Seiberg-Witten curves of $\widehat{D}$-type Little Strings

Little String Theories (LSTs) arise in string theory (or related dual descriptions) through different decoupling limits [Witten:1995zh, Aspinwall:1997ye, Seiberg:1997zk, Intriligator:1997dh, Hanany:1997gh, Brunner:1997gf] (see also [Aharony:1999ks, Kutasov:2001uf] for reviews). They follow an $\widehat{ADE}$-classification [Bhardwaj:2015oru] and can be engineered from numerous setups. For concreteness, we begin with a construction in M-theory [Hohenegger:2015btj, Hohenegger:2016eqy], consisting of $N$ M5-branes (along directions $x^{0,1,\ldots,5}$), separated along a circle $\mathbb{S}_{6}^{1}$ probing a transverse $\mathbb{R}^{4}_{\perp}/\Gamma$ orbifold with a finite subgroup of $SU(2)$, $\Gamma\subset SU(2)$:

	$\mathbb{S}^{1}_{0}$	$\mathbb{S}^{1}_{1}$		$\mathbb{R}^{3}_{\parallel}$		$\mathbb{S}_{5}^{1}$	$\mathbb{S}_{6}^{1}$		$\mathbb{R}^{4}_{\perp}\!\!\!$	$\!\!\!\!\!\!/\Gamma$
	$x^{0}$	$x^{1}$	$x^{2}$	$x^{3}$	$x^{4}$	$x^{5}$	$x^{6}$	$x^{7}$	$x^{8}$	$x^{9}$	$x^{10}$
$N$ $M5$	$=$	$=$	$=$	$=$	$=$	$=$	$\times$

In the low energy limit, this construction corresponds to a supersymmetric quiver gauge theory on $\mathbb{S}_{0}^{1}\times\mathbb{S}_{1}^{1}\times\mathbb{R}^{3}_{\parallel}\times\mathbb{S}_{5}^{1}$. The McKay correspondence [Douglas:1996sw] associates the orbifold group $\Gamma$ to an affine quiver group $\widehat{\Upsilon}$ as follows

Orbifold group $\Gamma$	$\mathbb{Z}_{M}$ for $M\geq 1$	$\mathbb{BD}_{M-2}$ for $M\geq 4$	$\mathbb{BT}$	$\mathbb{BO}$	$\mathbb{BI}$
Quiver group $\widehat{\Upsilon}$	$\widehat{A}_{M-1}$	$\widehat{D}_{M}$	$\widehat{E}_{6}$	$\widehat{E}_{7}$	$\widehat{E}_{8}$

where $\mathbb{BD}_{M}$ is the binary dihedral group of order $4M$ and $\mathbb{BT},\mathbb{BO}$ and $\mathbb{BI}$ are respectively the binary isomorphism groups of the tetrahedron, the cube and the dodecahedron. The quiver takes the form of the Dynkin diagram of $\widehat{\Upsilon}$ (see Fig. 1) and mathematically is described by $(\Upsilon_{0},\Upsilon_{1})$, where $\Upsilon_{0}$ is the set of nodes and $\Upsilon_{1}$ the set of edges connecting the latter. Fig. 1 also display an $n$-coloring of the quivers, i.e. a particular assignment of the rank of each gauge node. To explain this, we introduce the Cartan matrix of the quiver $(\Upsilon_{0},\Upsilon_{1})$ by

$$c_{ij}=2\delta_{ij}-\sum_{e\in\Upsilon_{1}:i\to j}1-\sum_{e\in\Upsilon_{1}:j\to i}1\,,\quad\quad\forall i,j\in\Upsilon_{0}\,.$$

(2.1)

The rank of the gauge group at each node in $\Upsilon_{0}$ is uniquely fixed⁴⁴4Here we do not assume any additional flavor groups. by demanding the vanishing of the beta-functions for the gauge couplings [Lawrence:1998ja]:

$$\sum_{j\in\Upsilon_{0}}c_{ij}\cdot n_{j}=0\,,\quad\quad\forall i\in\Upsilon_{0}\,,$$

(2.2)

with $n_{j}$ the rank of gauge group associated with the $j$-th node. A solution of the set of equations (2.2) is called an $n$-coloring of the quiver. For quivers of simply laced Lie groups, the coloring is fixed by a single integer $N$ which corresponds to the number of M5-branes probing the orbifold. Concretely, in Fig. 1, the nodes of the quiver represent gauge groups $\widehat{A}_{N-1}$, while lines denote matter in the bifundamental representation.⁵⁵5Certain quiver gauge theories, such as the $\widehat{A}_{0}$ theory with $M=1$ has matter in the adjoint- rather than the bifundamental representation [Hohenegger:2015btj]. However, we shall not consider this case in the current paper. We shall denote a LST in this framework as $(\widehat{\Upsilon},N)$.

Reviewing first the case of $\Gamma=\mathbb{Z}_{M}$, the so-called A-type LSTs engineered from the M-brane setup in Table 2.1 can be described by a doubly elliptic Seiberg-Witten curve [Braden:2003gv, Hollowood:2003cv]. To understand the origin of the latter, we recall [Haghighat:2013gba, Haghighat:2013tka, Hohenegger:2013ala, Hohenegger:2015cba, Hohenegger:2015btj, Hohenegger:2016eqy] that this M-brane configuration can be described by $MN+2$ independent parameters: we refer the reader for more details to [Hohenegger:2015btj, Hohenegger:2016eqy] and only mention that in the following we denote the (complexified) radii of $\mathbb{S}_{1}^{1}$ and $\mathbb{S}_{6}^{1}$ (measured in units of the radius of $\mathbb{S}_{0}^{1}$) by $\tau$ and $\rho$ respectively. Furthermore, following [Hohenegger:2015btj], we introduce a $U(1)_{S}$ deformation of $\mathbb{R}_{\perp}^{4}/\mathbb{Z}_{M}$ with respect to $\mathbb{S}_{0}^{1}$

$\displaystyle U(1)_{S}:\hskip 11.38092pt(w_{+},w_{-})\longrightarrow\left(e^{2\pi iS}\,w_{+}\,,e^{-2\pi iS}\,w_{-}\right)\,,$

with

$\displaystyle w_{\pm}=x^{7}\pm ix^{8}\,,$

(2.3)

which is compatible with the $\mathbb{Z}_{M}$ action. The parameters $(\tau,\rho,S)$ can be arranged into the period matrix $\Omega$ of a genus 2 Riemann surface and the SWC (parametrised by $(z_{1},z_{2})$) can be written in terms of genus $2$ theta functions $\Theta$ (see (A.1) for the definition) [Braden:2003gv, Hollowood:2003cv]:

$$\sum_{i=0}^{M-1}\sum_{j=0}^{N-1}a_{ij}\,\Theta\!\begin{bmatrix}i/M&j/N\\ 0&0\end{bmatrix}\!(z\,|\Omega)=0\,,\quad z=\begin{bmatrix}z_{1}\\ z_{2}\end{bmatrix},\quad\Omega=\begin{bmatrix}\frac{\tau}{M}&S\\ S&\frac{\rho}{N}\end{bmatrix}.$$

(2.4)

The $\{a_{ij}\}$ in (2.4) depend on the remaining parameters of the M-brane configuration mentioned above.⁶⁶6Indeed, due to an overall scaling symmetry of (2.4), the moduli space of the SWC is indeed $(NM+2)$-dimensional. Using a dual description in terms of F-theory compactified on a Calabi-Yau threefold $X_{N,M}$ [Hohenegger:2015btj, Hohenegger:2016yuv, Hohenegger:2016eqy] with double elliptic-fibration structure, the $\{a_{ij}\}$ have been characterised in [Kanazawa:2016tnt] as (the product of) generating functions of Gromov-Witten invariants.

From the perspective of the $\widehat{A}_{M}$ quiver gauge theory represented by the leftmost diagram in Fig. 1, the parameter $\tau$ is interpreted as a complexified gauge coupling (which is common to all $M$ nodes in the quiver), $\rho$ as the (common) parameter which extends all gauge group nodes to their affine forms $\widehat{A}_{N-1}$ and $S$ as the mass of the bifundamental hypermultiplets. The remaining parameters [Kanazawa:2016tnt] (appearing in the form of $a_{ij}$ in (2.4)) encode further structure of the gauge- and matter spectrum, but the details shall not be important in the following. At the classical level, the theory inherits modular properties from the genus $2$ formulation (2.4). The functional basis (i.e. the particular set $\Theta\!\left[\begin{smallmatrix}i/M&j/N\\ 0&0\end{smallmatrix}\!\right]\!(z\,|\Omega)$ of genus $2$ theta functions) over which the curve (2.4) is written, corresponds to theta functions of polarised Abelian varieties [book:Igusa, birkenhake2004complex]. The modular group acting on the period matrix $\Omega$ is therefore the paramodular group defined as [birkenhake2004complex]:

$$\Sigma_{M,N}:=Sp(4,\mathbb{Q})\cap\begin{bmatrix}\mathbb{Z}&\mathbb{Z}&\mathbb{Z}/M&\mathbb{Z}/N\\ \mathbb{Z}&\mathbb{Z}&\mathbb{Z}/M&\mathbb{Z}/N\\ M\mathbb{Z}&M\mathbb{Z}&\mathbb{Z}&M\mathbb{Z}/N\\ N\mathbb{Z}&N\mathbb{Z}&N\mathbb{Z}/M&\mathbb{Z}\end{bmatrix}\,,$$

(2.5)

with,

$$Sp(4,\mathbb{Q})=\left\{\Lambda=\begin{bmatrix}A&B\\ C&D\end{bmatrix}\in\mathbb{M}_{4}(\mathbb{Q})\,|\,\Lambda\begin{bmatrix}0&\mathds{1}\\ \mathds{-1}&0\end{bmatrix}\Lambda^{T}=\begin{bmatrix}0&\mathds{1}\\ \mathds{-1}&0\end{bmatrix}\right\},$$

(2.6)

where $\mathbb{M}_{4}(\mathbb{Q})$ is the space of $4\times 4$ matrices with rational entries and non-vanishing determinant. The group $\Sigma_{M,N}$ acts in the following way on the period matrix $\Omega$:

$$\Omega\to(A\Omega+B)(C\Omega+D)^{-1}\,,\quad\forall\begin{bmatrix}A&B\\ C&D\end{bmatrix}\in\Sigma_{M,N}\,.$$

(2.7)

It is interesting to note that the linearly acting subgroup of the paramodular group (i.e. the subgroup with $B=C=0$ in (2.7)) generates the web of dualities in LSTs that was described at the level of the instanton partition function in [Hohenegger:2016yuv, Bastian:2017ary, Bastian:2017ing]. In this respect, the group $\Sigma_{M,N}$ is a generalisation of the paramodular group found in [Bastian:2019hpx, Bastian:2019wpx] for generic $M,N>1$.

The $(\widehat{A}_{M-1},N)$ theory was shown to be dual to $(\widehat{A}_{M^{\prime}-1},N^{\prime})$ for any $N^{\prime},M^{\prime}\in\mathbb{N}$ such that $N^{\prime}M^{\prime}=NM$ and $\mathrm{gcd}(N^{\prime},M^{\prime})=\mathrm{gcd}(N,M)=:p$. The corresponding duality map at the level of the period matrix $\Omega$, was formulated in [Hohenegger:2016yuv] in terms of the following $Sp(4,\mathbb{Z})$ transformation

$$A=\begin{bmatrix}\frac{M}{p}&-\frac{N}{p}\\ 1&-1\end{bmatrix}\,,\quad\Omega\to(A^{T})^{-1}\Omega A^{-1}=\begin{bmatrix}\frac{\tau^{\prime}}{M^{\prime}}&S^{\prime}\\ S^{\prime}&\frac{\rho^{\prime}}{N^{\prime}}\end{bmatrix}\,.$$

(2.8)

This transformation acts naturally on the basis of genus 2 theta functions $\Theta\!\left[\begin{smallmatrix}i/M&j/N\\ 0&0\end{smallmatrix}\!\right]\!(z\,|\Omega)$ (for $i\in\{0\ldots M-1\}$ and $j\in\{0\ldots N-1$) over which (2.4) is formulated: using (A.2) and noticing the equivalence of the following sets

	$\displaystyle\Bigg{\{}A\cdot\left(\frac{i}{M},\frac{j}{N}\right),$	$\displaystyle\,0\leq i\leq M-1\,,\,0\leq j\leq N-1\Bigg{\}}$
		$\displaystyle\cong\Bigg{\{}\left(\frac{k}{p}\!\!\!\!\mod 1,\frac{lp}{MN}\!\!\!\!\mod 1\right),\,0\leq k\leq p-1\,,\,0\leq l\leq\frac{MN}{p}-1\Bigg{\}}\,,$		(2.9)

it indeed follows that (2.8) (along with a suitable duality map of the remaining parameters $\{a_{ij}\}$) leaves the SWC (2.4) invariant.

First of all, contrary to $\widehat{A}$-type, $\widehat{D}$-type LSTs do not allow for a mass-deformation parameter [Haghighat:2017vch], i.e. there is no equivalent to the parameter $S$ in the period matrix in (2.4). Therefore, also the genus 2 theta functions $\Theta$ are effectively reduced to two types of genus $1$ (i.e. Jacobi) theta functions. Secondly, the transverse $\widehat{D}$-type orbifold itself has two important effects, which are of importance to the SWC:

• (i)

  the orbifold gives rise to a $\mathbb{Z}_{2}$ symmetry, under which the SWC is invariant [Haghighat:2018dwe]

• (ii)

  $M5$-brane fractionalisation: as explained in [DelZotto:2014hpa] due to the orbifold, a single $M5$-brane splits into two ‘half-branes’ on the $\mathbb{S}_{6}^{1}$ circle of Table 2.1. At the level of the quantum states introduced in [Haghighat:2018gqf], each $M5$ state can be described by the product of two $1/2$ $M5$-brane states.

Finally, there is another point of view, which provides crucial information for the SWCs, namely a dual description of the ($\widehat{D}_{M},1)$ LSTs in terms of an $Sp(M-4)$–$SO(2M)$ quiver gauge theory, which is schematically shown in Figure 2. Indeed, for $M=4$ this duality was first argued in [Kim:2017xan] at the level of the partition function and in [Haghighat:2018dwe] at the level of the SWC using fiber-base duality, while for general $M\geq 4$, it was established at the level of the partition function in [Kim:2017xan].

To understand the implications of such a duality at the level of the SWC, we remark that, in the absence of a mass-deformation, curves with $\widehat{A}_{1}$ base can generally be written in the following form [Haghighat:2018dwe, Haghighat:2018gqf]:

$$0=\theta_{3}^{2}(0;\rho)\theta_{2}^{2}(0;\rho)\theta_{1}^{2}(z_{2};\rho)c_{1}(\tau)F_{1}(z_{1};\tau)-\theta_{4}^{2}(z_{2};\rho)F_{2}(z_{1};\tau),$$

(3.1)

where the Jacobi theta functions $\theta_{2,3}$ are defined in (A.1), $c_{1}(\tau)$ is a modular constant and $F_{1,2}$ correspond to elliptic gauge polynomials⁷⁷7The gauge polynomial is the polynomial whose roots are the Coulomb branch moduli. Elliptic gauge polynomials are their natural elliptic uplift. of each of the two nodes of the (right) quiver in Figure 2. For example, in [Haghighat:2018dwe], these polynomials have been found for $SO(8)$ and $Sp(0)$ nodes to be respectively

$\displaystyle F_{1}(z_{1};\tau)=\prod_{i=1}^{4}\frac{\theta_{1}(z_{1}-a_{i}^{SO};\tau)\theta_{1}(z_{1}+a_{i}^{SO};\tau)}{\theta_{1}^{2}(z_{1};\tau)}\,,$

and

$\displaystyle F_{2}(z_{1};\tau)=\frac{\theta_{1}^{2}(2z_{1};\tau)}{\theta_{1}^{8}(z_{1};\tau)}\,.$

(3.2)

We are now ready to discuss the general form of the SWC of the $(\widehat{D}_{M},1)$-LST. This curve is doubly elliptic and we shall denote the corresponding parameters by $\tau$ and $\rho$, while the coordinates shall be called $z_{1}$ (the coordinate on the $\tau$-parameterised elliptic curve) and $\mathbb{Z}_{2}$ even in $z_{2}$ (the coordinate on the $\rho$-parameterised elliptic curve). Following the discussion of the previous Subsection, we shall now provide a general ansatz for the $(\widehat{D}_{M},1)$-LST, which is based on the following principles:

1. 1.

   The SWC is independently $\mathbb{Z}_{2}$ even in $z_{1}$ and $z_{2}$. This symmetry is due to the $\mathbb{Z}_{2}$ symmetry of the $\widehat{D}$-type orbifold, explained in point (i) of the previous Subsection. The fact that it (a priori) can act independently on $z_{1}$ and $z_{2}$ is a consequence of the absence of a mass deformation in the $\widehat{D}$-type LSTs.

2. 2.

   Using (pairs of) genus 1 theta functions as building blocks, the general form of the $\widehat{D}$-type-LST SWC can be cast in the following form

   $$\sum_{0\leq i,j\leq M}\widetilde{T}_{i}(z_{2},\rho,\tau)\,\mathbb{M}_{ij}(\tau)\,\Phi_{j}(z_{1},\tau)=0\,.$$

   (3.3)

   Indeed, here $\{\Phi_{j}\}_{j\in\{0\ldots M\}}$ is a basis of $\mathbb{Z}_{2}$ even theta functions of degree $2M$ (defined in (A.7)) and $\widetilde{T}_{i}$ are $\mathbb{Z}_{2}$ even combinations of elliptic Weyl characters of $SU(1)$ and $SU(2)$ [Haghighat:2017vch, Nekrasov:2012xe], concretely

   $$\widetilde{T}_{i}(z_{2},\rho)=\begin{cases}\alpha_{i}\theta_{1}^{2}(z_{2};\rho),&\text{for }i\in\{0,1,M-1,M\}\,,\\ \alpha_{i}\theta_{1}^{2}(z_{2};\rho)+\beta_{i}\theta_{4}^{2}(z_{2};\rho),&\text{for }i\in\{2\ldots M-2\}\,,\end{cases}$$

   (3.4)

   with $\theta_{1}$ and $\theta_{4}$ defined eq. (A.1). $\alpha_{i},\beta_{i}$ are $2M-2=h^{\vee}(D_{M})$, which are independent of $z_{1,2}$, but may depend on $\tau$ and $\rho$ (as well as other Coulomb moduli of the $\widehat{D}_{M}$-LST).

3. 3.

   The form (3.3) of the SWC must permit an alternative decomposition either on sections of line bundles with polarisation $(2,2M)$ or on products of sections of line bundles with polarisation $(1,M)$. As detailed in [Haghighat:2018gqf], this is interpreted in term of quantum states as a consequence of the $M5$-brane fractionalisation discussed in point (ii) of the previous Subsection. This can alternatively be motivated by analysing the 4d SWC of the nodes involved in the dual construction (corresponding to the quiver on the right of Fig. 2). The gauge polynomial of $SO(2M)$ and $Sp(M-4)$ gauge theories can be decomposed into the product of two degree $M$ polynomials [DHoker:1996kdj]. We assume that this structure stems from a decomposition in term of sections of line bundle as explained above in the SWC of the LST.