Real Delta-Invariant in Algebraic Geometry

• Real delta-invariant is a numerical measure in algebraic and arithmetic geometry that captures analytic, geometric, and arithmetic data via Arakelov theory and theta function integrals.
• Explicit formulas provide effective bounds and stability thresholds for Riemann surfaces, log Fano varieties, and projective bundles, linking curvature formulas to moduli and K-stability.
• Current research extends the invariant to higher-dimensional varieties and principally polarized abelian varieties, refining criteria for K-stability, moduli compactifications, and arithmetic inequalities.

The real delta-invariant, denoted $\delta$, is a fundamental numerical invariant arising in several advanced settings in algebraic and arithmetic geometry. Its principal incarnations are (i) the $\delta$-invariant of Riemann surfaces or principally polarized abelian varieties, known as Faltings’ delta-invariant, and (ii) the valuative $\delta$-invariant of log Fano varieties and polarized varieties, which governs K-stability notions via thresholds on divisorial valuations. These seemingly disparate contexts are connected through Arakelov theory, birational geometry, and the study of moduli. The real delta-invariant encodes subtle analytic, geometric, and arithmetic data, and plays a crucial role in curvature formulas, stability thresholds, effective bounds, and moduli-theoretic applications.

1. Classical Delta-Invariant of Riemann Surfaces

For a compact connected Riemann surface $X$ of genus $g \ge 1$, the real delta-invariant $\delta(X)$ is defined via Arakelov geometry, involving the interplay of the Arakelov-Green function, theta functions, and the period matrix. Wilms established an explicit formula $$\delta(X) = -24\,H(X) + 2\,\varphi(X) - 8g\log(2\pi)$$ where $H(X)$ is the theta-integral over the Jacobian and $\varphi(X)$ is the Zhang–Kawazumi invariant, given by the double integral of the curvature form of the diagonal bundle (Wilms, 2016). These quantities are constructed as follows:

• Arakelov–Green Function: $G(P, Q)$, satisfying key invariance, normalization, and differential equations, encodes the canonical metric on $X$.
• Theta Function Norm and Integral: The Riemann theta function $\theta(z;\Omega)$ (with $\Omega$ the period matrix) is paired with a normalization to define $|\,\theta\,|(z)$. $H(X)$ is then an integral of the logarithm of this norm over the Jacobian.
• Zhang–Kawazumi Invariant: $\varphi(X) = \int_{X \times X} h^2$, with $h$ the curvature of the metric on the universal line bundle of the diagonal.

The delta-invariant emerges in Faltings' proof of the Mordell Conjecture, providing an “archimedean correction” term in the arithmetic Noether formula.

2. Explicit Formulas and Bounds in Special Cases

For hyperelliptic Riemann surfaces, precise closed formulas for $\delta(X)$ have been derived, expressing $\delta(X)$ in terms of branch points and theta-constants. Concretely, if $X \to \mathbb P^1$ is branched at $2g+2$ points $e_1, \ldots, e_{2g+2}$, then

$$\delta(X) = A_g + \frac{1}{2}\sum_{i<j}\log|e_i-e_j| - \frac{1}{2}\sum_{\epsilon}\log|\Theta_{\epsilon}(e_1,\ldots,e_{2g+2})|$$

where $A_g$ is an explicit genus-dependent constant and $\Theta_\epsilon(-)$ are products associated to half-integer characteristics (Wilms, 2015). Uniform lower bounds are established: for explicit constants $c_1, c_2 > 0$, $\delta(X) \ge c_1 g - c_2$, concretely, for example, $c_1 = 0.1, c_2 = 2$. This yields effective control over $\delta$ as genus grows. Furthermore, the Arakelov self-intersection $\omega^2$ of the relative dualizing sheaf of a hyperelliptic curve over a number field is bounded above by a function of $\delta$ plus the (normalized) discriminant, and improvements to Szpiro-type inequalities are obtained (Wilms, 2015).

3. Delta-Invariant for Log Fano and K-Stability

On a normal projective variety $X$ with a $\mathbb Q$-divisor $\Delta$ (log Fano pair), the real $\delta$-invariant is defined via an infimum over all prime divisors $E$ over $X$ of the ratio $\frac{A_{X,\Delta}(E)}{S_{X,\Delta}(E)}$, where $A_{X,\Delta}(E)$ is the log discrepancy and $S_{X,\Delta}(E)$ is the expected vanishing (volume) integral: $$S_{X,\Delta}(E) = \frac{1}{(-K_X-\Delta)^n} \int_0^{+\infty} \operatorname{vol}\left(f^*(-K_X-\Delta)-tE\right)\, dt$$

$$\delta(X, \Delta) = \inf_E \frac{A_{X,\Delta}(E)}{S_{X,\Delta}(E)}$$

This invariant governs K-semistability ($\delta \geq 1$) and uniform K-stability ($\delta > 1$)—central in the theory of Fano varieties and the existence of Kähler–Einstein metrics (Zhou, 2020, Ammar et al., 2024).

4. Approximation, Valuations, and Local Delta-Invariants

The $\delta$-invariant admits refined local and valuative versions. On a Fano surface (e.g., degree $2$ del Pezzo), the local stability threshold at a point $p$,

$$\delta_p(X) = \inf_{E: \text{center}_X(E) \ni p} \frac{A_X(E)}{S(-K_X;E)}$$

encapsulates local K-stability phenomena. Classification for degree $2$ del Pezzo surfaces demonstrates that for nine point-types, all but one case yield rational local $\delta$-values; $\delta_p(X)$ is irrational if and only if $p$ lies on a unique $(-1)$-curve, with explicit irrational value $\frac{6}{71}(11 + 8\sqrt{3})$ (Alberdi, 2023).

The global $\delta$-invariant can be approximated by divisorial valuations associated to log canonical places of plt complements. Given $\delta(X,\Delta) \le 1$, there exists a sequence of complements and divisors realizing the infimum in the definition of $\delta$ (Zhou, 2020). Under further (conjectural) boundedness properties, the infimum is achieved by a single prime divisor.

5. Delta-Invariant and Explicit Models: Projective Bundles

The delta-invariant has been explicitly computed for all ample line bundles on projective bundles $\mathbb P(E)$ over a smooth curve $C$, where $E$ has specified Harder–Narasimhan type. When $E$ is strictly semistable, for any ample $L \equiv a\xi + b f$, it holds that

$$\delta(\mathbb P(E), L) = \min\left\{ \frac{2}{a\mu + b}, \frac{n}{a} \right\}$$

where $\mu$ is the slope of $E$ and $n = \operatorname{rank} E$ (Ammar et al., 2024). A more general two-step Harder–Narasimhan case yields further explicit rational formulas. This enables a direct identification of the regions in the ample cone which yield K-semistability, tying together algebraic and differential geometric concepts (e.g., cscK metrics, slope stability of $E$).

6. Asymptotic, Extension, and Moduli Aspects

The delta-invariant extends canonically to indecomposable principally polarized abelian varieties (PPAVs), with explicit theta-integral and theta-derivative formulas, aligning with the Riemann surface case on Jacobians (Wilms, 2016). As genus $g \to \infty$, $\delta(X)$ grows at most linearly with $g$, and degenerations of PPAVs yield logarithmic singularities in $\delta$, matching geometric intuition for nodal degenerations.

Progress on explicit lower and upper bounds for $\delta$, realization by optimal divisors, and connections to the geometry of the dual complex of lc places have led to refinements in K-moduli compactifications, arithmetic inequalities, and the explicit classification of K-unstable Fano varieties.

7. Significance and Current Research Directions

The real delta-invariant serves as a bridge between analysis on algebraic curves, Arakelov theory, birational geometry, and the moduli theory of Fano varieties. Its analytic definition via theta functions and Green functions has led to effective control over Arakelov invariants and explicit height inequalities. In birational and K-stability theory, the valuative infimum criterion enables finite reduction of the stability problem, the construction of optimal destabilizers, and explicit criteria for moduli functors. Continuing research focuses on boundedness issues, the structure of possible irrational local invariants, further refinements for singular and higher-dimensional cases, and applications to arithmetic and geometric moduli problems (Wilms, 2016, Wilms, 2015, Zhou, 2020, Alberdi, 2023, Ammar et al., 2024).