Spectral Abscissa Optimization

• Spectral Abscissa Optimization is a technique that focuses on minimizing the largest real eigenvalue of a system’s matrix to improve stability.
• It employs gradient-based and non-smooth optimization methods to effectively navigate the non-convex landscape inherent in eigenvalue problems.
• This approach finds applications in control theory, structural dynamics, and network stability, ultimately enhancing safety and performance in engineering systems.

The second orbifold Chern class is a key invariant governing the geometry of complex analytic varieties and pairs with singularities, particularly in the context of varieties with quotient singularities or more general klt (“Kawamata log terminal”) singularities and pairs equipped with orbifold boundary divisors. It generalizes the classical second Chern class to singular and orbifold settings, encoding both the global topology and the local structure of quotient singularities and ramification data. The theory intertwines methods from algebraic geometry, complex geometry, and analytic approaches, yielding powerful applications to Chern class inequalities, moduli questions, and the structure of quotient and orbifold varieties.

1. Geometric Setting and Definitions

The second orbifold Chern class arises in several closely related frameworks:

• Let $X$ be a normal compact complex analytic variety (or a complex projective variety) of dimension $n$, with quotient singularities in codimension two, or more generally with klt singularities.
• Consider a boundary divisor $D = \sum_j d_j D_j$ on $X$ with $0 < d_j \leq 1$ and coefficients encoding orbifold ramification, so $(X, D)$ is an orbifold pair.
• For the sheaf-theoretic perspective, let $\mathcal{F}$ be a reflexive coherent sheaf of rank $r$ on $X$ (frequently $\mathcal{F} = \Omega^1_X(\log D)$).
• Let $\omega$ be a Kähler form on $X$ in analytic contexts.

Orbifold Modification: To precisely define orbifold Chern classes, one introduces an “orbifold modification” $f: Y \to X$, where $Y$ is a normal complex space with at worst quotient singularities and $f$ is a proper, bimeromorphic map that is an isomorphism outside a closed subset of codimension at least three. On $Y$, $f^*\mathcal{F}$ modulo torsion becomes locally free.

Definition of Second Orbifold Chern Class: The second orbifold Chern class of $\mathcal{F}$—denoted $c_2^{\mathrm{orb}}(\mathcal{F})$—is defined in homology by the pushforward: $$c_2^{\mathrm{orb}}(\mathcal{F}) := f_*\left(c_2(f^*\mathcal{F}/\text{tors})\right) \in H_{2n-4}(X, \mathbb{Q}).$$ This construction is independent of the choice of orbifold modification $f$ and recovers the classical Chern class when $X$ is smooth and $\mathcal{F}$ is locally free (Ou, 25 Dec 2025).

2. Analytic and Intersection-Theoretic Constructions

Complex Analytic Description: The analytic approach to $c_2^{\mathrm{orb}}$ employs orbifold or singular Hermitian metrics, adapted to both the singularities of $X$ and the orbifold structure introduced by $D$. For a reflexive sheaf $\mathcal{E}$ (for instance, a $\mathbb{Q}$-sheaf), the curvature tensor is well defined on the smooth locus with transversal slice arguments along the singularities, and the Chern-Weil forms extend as closed positive currents with locally integrable coefficients. The current associated to the second orbifold Chern class is defined (in the Bott–Chern or de Rham sense) by pulling back to a resolution or orbifold modification and descending via pushforward, yielding a cohomology class: $$c_2^{\mathrm{orb}}(\mathcal{E}) = [\operatorname{tr}(\Theta^2) - \operatorname{tr}(\Theta) \wedge \operatorname{tr}(\Theta)]$$ where $\Theta$ is the orbifold-adapted curvature 2-form, and the representative is computed on $Y$ and pushed forward to $X$ (Guenancia et al., 13 Jan 2026).

Intersection-Theoretic Calculation: If $X$ is projective (or compact Kähler), the intersection number of $c_2^{\mathrm{orb}}(\mathcal{F})$ with a product of ample (or Kähler) classes $\omega$ is given by

$$\int_X c_2^{\mathrm{orb}}(\mathcal{F}) \wedge \omega^{n-2}$$

This generalizes the intersection pairing for smooth varieties.

3. Main Theorems and Inequalities

3.1. Orbifold Bogomolov–Gieseker Inequality

A central result states the following generalization of the Bogomolov–Gieseker inequality to orbifold settings, under stability hypotheses for a reflexive sheaf $\mathcal{F}$ with respect to a Kähler class $\omega$:

Theorem (Orbifold Bogomolov–Gieseker).

Let $X$ be a compact Kähler variety with quotient singularities in codimension two, and $\mathcal{F}$ an $\omega$-stable reflexive sheaf of rank $r$ on $X$. Then

$$2r \cdot c_2^{\mathrm{orb}}(\mathcal{F}) \cdot [\omega]^{n-2} \geq (r-1) \cdot c_1^{\mathrm{orb}}(\mathcal{F})^2 \cdot [\omega]^{n-2}.$$

Here, $c_1^{\mathrm{orb}}$ and $c_2^{\mathrm{orb}}$ are the orbifold first and second Chern classes as above, and intersection products denote pairing with suitable powers of the Kähler class (Ou, 25 Dec 2025, Guenancia et al., 13 Jan 2026).

3.2. Pseudo-Effectivity and Positivity

For orbifold pairs $(X,D)$, the second orbifold Chern class $c_2(X,D)$ satisfies:

Theorem (Pseudo-Effectivity for Movable Log Canonical Class).

Let $(X, D)$ be a threefold with mild singularities (klt and projective), and suppose that $K_X + D$ is movable. Then $c_2(X,D)$ is pseudo-effective, i.e.

$$c_2(X,D) \cdot A \geq 0 \quad \text{for every ample divisor } A$$

This generalizes Miyaoka's pseudo-effectivity theorem for the second Chern class of minimal models (Rousseau et al., 2016).

3.3. Characterization of the Equality Case

The case of equality in the orbifold Bogomolov-Gieseker inequality is precisely characterized: if equality holds for $\mathcal{E}$ stable, then $X$ is a finite (global) quotient of an abelian variety by a finite group acting freely in codimension one, and $D = 0$ (Guenancia et al., 13 Jan 2026). This provides a direct numerical and geometric criterion for the global-quotient case.

4. Explicit Formulas and Computations

Given an orbifold pair $(X,D)$, for example with $D = \sum_{i=1}^k (1-1/m_i)D_i$, the orbifold Chern classes can be calculated via log geometry: $$c_1(X,D) = c_1(X) - \sum_i \left(1 - \frac{1}{m_i} \right) [D_i]$$

$$c_2(X,D) = c_2(X) - \sum_i \left(1 - \frac{1}{m_i}\right) c_1(D_i) + \sum_{i<j} \left(1 - \frac{1}{m_i}\right)\left(1 - \frac{1}{m_j}\right) [D_i] \cdot [D_j]$$

where $c_1(X)$, $c_2(X)$ denote the usual Chern classes of the smooth locus, and $[D_i]$ the classes of the boundary components. Local contributions at points with nontrivial isotropy groups are computed via orbifold structure sheaves and local group actions (Rousseau et al., 2016).

In homological terms, for a reflexive sheaf $\mathcal{F}$, the orbifold second Chern class is: $$c_2^{\mathrm{orb}}(\mathcal{F}) = f_*( c_2(f^* \mathcal{F} / \text{tors}) )$$ with $f:Y\to X$ an orbifold modification, $f^*\mathcal{F} / \text{tors}$ locally free on $Y$, and $f_*$ the pushforward (Ou, 25 Dec 2025).

5. Applications, Examples, and Characterizations

5.1. Torus Quotients and Vanishing Criteria

In the case of compact complex threefolds with klt singularities, the vanishing of both the first and second orbifold Chern classes characterizes global torus quotients: $$c_1^{\mathrm{orb}}(X) = 0, \quad c_2^{\mathrm{orb}}(X) = 0 \implies X \simeq T/G$$ for a complex torus $T$ and $G$ a finite group acting holomorphically, freely in codimension one (Graf et al., 2017).

5.2. Minimal Models, Effective Non-Vanishing, Finiteness of Subvarieties

The pseudo-effectivity of $c_2(X,D)$ for movable $K_X+D$ yields, as corollaries, solutions to the effective non-vanishing problem (Kawamata's conjecture) and sharp restrictions on the number and type of minimal subvarieties of general type contained in $X$:

• If $K_X+H$ is nef and $H$ is ample, $H^0(X,K_X+H) \neq 0$ (Rousseau et al., 2016).
• Only finitely many Fano, Calabi–Yau, or Abelian subvarieties of codimension one, mildly singular, whose classes lie in the movable cone, can occur.

6. Relation to Birational and Homological Chern Classes

For group actions free in codimension two, a "birational" second Chern class can be defined, which captures Chern class data sensitive to deeper quotient singularities or resolutions: $$c_2^{\text{bir}}(\mathcal{F}) := \lim_{f} f_*\left( c_2(f^*\mathcal{F}/\mathrm{tors}) \right)$$ where the limit runs over suitable modifications (Graf et al., 2017). Both orbifold and birational second Chern classes relate to the Schwartz–MacPherson Chern class for singular spaces, and generalize classical smooth Chern class theory via pushforward and functoriality properties.

7. Broader Context: Chern Class Inequalities and Moduli

The extension of Chern class inequalities, such as the Miyaoka–Yau and Bogomolov–Gieseker inequalities, to orbifold and singular settings has enabled new advances in the study of the minimal model program, moduli of varieties of general type, and the classification of special varieties with vanishing or numerically trivial Chern classes. These orbifold invariants provide numerically effective and topologically meaningful constraints even in the presence of quotient and log singularities, guiding recent progress in birational classification, abundance conjectures, and the geometry of Kähler and projective varieties with mild or canonical singularities (Rousseau et al., 2016, Graf et al., 2017, Ou, 25 Dec 2025, Guenancia et al., 13 Jan 2026).