Drinfeld–Sokolov (DS) reduction: $$W_k(\mathfrak{g}, f) = H^{\bullet}_{\rm DS,f}(L_k(\mathfrak{g})).$$ The associated variety is given by

$$X_{W_k(\mathfrak{g}, f)} = X_{L_k(\mathfrak{g})} \cap (f + \mathfrak{g}^e),$$

where $\mathfrak{g}^e$ is the centralizer of $e$ ($[e,f]=h$ for an $\mathfrak{sl}_2$ triple).

Minimal Nilpotent $f = f_\theta$ for $sl_3$

$$X_{W_k(sl_3, f_\theta)} = \{B \in f_\theta + \mathfrak{g}^e \mid \det B = 0\}$$

is a three-dimensional affine subset of the Slodowy slice at $f_\theta$. The variety exhibits infinitely many symplectic leaves, so $W_k(sl_3, f_\theta)$ is not quasi-lisse (Jiang et al., 2024).

Regular Nilpotent $f = f_{\text{reg}}$

$$X_{W_k(sl_3, f_{\text{reg}})} = X_{L_k(sl_3)} \cap (f_{\text{reg}} + \mathfrak{g}^e_{\text{reg}})$$

is a one-parameter nilpotent line isomorphic to $\mathbb{C}$, again with infinitely many symplectic leaves and non-quasi-lisse structure.

5. Module Theory and Classification

Singular vector techniques, in conjunction with Zhu’s algebra, enable detailed classification of highest-weight modules for simple affine VOAs:

• For type $D_4$ at $k = -2$, the only irreducible ordinary $L_{D_4}(-2,0)$-module is the adjoint module itself. Among weak modules in the weight category $\mathcal{O}$, five are identified, corresponding to $L_{D_4}(-2,0)$ and highest weights $-2\omega_1$, $-2\omega_3$, $-2\omega_4$, $-\omega_2$, where $\omega_i$ are fundamental weights. All ordinary modules are semisimple (Perse, 2012).
• For $sl_3$, the non-admissible levels $k = -3 + 2/q$ support simple affine VOAs $L_k(sl_3)$ with associated varieties given by Dixmier sheets, confirming that such levels, while not rational, still yield “nice” geometric invariants (Jiang et al., 2024).

6. Broader Implications and Connections

These findings demonstrate that, beyond admissible or rational levels, certain negative or fractional levels lead to simple affine VOAs whose maximal ideals are generated by explicit singular vectors. The associated varieties are conic, typically dramatically smaller than the full nilpotent cone yet richer than a single nilpotent orbit, revealing intricate Poisson and geometric structure.

The connection with Poisson geometry (via associated varieties, Dixmier sheets, Slodowy slices), as well as representation-theoretic techniques (singular vectors, character formulae, block equivalence), are central. These results confirm and extend physical expectations from 4D/2D dualities, indicating that affine VOAs at non-admissible levels can produce distinguished invariants of geometric and physical significance.

The methodology, integrating Li-filtration, singular vector generation, and Zhu’s algebra computations, is robust and has been effective in type $A$, $D$, and beyond, with potential for further generalizations to other Lie types and levels (Jiang et al., 2024, Perse, 2012).

---

Summary of Key Structures in Simple Affine VOAs and Related Objects:

Algebra	Level $k$	Maximal Ideal Generators	Associated Variety	Quasi-Lisse
$L_k(sl_3)$	$-3+\frac{2}{2m+1}$	2 singular vectors, weight $3q$	Dixmier sheet of rank 1 in $sl_3^*$	No
$W_k(sl_3,f_\theta)$	as above	Quantum Hamiltonian reduction	3D affine in Slodowy slice at $f_\theta$	No
$L_{D_4}(-2,0)$	$-2$	3 singular vectors	Not specified	(—)

This synthesis underscores the central role of explicit singular vector construction, associated variety computation, and the geometric classification of simple affine vertex operator algebras across diverse types and levels (Jiang et al., 2024, Perse, 2012).