Papers
Topics
Authors
Recent
Search
2000 character limit reached

Characteristic-Zero Lifts of Partial Hasse Invariants

Updated 30 January 2026
  • Characteristic-zero lifts of partial Hasse invariants extend classical modular invariants from characteristic p to zero, unifying crystalline, modular, and differential approaches.
  • They are constructed via crystalline Dieudonné theory and explicit computational techniques, which yield refined arithmetic stratifications in Shimura varieties and Hilbert modular settings.
  • These lifts detect μ-ordinary and non-ordinary loci, providing practical tools for analyzing Newton strata and guiding the reduction of twisted modular forms.

Characteristic-zero lifts of partial Hasse invariants refer to the canonical extension of modular invariants, originally defined in characteristic pp, to characteristic zero. These lifts are crucial in relating the geometry and arithmetic of Shimura varieties, pp-divisible groups, and Hilbert modular varieties across various characteristics. They generalize the classical Hasse invariant and provide a fine arithmetic stratification, particularly detecting the μ\mu-ordinary and non-ordinary loci. Central to modern advances, their construction unifies crystalline, modular, and differential approaches, with explicit computational techniques and enumerative formulas.

1. Definition and Local Construction of Partial Hasse Invariants

Let pp be a prime and F/QpF/\mathbb{Q}_p a finite unramified extension with ring of integers O=OF\mathcal{O} = \mathcal{O}_F and residue field κ=O/p\kappa = \mathcal{O}/p. For a pp-divisible group GG over a κ\kappa-scheme SS equipped with an O\mathcal{O}-action, the conormal sheaf of the Cartier dual GDG^D admits a decomposition

ωGD=τHom(κ,κ)ωGD,τ\omega_{G^D} = \bigoplus_{\tau \in \operatorname{Hom}(\kappa, \kappa)}\omega_{G^D, \tau}

with each summand locally free of rank qτq_\tau. The signature (pτ,qτ)(p_\tau,q_\tau), determined by ranks of the respective conormal and cotangent sheaves, specifies the nature of the group.

Partial Hasse invariants are constructed using the crystalline Dieudonné theory. Specifically, for each τ0\tau_0 with qτ0=qq_{\tau_0} = q,

Ha~τ0:qωGD,τ0qωGD,τ0(pf)det(ωGD,τ0)pf\widetilde{\mathrm{Ha}}_{\tau_0}: \wedge^{q}\omega_{G^D,\tau_0} \to \wedge^q\omega_{G^D,{\tau_0}}^{(p^f)} \simeq \det(\omega_{G^D,\tau_0})^{\otimes p^f}

divides the wedge of Frobenius by an explicit pp-power, and twists yield

Ha~τ0(G)H0(S,det(ωGD,τ0)(pf1)).\widetilde{\mathrm{Ha}}_{\tau_0}(G) \in H^0\Big(S, \det(\omega_{G^D,\tau_0})^{\otimes(p^f-1)}\Big).

These invariants detect the locus where the Newton and Hodge polygons meet at qτq_\tau; their vanishing defines divisors that stratify the moduli space according to slopes and pp-rank strata (Hernandez, 2016).

2. Properties: Functoriality, Cartier Divisors, and Product Structure

Partial Hasse invariants possess extensive functoriality and compatibility properties:

  • Base Change: Their formation commutes with any morphism of smooth bases.
  • Duality: There is a canonical identification:

det(ωG,τ)(pf1)det(ωGD,τ)(pf1)\det(\omega_{G,\tau})^{\otimes(p^f-1)} \simeq \det(\omega_{G^D,\tau})^{\otimes(p^f-1)}

with the invariants coinciding under dualization.

  • Products: For G=G1×G2G = G_1 \times G_2 and kτ(G)=kτ(G1)+kτ(G2)k_\tau(G) = k_\tau(G_1) + k_\tau(G_2),

Ha~τ(G1×G2)=Ha~τ(G1)Ha~τ(G2).\widetilde{\mathrm{Ha}}_\tau(G_1 \times G_2) = \widetilde{\mathrm{Ha}}_\tau(G_1) \otimes \widetilde{\mathrm{Ha}}_\tau(G_2).

  • Cartier Divisor: The vanishing locus of each Ha~τ\widetilde{\mathrm{Ha}}_\tau defines a reduced, effective Cartier divisor on the moduli stack, with the complementary μ\mu-ordinary locus Zariski dense by Wedhorn's purity.

The global μ\mu-ordinary Hasse invariant is defined as

μHa~(G)=τHa~τ(G)\widetilde{^\mu\mathrm{Ha}}(G) = \bigotimes_{\tau} \widetilde{\mathrm{Ha}}_\tau(G)

and cuts out the non-μ\mu-ordinary locus in codimension 1, guaranteeing the divisors are reduced (Hernandez, 2016).

3. Characteristic-Zero Lifting: Crystalline and Modular Approaches

Given a smooth O\mathcal{O}-scheme S~\widetilde{S} with special fiber SS, and a pp-divisible O\mathcal{O}-group G/SG/S lifting uniquely to G~/S~\widetilde{G}/\widetilde{S}, the corresponding Dieudonné crystal extends and carries the same partial Hasse structures. By uniqueness, each Ha~τ(G)\widetilde{\mathrm{Ha}}_\tau(G) lifts canonically to a global section over S~\widetilde{S}: H0 ⁣(S~,det(ωG~D,τ)(pf1)).H^0\!\big(\widetilde{S},\,\det(\omega_{\widetilde{G}^D,\tau})^{\otimes(p^f-1)}\big). This provides a characteristic-zero lift, unifying and generalizing classical Eisenstein series-type constructions in modular and Siegel settings (Hernandez, 2016).

In the context of Hilbert modular varieties, Bogo–Li establish a construction of characteristic-zero lifts for genus-zero, non-compact curves YMFY \hookrightarrow \mathcal{M}_F via solutions Fj(t)F_j(t) to Picard–Fuchs differential equations, with the reduction mod pp congruent to the polynomial defining the non-ordinary locus. They define explicit holomorphic twisted modular forms hp,j(τ)h_{p,j}(\tau) with pp-integral tt-expansions that reduce to the partial Hasse polynomials php,j(t)\mathrm{ph}_{p,j}(t), providing uniqueness and modularity in the lift (Bogo et al., 23 Jan 2026).

4. Explicit Computations and Enumerative Formulas

On genus-zero curves in Hilbert modular varieties, pulling back the partial Hasse invariants yields monic polynomials in a Hauptmodul tt. The construction via Picard-Fuchs theory shows that for each partial Hasse polynomial php,j(t)\mathrm{ph}_{p,j}(t) corresponding to divisor DjD_j,

degphp,j(t)=dimCMwj(Γ,φ)1,\deg \mathrm{ph}_{p,j}(t) = \dim_{\mathbb{C}} M_{w_j}(\Gamma, \varphi) - 1,

with explicit weights wj=(p1,0)w_j = (p-1,0) or (0,p1)(0,p-1) (split) and wj=(1,p)w_j = (-1,p) or (p,1)(p,-1) (inert). Consequently, the number of non-ordinary points is given by

#{Yt0 non-ordinary}=1p1(dimMk(Γ,φ)correction)\#\{ Y_{t_0} \text{ non-ordinary}\} = \frac{1}{p-1}\left(\dim M_k(\Gamma,\varphi) - \text{correction}\right)

and exhibits linear growth in pp modulo corrections, recapitulating and generalizing Deuring’s formula in the modular curve context (Bogo et al., 23 Jan 2026).

For families such as the Teichmüller curve W13W_{13}, precise formulas for the degrees and counts of supersingular fibers emerge, dependent on the congruence class of pp and the topology of W13W_{13}.

5. Reduction of Modular and Twisted Modular Forms Modulo pp

The reductions of twisted modular forms, particularly those with pp-integral tt-expansions, are governed by rational functions whose denominators are powers of the partial Hasse polynomials and additional factors from elliptic fixed points. Explicitly, for fj(τ)M(k,kp)(Γ,φ)f_j(\tau) \in M_{(-k, kp)}(\Gamma, \varphi) with sufficiently large pp, one obtains

fj(t)Pf(t)php,j(t)ki(t(ei)t)αj,i(modp)f_j(t) \equiv \frac{P_f(t)}{\mathrm{ph}_{p,j}(t)^k\prod_i (t(e_i)-t)^{\alpha_{j,i}}} \pmod{p}

where Pf(t)P_f(t) lies in the ring of integers of the coefficient field. Conversely, every such rational function arises from some twisted modular form, providing a description of the mod pp reduction behavior (Bogo et al., 23 Jan 2026).

6. Hypergeometric and Differential-Equation-Based Constructions

Several family-specific constructions utilize hypergeometric series and explicit determinants, notably for the triangle curve Δ(2,5,)\Delta(2,5,\infty). The Jacobian family

Cη:y2=x55x3+5x2ηC_\eta: y^2 = x^5 - 5x^3 + 5x - 2\eta

yields modular forms Q(τ),R(τ)Q(\tau), R(\tau) and twisted modular forms B(τ)B(\tau), with recursion

(15Qx4+5Q2x82Rx10)1/2=n0dn(Q,R)xn,(1-5Qx^4 + 5Q^2x^8 - 2Rx^{10})^{-1/2} = \sum_{n\ge 0} d_n(Q,R)x^n,

providing explicit characteristic-zero lifts for both inert and split cases: \begin{align*} h_{p,1}(\tau) &= Bp(\tau) d_{p-3}(Q(\tau),R(\tau)), \ h_{p,2}(\tau) &= B{-1}(\tau) d_{3p-1}(Q(\tau), R(\tau)), & p \text{ inert}, \ h_{p,1}(\tau) &= d_{3p-3}(Q(\tau), R(\tau)), \ h_{p,2}(\tau) &= B{p-1}(\tau) d_{p-1}(Q(\tau), R(\tau)), & p \text{ split}, \end{align*} which are rewritten in terms of classical hypergeometric functions. In each good prime case, squares of these forms reduce mod pp to constants, ensuring their modular interpretations and lifting properties (Bogo et al., 23 Jan 2026).

7. Applications: μ\mu-Ordinary Locus and Shimura Varieties

In integral models of Shimura varieties of PEL type, especially those reducing to GLh(OF)\mathrm{GL}_h(\mathcal{O}_F) at pp, the partial Hasse invariants detect Newton strata, with the μ\mu-ordinary stratum given by the non-vanishing locus. The construction recovers established invariants on the special fiber and aligns with explicit calculations via Serre–Tate theory and display formulas. In these moduli spaces, the explicit nature and functoriality of characteristic zero lifts afford concrete tools for analyzing geometric and arithmetic structures in mixed characteristic settings (Hernandez, 2016).


For further details and explicit constructions of characteristic-zero lifts of partial Hasse invariants, see (Hernandez, 2016) and (Bogo et al., 23 Jan 2026).

Definition Search Book Streamline Icon: https://streamlinehq.com
References (2)

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to Characteristic-Zero Lifts of Partial Hasse Invariants.