---
title: Generalized Schur Index
url: https://www.emergentmind.com/topics/generalized-schur-index
type: topic
---

# Generalized Schur Index

Searching arXiv for recent papers directly relevant to the generalized Schur index and closely related Schur-index generalizations.
The generalized Schur index, more precisely the generalized Schur partition function \(\hat{\mathcal Z}(q,\alpha)\), is a one-parameter deformation of the Schur limit of the four-dimensional \(\mathcal N=2\) superconformal index, obtained through a double-scaled limit of the full index in which the ordinary Schur index is recovered at \(\alpha=1\) [2506.13764]. In the formulation introduced in 2025, it interpolates between the ordinary Schur index and a Coulomb-branch-like limit, and its coefficients in the \(q\)-expansion depend continuously on a parameter \(\alpha\) [2506.13764]. Earlier literature did not use the same name, but developed several structurally related generalizations of the Schur index: flavor refinements, loop-decorated indices, orbifold Schur indices on \(S^3/\mathbb Z_n\times S^1\), and exact grand-canonical or modular reformulations [2208.01426] [1510.02480] [1710.08853]. In current usage, the phrase therefore has both a narrow sense—\(\hat{\mathcal Z}(q,\alpha)\) as defined in 2025—and a broader sense encompassing Schur-sector observables modified by extra parameters, defects, or background geometry.

## 1. Definition and limiting construction

The starting point is the standard \(4d\ \mathcal N=2\) superconformal index on \(\mathbb S^3\times \mathbb S^1\),
\[
{\cal I} = \mathrm{Tr}_{\mathbb S^3} \,(-1)^F \left(\frac{qp}{t}\right)^{-r} p^{j_2-j_1}\, q^{j_1+j_2}\, t^{R}\, \prod_{i=1}^{\mathrm{rank}\,G_F}u_i^{F_i},
\]
with \(r\) the \(U(1)_r\) charge, \(R\) the Cartan of \(SU(2)_R\), \(j_1,j_2\) the Cartans of the \(SU(2)\times SU(2)\) isometry of \(\mathbb S^3\), and \(F_i\) flavor Cartan charges [2506.13764]. The ordinary Schur index is obtained by setting \(q=t\), giving
\[
{\cal I}_S = \mathrm{Tr}\,(-1)^F\,q^{R+j_1+j_2}\, \prod_{i=1}^{\mathrm{rank}\,G_F}u_i^{F_i},
\]
and the contributing operators obey the Schur shortening conditions
\[
E-(j_1+j_2)-2R=0,\qquad r+j_1-j_2=0
\]
[2506.13764].

The generalized Schur partition function is defined by a double-scaled limit that probes the order-of-limits interpolation between the Schur specialization \(q=t\) and the specialization \(qp=t\), the latter being associated in the paper with turning on masses \(m\,Q\widetilde Q\) or Coulomb-branch vevs \(\langle \mathrm{Tr}\,\Phi^k\rangle\neq 0\) [2506.13764]. The limit is
\[
p=1-\epsilon,\qquad t=(qp)^{1+\frac{\alpha}{\log q}\epsilon},
\]
followed by \(\epsilon\to 0\) [2506.13764]. For Lagrangian theories, after stripping off the universal pole singularity, the resulting partition function is
\[
{\cal Z}(q,\alpha) = (q;q)^{2\,\mathfrak r_G} \oint d\boldsymbol\zeta_G\, \left( \frac{ \prod_{\beta}(1-e^{\beta(\zeta)})(q e^{\beta(\zeta)};q)^2 }{ \prod_{\rho}(q^{1/2}e^{\rho(\zeta)};q) } \right)^\alpha
\]
[2506.13764]. Here \(\mathfrak r_G\) is the rank of the gauge group, \(\beta\) runs over nonzero roots, \(\rho\) runs over weights of the matter representation, and \(d\boldsymbol\zeta_G\) is the gauge Haar measure in fugacity variables [2506.13764].

At \(\alpha=1\), this is exactly the standard Schur matrix-integral formula, while at \(\alpha=0\) one obtains
\[
{\cal Z}(q,0)=(q;q)^{2\mathfrak r_G},
\]
identified in the paper with the Schur index of free \(U(1)\) vectors on a generic Coulomb-branch locus [2506.13764]. The generalized quantity therefore interpolates continuously between the Coulomb-generic answer and the ordinary Schur index. The normalized object \(\hat{\mathcal Z}(q,\alpha)\) has a standard \(q\)-expansion,
\[
\hat{\mathcal Z}(q,\alpha)=\sum_{n\ge 0} c_n(\alpha)\, q^n,
\]
whose coefficients depend continuously on \(\alpha\) and are, in general, not integer [2506.13764]. The paper is explicit that for non-integer \(\alpha\) the resulting \(q\)-series coefficients are “in general not integer,” so \(\hat{\mathcal Z}(q,\alpha)\) is not generally an index in the strict counting sense [2506.13764].

## 2. Relation to the ordinary Schur index and earlier Schur-type generalizations

The generalized Schur partition function is a recent construction, but it sits within a larger family of Schur-sector observables. The ordinary Schur index itself is a one-fugacity specialization of the \(4d\) superconformal index, often written as
\[
\mathcal I_{\rm Schur}(q) = \operatorname{Tr}_{\mathcal H(S^3)} (-1)^F\, q^{E-R},
\]
with Schur operators contributing when
\[
\delta \equiv E-2j_2-2R+r=0
\]
[1507.08659]. This quantity is protected, depends only on \(q\), and simplifies the special-function structure from elliptic gamma functions to theta functions [1507.08659].

Several earlier works extended the Schur framework without introducing \(\hat{\mathcal Z}(q,\alpha)\). One direct refinement is the flavored Schur index of \(4d\ \mathcal N=4\ U(N)\) SYM, equivalently the Schur index of \(4d\ \mathcal N=2^*\ U(N)\) SYM, with flavor fugacity \(t\) or \(\xi=q^{-1/2}t^2\) [2208.01426]. Its exact matrix-integral form is
\[
\mathcal{I}^{U(N)}(t;q) = \frac{1}{N!}\frac{(q)_{\infty}^{2N}}{(q^{\frac12} t^{\pm 2};q)_{\infty}^N} \oint_{|\sigma_i|=1} \prod_{i=1}^N \frac{d\sigma_i}{2\pi i\sigma_i} \frac{ \prod_{i\neq j} \left( \frac{\sigma_i}{\sigma_j};q \right)_{\infty} \left( q\frac{\sigma_i}{\sigma_j};q \right)_{\infty} } { \prod_{i\neq j} \left(q^{\frac12}t^{-2} \frac{\sigma_i}{\sigma_j};q \right)_{\infty} \left( q^{\frac12}t^{2} \frac{\sigma_i}{\sigma_j};q \right)_{\infty} }
\]
[2208.01426]. In that literature, the nearest analogue of a generalized Schur index is therefore a flavor- or mass-deformed Schur index.

Another class of generalizations inserts supersymmetric line operators. For \(\mathcal N=4\) SYM, the Schur index with Polyakov or Wilson loops is defined by inserting characters into the Schur matrix model,
\[
\mathcal I_R(N) =\frac{1}{\mathcal I(N)} \frac{q^{-N^2/4}\eta^{3N}(\tau)}{N!\pi^N} \int_0^\pi d^N \alpha\, \Tr_R(e^{2i\alpha}) \Tr_{\bar R}(e^{2i\alpha}) \frac{\prod_{i <j} \vartheta_1^2( \alpha_i-\alpha_j)} {\prod_{i,j} \vartheta_4(\alpha_i-\alpha_j)}
\]
[1510.02480]. This construction was described there as an “enrichment” of the index by loop operators, and it is a defect-sensitive protected Schur-sector observable.

A geometric generalization is the orbifold Schur index on
\[
S^3/\mathbb Z_n\times S^1,
\]
defined with an orbifold action modified by an \(SU(2)_R\times U(1)_r\) twist so that both Schur supercharges are preserved [1710.08853]. The orbifold Schur one-particle index is
\[
i_n^h(q,z) = \frac1n\sum_{k=0}^{n-1} i(\omega_n^k q,\omega_n^{kh}z),
\]
and the full orbifold Schur index of a Lagrangian theory is
\[
I_n = \sum_h e^{\varepsilon(h)} \int d\mu\ \mathrm{Pexp}\, i_n^h
\]
[1710.08853]. This orbifold construction reduces to the ordinary Schur index when \(n=1\).

These earlier directions show that “generalized Schur index” had a broad informal meaning even before 2025. The narrow modern usage, however, is tied to the double-scaled limit \(\hat{\mathcal Z}(q,\alpha)\) [2506.13764].

## 3. Exact structures, modularity, and computational frameworks

The generalized Schur partition function inherits from the Schur index a strong interplay with exact matrix models, Fermi-gas formalisms, and modular structures. For the ordinary Schur index of \(\mathcal N=4\) \(U(N)\) SYM, the matrix integral
\[
\mathcal{I}(N) =\frac{q^{-N^2/4}\eta^{3N}(\tau)}{N!\,\pi^N} \int_0^\pi d^N \alpha \, \frac{\prod_{i<j}\vartheta_1^2(\alpha_i-\alpha_j)} {\prod_{i,j}\vartheta_4(\alpha_i-\alpha_j)}
\]
was rewritten as the partition function of \(N\) non-interacting fermions on a circle [1507.08659]. The resulting grand index,
\[
\hat{\Xi}(\kappa) \equiv 1+\sum_{N=1}^\infty \mathcal I(N)\,q^{N^2/4}\,\kappa^N = \frac{1}{\vartheta_4} \left[ \vartheta_3\!\left(\arccos \frac{\kappa}{2}\right) + \vartheta_2\!\left(\arccos \frac{\kappa}{2}\right) \right],
\]
encodes exact finite-\(N\) data and large-\(N\) asymptotics [1507.08659]. The same Fermi-gas technology extends to loop-decorated indices [1510.02480], circular quiver Schur indices [1510.07041], and flavored \(\mathcal N=2^*\) Schur indices [2208.01426].

For flavored \(\mathcal N=2^*\) Schur indices, exact formulas are organized by spectral traces and Young diagrams. The canonical partition function is
\[
\mathcal{Z}(N,u;\xi;q) =\sum_{\lambda}(-1)^{N-r} \prod_{i=1}^{r} \frac{1}{\lambda_i^{m_i} (m_i !)} Z_{\lambda_i}(u;\xi;q)^{m_i},
\]
where \(\lambda\) is a partition of \(N\), and the full index is
\[
\mathcal{I}^{U(N)}(\xi;q)= \frac{(-1)^N \xi^{N^2/2} \theta(u;q)}{\theta(u\xi^{-N};q)} \mathcal{Z}(N;u;\xi;q)
\]
[2208.01426]. The spectral zeta functions are expressed in terms of twisted Weierstrass functions, and the normalized index lies in the polynomial ring generated by the Kronecker theta function and the Weierstrass functions which contains the polynomial ring of the quasi-Jacobi forms [2208.01426]. This suggests that generalized Schur-type quantities naturally organize into Jacobi and quasi-Jacobi structures.

The unflavored \(\mathcal N=4\) Schur index also exhibits modular anomaly equations. For \(SU(N)\), the unflavored exact Schur indices satisfy rank-recursive modular anomaly equations, such as
\[
\partial_{E_2} I_{2N+1} = \sum_{k=1}^N c_k\, I_{2N+1-2k}, \qquad
\partial_{E_2} I_{2N} = \sum_{k=1}^N c_k\, I_{2N-2k},
\]
with
\[
c_k=\frac{((k-1)!)^2}{(2k)!}
\]
[2205.00818]. In that setting, exact indices are reconstructed from the anomaly equation together with vanishing conditions. For non-\(A_n\) gauge groups \(B_n,C_n,D_n,G_2\), analogous unflavored Schur indices have also been computed using character expansion and Fermi-gas methods, but the modular anomaly equations become substantially more complicated [2311.08714].

A plausible implication is that the generalized Schur partition function \(\hat{\mathcal Z}(q,\alpha)\), though defined differently, enters a pre-existing web of Schur-sector exact structures rather than standing in isolation.

## 4. RG flows, special values of \(\alpha\), and theory-to-theory matching

A central claim of the generalized Schur construction is that \(\hat{\mathcal Z}(q,\alpha)\) is invariant, up to a nontrivial redefinition of \(\alpha\), under certain mass deformations, vev deformations, and Coulomb-branch flows between \(\mathcal N=2\) SCFTs [2506.13764]. The general form of the relation is
\[
\hat{\mathcal Z}_1(q,\alpha_1) = \hat{\mathcal Z}_2(q,\alpha_2(\alpha_1)),
\]
and in particular
\[
\hat{\mathcal Z}_{\mathcal T_2}(q,1) = \hat{\mathcal Z}_{\mathcal T_1}(q,\alpha)
\]
for certain pairs of theories \(\mathcal T_1,\mathcal T_2\) [2506.13764].

The paper proposes necessary conditions for such a relation. If
\[
\hat{\mathcal Z}_{\mathcal T_2}(q,1)=\hat{\mathcal Z}_{\mathcal T_1}(q,\alpha),
\]
then one should have
\[
c^{(2)}=c^{(1)}\alpha+\frac{1}{6}(1-\alpha)\mathfrak r,
\qquad
\Delta_i^{(2)}=(\Delta_i^{(1)}-1)\alpha+1,
\]
and, via the Shapere–Tachikawa relation,
\[
a^{(2)}=a^{(1)}\alpha+\frac{5}{24}(1-\alpha)\mathfrak r
\]
[2506.13764]. Here \(\mathfrak r\) is the Coulomb-branch rank. These formulas express the matching in terms of central charges and Coulomb scaling dimensions.

The best-known example is \(SU(2)\) \(\mathcal N=2\) SQCD with four flavors, denoted \(\mathfrak d_4\). Its normalized generalized Schur partition function is
\[
\hat {\cal Z}_{\mathfrak d_4}(q,\alpha) = \frac{(q;q)^2}{\mathtt N(\alpha)} \oint \frac{dz}{4\pi i z} \left( \frac{\Delta(z)(q z^{\pm2};q)^2}{(q^{1/2}z^{\pm1};q)^8} \right)^\alpha,
\]
with
\[
\Delta(z)=(1-z^2)(1-z^{-2}), \qquad \mathtt N(\alpha)=\oint\frac{dz}{4\pi i z}\,\Delta(z)^\alpha
\]
[2506.13764]. Special values of \(\alpha\) reproduce the ordinary Schur indices of all rank-one theories in the Deligne–Cvitanović series:
\[
\hat {\cal Z}_{\mathfrak d_4}(q,\tfrac15) = {\cal Z}_{\mathfrak a_0}(q,1),\quad
\hat {\cal Z}_{\mathfrak d_4}(q,\tfrac13) = {\cal Z}_{\mathfrak a_1}(q,1),\quad
\hat {\cal Z}_{\mathfrak d_4}(q,\tfrac12) = {\cal Z}_{\mathfrak a_2}(q,1),
\]
\[
\hat {\cal Z}_{\mathfrak d_4}(q,2) = {\cal Z}_{\mathfrak e_6}(q,1),\quad
\hat {\cal Z}_{\mathfrak d_4}(q,3) = {\cal Z}_{\mathfrak e_7}(q,1),\quad
\hat {\cal Z}_{\mathfrak d_4}(q,5) = {\cal Z}_{\mathfrak e_8}(q,1)
\]
[2506.13764]. Equivalently,
\[
\hat {\cal Z}_{\mathfrak g}(q,\alpha) = \hat{\mathcal Z}_{\mathfrak d_4}\!\left(q,\frac{h^\vee_{\mathfrak g}}{6}\,\alpha\right),
\]
where \(h^\vee_{\mathfrak g}\) is the dual Coxeter number [2506.13764].

Higher-rank examples exhibit analogous behavior. For \(SU(N)\) with \(2N\) fundamentals, the generalized partition function reproduces Schur indices of \(R_{2,2N-1}\) at \(\alpha=2\), \((A_1,D_{2N})\) at \(\alpha=\frac1N\), and \((A_1,A_{2N-1})\) at \(\alpha=\frac{1}{N+1}\) [2506.13764]. For \(USp(2N)\) with \(2N+2\) fundamentals, it reproduces Schur indices of \(D_2(SU(2N+1))\) at \(\alpha=\frac12\), \((A_1,D_{2N+1})\) at \(\alpha=\frac{1}{2N+1}\), and \((A_1,A_{2N})\) at \(\alpha=\frac{1}{2N+3}\) [2506.13764]. These results motivate the view that \(\hat{\mathcal Z}(q,\alpha)\) carries RG-flow-covariant information not visible in the ordinary Schur index alone.

## 5. Modular differential equations and quantum monodromy traces

A major development after the introduction of \(\hat{\mathcal Z}(q,\alpha)\) is the observation that, as a function of \(\alpha\), it appears to satisfy a modular linear differential equation of fixed order, with coefficients depending on \(\alpha\) [2512.02102]. This generalizes the well-known relation between ordinary Schur indices and modular differential equations in many VOA-associated examples.

For \(SU(2)\) with \(N_f=4\), the generalized Schur limit solves a second-order MLDE. Its \(q\)-series begins
\[
\hat {\mathcal Z}_{\mathfrak d_4}(q,\alpha) = 1+ \frac{2(-1+5\alpha)(1+6\alpha)}{1+\alpha}\,q
+ \frac{-2-3\alpha+29\alpha^2+150\alpha^3+1800\alpha^4} {(1+\alpha)(2+\alpha)}\,q^2
+ \cdots
\]
and it coincides with the hypergeometric modular expression
\[
K(q)^{\frac{1+6\alpha}{12}}\, {}_2F_1\!\left( \frac{1+6\alpha}{12}, \frac{5+6\alpha}{12}, 1+\alpha, K(q) \right),
\]
where
\[
j(q)=\frac{1728E_4(q)^3}{E_4(q)^3-E_6(q)^2}, \qquad K(q)=\frac{1728}{j(q)}
\]
[2512.02102]. The corresponding MLDE is
\[
D_q^{(2)}-5(6\alpha+1)(6\alpha-1)\,\mathbb E_4(q)
\]
[2512.02102].

Higher-rank examples exhibit fixed-order MLDEs as well: \(USp(4)\) with \(N_f=6\) gives a third-order MLDE, \(USp(6)\) with \(N_f=8\) a fourth-order MLDE, \(SU(3)\) with \(N_f=6\) a fourth-order twisted MLDE, and \(SU(4)\) with \(N_f=8\) a sixth-order MLDE [2512.02102]. The relevant modular group is the full modular group for ordinary integer-power \(q\)-series and \(\Gamma^0(2)\) for half-integer-power examples [2512.02102].

The same work also observes a relation between generalized Schur limits at certain negative integer values of \(\alpha\) and traces of higher powers of the quantum monodromy operator \(M(q)\). The ordinary Schur index is known to satisfy
\[
\mathcal I_{\mathrm{Schur}}(q) = (q;q)_\infty^{2\mathfrak r}\,\mathrm{Tr}\,M(q)^{-1}
\]
[1506.00265] [2512.02102]. For \((A_1,G)\) Argyres–Douglas theories with \(G=A_n,D_n\), the 2025 work proposes that, in examples,
\[
\hat{\mathcal Z}(q,\alpha) = (q;q)_\infty^{2\mathfrak r}\,\mathrm{Tr}\,M(q)^{-\alpha},
\qquad \alpha\in\mathbb Z_{\le 1}, \quad \alpha>\alpha^*
\]
[2512.02102]. Here \(M(q)\) is the quantum monodromy operator constructed from Coulomb-branch BPS data,
\[
M(q)=\prod_{\gamma\in\Gamma}^{\curvearrowleft}\Psi(q,X_\gamma),
\]
with
\[
\Psi(q,X_\gamma) = \prod_{i\ge 0}(1+q^{i/2+1}X_\gamma)
= \sum_{k\ge 0}\frac{q^{k^2/2}}{(q;q)_k}X_\gamma^k
\]
[2512.02102]. The quantum torus variables satisfy
\[
X_{\gamma_1}X_{\gamma_2} = q^{\frac{\langle\gamma_1,\gamma_2\rangle}{2}} X_{\gamma_1+\gamma_2}
= q^{\langle\gamma_1,\gamma_2\rangle} X_{\gamma_2}X_{\gamma_1},
\]
and the trace is defined by
\[
\mathrm{Tr}\,X_\gamma=
\begin{cases}
1,& \gamma=0,\\
0,& \gamma\neq 0.
\end{cases}
\]
[2512.02102].

For rank-one Deligne–Cvitanović theories, special negative rational values of \(\alpha\) yield integer \(q\)-series identified with VOA vacuum characters previously found from higher monodromy traces [2512.02102]. This suggests a broader Higgs-branch/Coulomb-branch correspondence: the generalized Schur limit is naturally related to the Schur/Higgs/VOA sector, whereas \(M(q)\) is constructed from Coulomb-branch BPS-wall-crossing data [2512.02102]. A plausible implication is that \(\hat{\mathcal Z}(q,\alpha)\) may provide a bridge between these sectors beyond the \(\alpha=1\) case, but this remains conjectural in the current literature.

## 6. Variants, applications, and open problems

A number of further constructions illuminate the broader landscape in which the generalized Schur index sits. One is the giant graviton expansion of the flavored Schur index of \(\mathcal N=4\ U(N)\) SYM. In the Schur limit, the finite-\(N\) index admits the exact expansion
\[
I^{U(N)}(u;q) = I^{\rm KK}(u;q) \sum_{n=0}^{\infty}\sum_{p=0}^{n} (uq)^{(n-p)N} I_{n-p}^{\rm D3}(u;q)\, q^{2(n-p)p}\, (u^{-1}q)^{pN}I_p^{\rm D3}(u^{-1};q),
\]
with
\[
I^{\rm KK}(u;q) = \frac{(q^2)_\infty}{(uq)_\infty (u^{-1}q)_\infty}
\]
[2403.06509]. A key feature is that the wrapped D3-brane index is an analytic continuation of the flavored Schur index of \(U(n)\) SYM:
\[
I_n^{\rm D3}(u;q) = I^{U(n)}\!\left(u^{-1/2}q^{-3/2};\,u^{-1/2}q^{1/2}\right)
\]
[2403.06509]. Near the unflavored point \(u=1\), the nontrivial functions appearing in the brane index are governed by quasimodular forms [2403.06509]. This is not a generalized Schur index in the 2025 sense, but it shows that exact finite-\(N\) Schur-sector data admits analytic continuation and modular completion.

The Schur index of \(\mathcal N=4\) SYM with more general gauge groups also extends the exact Schur program. For the unflavored Schur indices of \(B_n,C_n,D_n,G_2\) gauge groups, character expansion and Fermi-gas methods yield high-order \(q\)-series and many exact formulas [2311.08714]. For \(G_2\), one exact formula is
\[
\mathcal{I}_{G_2}(q)=
\left(\frac{\theta_4 }{\eta^3} \right)^{2}\left(\mathcal{I}_{A_1}^2-\frac{1}{288}(8E_2^{(3)}+\theta_2^4-8\theta_3^4+3\theta_2^4\theta_3^4)\right)
\]
[2311.08714]. This again underlines that Schur-sector observables are strongly constrained by modularity, but also that generalizations to other gauge groups may introduce more complicated anomaly structures.

The orbifold Schur index illustrates a different type of difficulty. On the ultraviolet side, the orbifold Schur index is straightforwardly defined for Lagrangian theories, but a proposed infrared generalization of the Cordova–Shao formula works for free hypermultiplets only when the background data are tuned to be orbifold invariant, and fails for theories with dynamical vector multiplets [1710.08853]. This example serves as a warning against assuming that every Schur-type generalization inherits the full exact structure of the ordinary Schur index.

Several open problems are explicit in the current literature. The generalized Schur partition function \(\hat{\mathcal Z}(q,\alpha)\) is defined by a limiting procedure rather than by a direct cohomological trace, and for generic \(\alpha\) it lacks an established operator-counting interpretation [2506.13764]. Its MLDE structure is conjectural beyond the computed examples [2512.02102]. The relation to higher monodromy traces is strongly suggestive but not derived from first principles [2512.02102]. The earlier Schur-index literature also points toward further extensions involving flavor fugacities, line operators, necklace quivers, and analysis beyond the Schur limit or to the lens-space index [1507.08659].

These developments suggest that the generalized Schur index, in the narrow sense of \(\hat{\mathcal Z}(q,\alpha)\), is best viewed as a protected partition function interpolating between Schur and Coulomb-like limits and exhibiting unexpected covariance under specific RG flows [2506.13764]. In the broader sense, it is part of a family of Schur-sector observables whose refinements by flavor, defects, background geometry, rank-generating variables, and modular reorganization continue to reveal nontrivial links among superconformal indices, chiral algebras, BPS wall-crossing, and exact spectral methods [1506.00265] [2208.01426] [1510.02480] [1710.08853] [2512.02102].

Source: https://www.emergentmind.com/topics/generalized-schur-index