Brill–Noether Number Overview

• The Brill–Noether number is a fundamental invariant that encodes the expected dimension of loci of line bundles and vector bundles with prescribed global sections on algebraic curves and related structures.
• It underpins existence results like the Kempf–Kleiman–Laksov and Griffiths–Harris theorems, predicting the non-emptiness and dimension of moduli spaces in smooth, singular, and combinatorial contexts.
• Extensions to higher dimensions and ranks incorporate corrections for singularities and irregularity, influencing modern approaches to moduli geometry and applications in algebraic and combinatorial settings.

The Brill–Noether number is a fundamental invariant arising in the study of special divisors and linear series on algebraic curves and, by extension, on more general discrete and geometric objects. It encodes the expected dimension of loci parameterizing line bundles (or vector bundles in higher rank settings) with prescribed numbers of global sections and plays a central role in the existence, structure, and geometry of Brill–Noether varieties. The definition and consequences of the Brill–Noether number extend from classical smooth projective curves to singular curves, combinatorial graphs, irregular varieties, and higher-rank vector bundle moduli. Throughout, the sign and value of the Brill–Noether number control non-emptiness, dimension, and genericity properties of associated moduli spaces.

1. Classical Definition and Geometric Significance

For a smooth projective curve $C$ of genus $g$, integers $d\ge0$ and $r\ge0$, the Brill–Noether locus

$$W^r_d(C) = \{\,L \in \mathrm{Pic}^d(C) : h^0(C, L) \ge r+1\,\}$$

parametrizes line bundles of degree $d$ admitting at least $r+1$ independent global sections—that is, $g^r_d$ linear series. The expected dimension of $W^r_d(C)$ is given by the Brill–Noether number

$\rho(g, r, d) = g - (r+1)(g-d+r)$

(Caporaso, 2011, Pflueger, 2013, Cotterill et al., 2023, Lange et al., 2017).

This integer controls whether special linear series exist on a general curve of genus $g$ and determines the dimension of their moduli spaces. The principal structure theorems are:

• Existence Theorem (Kempf–Kleiman–Laksov): If $\rho \ge 0$, then $W^r_d(C) \ne \varnothing$ for every smooth $C$.
• Brill–Noether Theorem (Griffiths–Harris): If $\rho < 0$, then for a general $C$, $W^r_d(C) = \varnothing$.

For the universal moduli space $W^r_{d,g}$ parametrizing triples $(C,L,V)$, the expected dimension is $\dim W^r_{d,g} = 3g-3 + \rho(g,d,r)$ (Pflueger, 2013). In the range $0 \le \rho \le g$, the locus is irreducible and surjects onto the moduli space of curves.

2. Brill–Noether Number in Combinatorial and Degenerate Settings

The Brill–Noether number is equally fundamental in combinatorial divisor theory on finite connected graphs $\Gamma$ of first Betti number $g$:

• Let $D \in \mathrm{Div}(\Gamma)$ be a divisor of degree $d$.
• Define the combinatorial rank $r_\Gamma(D)$ (with loop-sensitive refinements as needed).
• The combinatorial Brill–Noether locus

$$W^r_d(\Gamma) = \{\, [D] \in \mathrm{Jac}^d(\Gamma) : r_\Gamma(D) \ge r \,\}$$

mirrors the curve-theoretic setup, and is governed by the same Brill–Noether number $\rho(g,r,d)$ (Caporaso, 2011).

A key combinatorial analog holds: for $\rho \ge 0$, $W^r_d(\Gamma) \neq \varnothing$ for every $\Gamma$. Conversely, for $\rho < 0$ there exists $\Gamma$ with $W^r_d(\Gamma) = \varnothing$, and such a graph obstructs the existence of special divisors on general curves of genus $g$ (Caporaso, 2011).

Baker’s Specialization Lemma (and Caporaso’s refinements) formally relate the rank of divisors on curves and their specializations on graphs, underpinning the transfer of Brill–Noether theorems between geometric and combinatorial contexts.

3. Extensions: Irregular Varieties and Surfaces

On higher-dimensional varieties—especially surfaces of maximal Albanese dimension—the Brill–Noether number adapts to incorporate irregularity and intersection data. For a smooth projective surface $S$ of irregularity $q = h^1(\mathcal{O}_S)$, and a curve $C \subset S$ of arithmetic genus $p_a(C)$, the Brill–Noether loci

$$W^r(C, S) = \{\, \eta \in \mathrm{Pic}^0(S) : h^0(S, \mathcal{O}_S(C) \otimes \eta) \ge r+1 \,\}$$

have expected dimension

$\rho(C, r) = q - (r+1)(p_a(C) - C^2 + r)$

(Lopes et al., 2011).

Criteria for non-emptiness and dimension closely parallel the classical curve case: under mild hypotheses, if $\rho(C, r)\ge 0$, $W^r(C, S)\ne\varnothing$ and each component has dimension at least $\min\{ q,\,\rho(C, r) \}$.

This formulation provides lower bounds for $h^0(K_D)$ when $D$ moves linearly, and detailed inequalities for curves not moving in linear systems, extending the impact of the Brill–Noether number to the broader context of irregular and higher-dimensional varieties.

4. Negative Brill–Noether Number and Moduli Geometry

A central question is the behavior of $W^r_{d,g}$ when $\rho(g,d,r)<0$. Classical theory asserts generic emptiness, but significant subtleties emerge in moduli geometry:

• For not-too-large $|\rho|$, there exist irreducible components $Z\subset W^r_{d,g}$ of dimension $3g-3+\rho$, with image in $\mathcal{M}_g$ of codimension $-\rho$ (Pflueger, 2013).
• Pflueger establishes that for $0 < -\rho \le \frac{r}{r+2}g - 3r + 3$, such components arise (Theorem 1.1 of (Pflueger, 2013)).
• The construction depends on partition-theoretic combinatorics ("twisted Weierstrass points"), limit linear series on nodal curves, and explicit recursions controlling the “difficulty” $\epsilon(P)$ of the relevant partition—expected to be $O(1)$ for $|\rho|$ up to $g + C(r)$.

This suggests that in the negative regime, the geometry of Brill–Noether loci is richer than the naive dimension count indicates, and moduli components may exist even below the Brill–Noether line.

5. Brill–Noether Number for Singular and Cuspidal Curves

For curves with singularities—especially cusps—the Brill–Noether number receives corrections reflecting local constraints. For a curve $C_0$ with a cusp characterized by a numerical semigroup $S$, the expected dimension of $W^r_d(C_0)$ is governed not just by $\rho(g,r,d)$ but also by local ramification and non-linear combinatorial data from $S$: $\rho_S(g,r,d;\,(r_i)) = \rho(g,r,d) - \tau_S(r)$ where $\tau_S(r)$ comprises the sum of ramification losses and contributions from Betti elements of $S$ (Cotterill et al., 2023).

The effective Brill–Noether number

$$\rho_{\mathrm{eff}}(g,r,d;S) = \max_{(r_i)\in S,\,\sum r_i=d}\Big\{ \rho(g,r,d) - \text{ramification losses} - \text{semigroup-theoretic extra conditions} \Big\}$$

determines both non-emptiness and dimension, establishing a direct link between the local analytic type of singularities and the global Brill–Noether count.

6. Higher Rank Brill–Noether Numbers and Clifford Phenomena

In higher rank, the Brill–Noether number generalizes to

$\beta(n,d,k) = n^2(g-1)+1 - k(k-d + n(g-1))$

parametrizing vector bundles $E$ of rank $n$, degree $d$, with $h^0(E)\ge k$ (Lange et al., 2017). In rank $n=1$, one recovers the classical Brill–Noether number.

Latest genus-6 theory identifies phenomena with some $B(n,d,k)$ loci non-empty despite negative $\beta(n,d,k)$. Clifford dimension >1 curves (e.g., plane quintics) and canonical loci further necessitate refined Brill–Noether counts.

A plausible implication is that higher rank Brill–Noether theory diverges from its rank-one counterpart, with negative Brill–Noether numbers indicating new features in moduli behavior—components "below the Brill–Noether line"—and connections to higher Clifford indices and Clifford-type inequalities.

7. Formulas, Examples, and Applications

Several general and explicit formulas underpin Brill–Noether theory:

Setting	BN number formula	Dim. criterion
Smooth curves	$\rho(g,r,d)=g-(r+1)(g-d+r)$	$\rho\ge0\implies$ nonempty
Graphs	$\rho(g,r,d)=g-(r+1)(g-d+r)$	Same as curves
Surfaces	$\rho(C,r)=q-(r+1)(p_a(C)-C^2+r)$	$\rho(C,r)\ge0\implies$ nonempty
Higher rank	$\beta(n,d,k)=n^2(g-1)+1-k(k-d+n(g-1))$	$\beta\ge0$ usually

Notable applications include lower bounds for $h^0(K_D)$ in linear systems, inequalities for non-moving curves, and a rich supply of examples where the computed Brill–Noether number matches or contrasts with actual family dimensions (e.g., symmetric products, Fano surfaces, cuspidal curves).

The extension to singularities and higher dimensions illustrates the adaptability and centrality of the Brill–Noether number across modern algebraic geometry and combinatorics.

---

For foundational, combinatorial, and advanced moduli-theoretic treatments see (Caporaso, 2011, Pflueger, 2013, Lopes et al., 2011, Cotterill et al., 2023, Lange et al., 2017).