Griffiths' Conjecture in Complex Geometry

• Griffiths' Conjecture is a collection of hypotheses in complex geometry linking the ampleness of holomorphic vector bundles with the existence of Hermitian metrics having Griffiths positive curvature.
• The conjecture extends to the positivity of Chern–Weil forms and the quasi-projectivity of period map images, connecting analytic methods with algebraic geometry.
• Recent progress involves analytic approaches such as Hermitian–Yang–Mills/Monge–Ampère systems and localized L2 extension estimates, illuminating both resolved and open aspects.

In the literature represented here, the expression “Griffiths’ Conjecture” does not denote a single universally fixed statement. It is used for several distinct but related problems in complex geometry, Hodge theory, and Nevanlinna theory. The most common usage in the present corpus is the vector-bundle conjecture asserting that a holomorphic vector bundle is ample if and only if it admits a smooth Hermitian metric whose Chern curvature is Griffiths positive; other uses concern positivity of Chern–Weil forms, quasi-projectivity of period-map images, and holomorphic curves in logarithmic settings (Demailly, 2020, Bakker et al., 2018, Dong et al., 2022).

1. Terminological scope

A concise way to organize the main usages is the following.

Usage of the name	Representative statement	Status in the papers here
Vector-bundle positivity	Ample $E$ iff $E$ admits a Griffiths positive Hermitian metric	Open in general; proved on compact Riemann surfaces (Demailly, 2020, Murakami, 27 Sep 2025)
Positivity of Chern–Weil forms	Positive Schur polynomials in Chern forms of a Griffiths semipositive bundle should be positive forms	Partial results and new positive cases (Diverio et al., 2020, Fagioli, 2020)
Period maps	The image of a period map should be quasi-projective	Proved (Bakker et al., 2018)
Nevanlinna/algebraic-geometry usage	A Griffiths conjecture concerning holomorphic curves and divisors	Abstract claims a proof, but the supplied record lacks theorem statements (Dong et al., 2022)

This multiplicity of meanings is not merely terminological. It reflects several distinct research programs associated with Phillip Griffiths and with Griffiths-inspired methods. A recurring source of confusion is the conflation of these conjectures with the Green–Griffiths conjecture, the Griffiths–Harris conjectures, or statements about Griffiths numbers; these are separate topics.

2. The ampleness-versus-curvature conjecture for vector bundles

In the formulation used by Demailly, Griffiths’ conjecture states that a holomorphic vector bundle $E$ on a projective complex manifold is ample if and only if it admits a smooth Hermitian metric $h$ such that its Chern curvature tensor $\Theta_{E,h}$ is positive in the sense of Griffiths (Demailly, 2020). With local curvature coefficients $c_{jk\lambda\mu}$, Griffiths positivity means that

$$\widetilde{\Theta}_{E,h}(\xi\otimes v) = \sum_{j,k,\lambda,\mu} c_{jk\lambda\mu}\,\xi_j\bar\xi_k\,v_\lambda\bar v_\mu >0$$

for all nonzero $\xi\otimes v\in T_X\otimes E$ (Demailly, 2020). The implication

$$\Theta_{E,h}\ \text{Griffiths positive} \Longrightarrow E\ \text{ample}$$

is classical; the difficult direction is the converse.

Demailly reformulated the conjecture analytically through a coupled Hermitian–Yang–Mills/Monge–Ampère type system. The proposal is not itself a proof; rather, it reduces the conjecture to a global existence problem for a specifically designed elliptic system whose solutions would yield metrics that are even dual Nakano positive, hence Griffiths positive (Demailly, 2020). This analytic program was refined in several directions. Naumann gave a sufficient condition under which a positive Hermitian metric on $\mathcal O_{\mathbb P(E^*)}(1)$ induces a Griffiths positive $E$0-metric on $E$1, and suggested studying a relative Kähler–Ricci flow on $E$2 for the fibration $E$3 (Naumann, 2017). Pingali proved that a “cushioned” Hermitian–Einstein-type equation proposed in Demailly’s approach has, on a stable bundle, only conformal rescalings of the Hermitian–Einstein metric as solutions, indicating that this particular PDE must be modified if it is to produce genuinely new Griffiths-positive metrics (Pingali, 2021).

Within this line of work, the one-dimensional case is now settled analytically. Murakami proved that for a holomorphic vector bundle $E$4 on a compact Riemann surface, the following are equivalent: the Demailly system is solvable continuously in $E$5, $E$6 admits a Hermitian metric with Griffiths positive curvature, and $E$7 is ample (Murakami, 27 Sep 2025). This confirms Demailly’s program on compact Riemann surfaces. The same paper identifies the decisive analytic issue as an a priori lower bound for the scalar unknown $E$8 in the Demailly system, obtained via Uhlenbeck–Yau-type compactness and a contradiction with positivity of degrees of quotient bundles on curves (Murakami, 27 Sep 2025).

A complementary analytic perspective replaces curvature inequalities by extension inequalities. A smooth Hermitian holomorphic vector bundle is Griffiths positive at a point if and only if it satisfies a quantitative local $E$9-extension estimate with the sharp second-order defect

$E$0

where $E$1 is the Griffiths lower curvature bound and $E$2 is a holomorphic cylinder (Liu et al., 2022). This does not solve the ampleness problem, but it gives a precise analytic characterization of Griffiths positivity itself.

3. The conjecture on positivity of Chern–Weil and Schur forms

A second major usage of the name concerns positive characteristic forms. In this formulation, one asks: if $E$3 is Griffiths (semi)positive, do all positive polynomials in the Chern forms of $E$4 define positive $E$5-forms? Here “positive polynomials” means positive linear combinations of Schur polynomials (Diverio et al., 2020). This question is presented as a pointwise Hermitianized analogue of the Fulton–Lazarsfeld theorem on numerically positive polynomials for ample vector bundles (Diverio et al., 2020).

Recent work has produced substantial evidence without proving the full conjecture. A pointwise Hermitian universal Gysin formula on flag bundles shows that for a Griffiths semipositive bundle, many push-forwards of powers of tautological Chern forms are closed strongly positive forms, and the corresponding universal polynomials lie in the Schur-positive cone $E$6 (Diverio et al., 2020). The same work emphasizes that it does not prove the full conjecture; rather, it constructs a large family of new positive examples (Diverio et al., 2020).

A complementary rank-$E$7 result proves positivity of

$E$8

for Griffiths semipositive Hermitian holomorphic vector bundles of rank $E$9, and also proves positivity of $h$0 in arbitrary rank (Fagioli, 2020). In particular, in rank $h$1 one obtains the chain of inequalities

$h$2

as pointwise $h$3-forms (Fagioli, 2020). These results concern the positivity of Schur forms attached to a given Griffiths semipositive metric; they should not be conflated with the ampleness-versus-curvature existence conjecture, even though both belong to Griffiths positivity theory.

4. The period-map conjecture

Another well-established use of the name is the conjecture that the image of a period map is quasi-projective. In the formulation proved by Bakker, Brunebarbe, and Tsimerman, if $h$4 is a reduced separated algebraic space of finite type over $h$5 and

$h$6

is a period map, then $h$7 factors as

$h$8

for a dominant morphism $h$9 of algebraic spaces and a closed immersion $\Theta_{E,h}$0, and the Griffiths $\Theta_{E,h}$1-bundle restricted to $\Theta_{E,h}$2 is ample; in particular $\Theta_{E,h}$3 is quasi-projective (Bakker et al., 2018).

The proof combines three ingredients: definability of period maps in the o-minimal structure $\Theta_{E,h}$4, a GAGA theorem for definable coherent sheaves on complex algebraic spaces, and Artin-style algebraization of proper definable images (Bakker et al., 2018). In this context, “Griffiths’ conjecture” is thus a statement about the algebraicity and positivity of period-map images, not about vector-bundle curvature.

This usage is historically close to Griffiths’ work on period domains and Hodge bundles. It is also one of the few versions in the present corpus that is fully resolved in arbitrary dimension.

5. The Nevanlinna-theoretic usage in “On Griffiths conjecture”

The arXiv record for “On Griffiths conjecture” states only that, by using techniques of holomorphic jets and Jacobian fields, the paper devises a non-equidistribution theory of holomorphic curves into complex projective varieties intersecting normal crossing divisors, and on that basis proves the Griffiths conjecture and the Green–Griffiths conjecture in Nevanlinna theory and algebraic geometry (Dong et al., 2022). The same record also states that the supplied material contains no theorem statements, notation, or proofs beyond the abstract, so the precise formulation of the “Griffiths conjecture” meant there cannot be reconstructed from the available text (Dong et al., 2022).

What can be said safely is that this title places the conjecture in the entire-curve and logarithmic-geometry tradition that also includes work on jet differentials, algebraic differential equations, and algebraic degeneracy of holomorphic curves. Merker proved the “first half” of the Green–Griffiths conjectural picture for smooth projective hypersurfaces of general type by constructing nonzero global Green–Griffiths jet differentials annihilating entire curves at the optimal threshold $\Theta_{E,h}$5 (Merker, 2010). Bérczi later established an effective algebraic degeneracy theorem for generic projective hypersurfaces of degree at least $\Theta_{E,h}$6 via equivariant localization on the Demailly–Semple jet tower (Berczi, 2015). For complements, Ascher, Turchet, and Yeong proved that the complement of a very general pair of hypersurfaces of total degree $\Theta_{E,h}$7 in $\Theta_{E,h}$8 is algebraically hyperbolic modulo a proper closed subset, thereby providing evidence toward Green–Griffiths–Lang phenomena in the logarithmic setting (Ascher et al., 2024).

A plausible implication is that the conjecture referred to in (Dong et al., 2022) belongs to this jet-theoretic and Nevanlinna-theoretic cluster. However, the precise hypotheses and statement are unavailable in the supplied record and should not be inferred more specifically than the abstract warrants.

6. Related but distinct conjectures

Several nearby statements are often confused with “Griffiths’ Conjecture,” but the present literature distinguishes them sharply.

The Griffiths–Harris conjecture in Noether–Lefschetz theory predicts the minimal codimension of the locus of smooth degree-$\Theta_{E,h}$9 surfaces in $c_{jk\lambda\mu}$0 with Picard number at least $c_{jk\lambda\mu}$1; an asymptotic form for fixed $c_{jk\lambda\mu}$2 and $c_{jk\lambda\mu}$3 was proved by considering surfaces containing $c_{jk\lambda\mu}$4 coplanar lines (Dan, 2014). A different Griffiths–Harris conjecture concerns the Abel–Jacobi map for codimension-two cycles on general hypersurfaces $c_{jk\lambda\mu}$5 of degree $c_{jk\lambda\mu}$6; Yang proved its infinitesimal form by identifying the differential of the Abel–Jacobi map with

$c_{jk\lambda\mu}$7

and showing that its image vanishes for general such hypersurfaces (Yang, 2018). Ravindra treated yet another Griffiths–Harris problem on curves on threefold hypersurfaces, proving that any arithmetically Gorenstein curve on a general smooth hypersurface $c_{jk\lambda\mu}$8 of degree at least $c_{jk\lambda\mu}$9 is a complete intersection (Ravindra, 2010).

The Green–Griffiths conjecture is likewise distinct. In one standard form, it predicts that a projective variety of general type should contain a proper algebraic subset containing the image of every nonconstant entire curve; this is the setting of jet-differential work such as (Merker, 2010) and (Berczi, 2015). In a different Hodge-theoretic form, the expression “Green–Griffiths conjecture” denotes the algebraicity of the zero locus of an admissible normal function, a theorem surveyed by Charles (Charles, 2013). These are not the same as Griffiths’ conjectures on vector bundles or period maps.

Finally, Griffiths numbers in singularity theory define another unrelated usage. Yau conjectured inequalities for the $$\widetilde{\Theta}_{E,h}(\xi\otimes v) = \sum_{j,k,\lambda,\mu} c_{jk\lambda\mu}\,\xi_j\bar\xi_k\,v_\lambda\bar v_\mu >0$$0-st Griffiths number and Hironaka number of isolated singularities, and Lu and Yau showed that these conjectures fail for isolated rigid Gorenstein singularities in dimension $$\widetilde{\Theta}_{E,h}(\xi\otimes v) = \sum_{j,k,\lambda,\mu} c_{jk\lambda\mu}\,\xi_j\bar\xi_k\,v_\lambda\bar v_\mu >0$$1, while one inequality survives for irregular singularities (Du et al., 2014).

The general encyclopedic lesson is therefore one of nomenclature as much as of mathematics: “Griffiths’ Conjecture” names several technically distinct conjectures whose common feature is their origin in Griffiths’ geometric program. In current usage, the ampleness-versus-Griffiths-positivity problem remains the default meaning in vector-bundle theory, the period-map version is solved, the Chern–Weil-form version remains partially open, and the Nevanlinna-theoretic usage of (Dong et al., 2022) can presently be described only at the level of its abstract.