---
title: Twisted Holomorphic 1-Forms
url: https://www.emergentmind.com/topics/twisted-holomorphic-1-forms
type: topic
---

# Twisted Holomorphic 1-Forms

Twisted holomorphic \(1\)-forms are holomorphic or meromorphic \(1\)-forms defined relative to a flat line bundle, or equivalently relative to a twisted differential rather than the ordinary exterior differential. In the literature summarized here, the term appears in three closely related settings: on the punctured projective line as \(d_\alpha\)-closed meromorphic \(1\)-forms; on compact complex surfaces as sections of \(\Omega_X^1(\log D)\otimes L\), where \(L\) is a flat holomorphic line bundle; and on marked Riemann surfaces as meromorphic sections of \(K_X\otimes L_\chi\), equivalently dilation surfaces with scaling. These formulations connect twisted de Rham cohomology, logarithmic residues, resonance phenomena, Lee classes of locally conformally symplectic structures, and moduli spaces carrying \(GL(2,\mathbb R)\)- and \(SL(2,\mathbb R)\)-actions [1812.09791], [2204.02122], [2507.10685].

## 1. Foundational definitions

On the punctured sphere, one starts with
\[
U=\mathbb P^1\setminus\{z_1,\dots,z_n,\infty\},
\qquad
\omega=\sum_{j=1}^n \alpha_j\,\frac{dt}{t-z_j}\in \Omega^1(U),
\]
and defines
\[
d_\alpha:\Omega^p(U)\to \Omega^{p+1}(U),
\qquad
d_\alpha(\phi)=d\phi+\omega\wedge\phi.
\]
The resulting twisted de Rham complex is
\[
0\to \Omega^0(U)\xrightarrow{d_\alpha}\Omega^1(U)\to 0.
\]
In this model, a twisted holomorphic \(1\)-form is a global meromorphic \(1\)-form \(\eta\) on \(\mathbb P^1\) with poles only at the \(z_j\) and \(\infty\) such that \(d_\alpha\eta=0\) [1812.09791].

On a compact complex manifold \(X=(M,J)\), Apostolov–Dloussky fix a closed real or complex \(1\)-form \(\alpha\) representing a de Rham class \(a=[\alpha]\in H^1_{\mathrm{dR}}(M)\). Via the exponential exact sequence this determines a topologically trivial flat real line bundle
\[
L_a\longleftrightarrow (M\times \mathbb R,\nabla=d+\alpha),
\]
whose complexification \(L=L_a\otimes \mathbb C\) is a flat holomorphic line bundle. The twisted differential is then
\[
d_\alpha=d-\alpha\wedge\cdot,
\]
and it splits as
\[
d_\alpha=\partial_\alpha+\overline\partial_{-\alpha},
\qquad
\partial_\alpha=\partial-\alpha^{1,0}\wedge,
\qquad
\overline\partial_{-\alpha}=\overline\partial+\alpha^{0,1}\wedge.
\]
Accordingly, forms with values in \(L\) are identified with ordinary forms equipped with the twisted differential [2204.02122].

On a marked Riemann surface \(X\cong \Sigma_{g,n}\), Apisa–Salter fix a character
\[
\chi\in \operatorname{Hom}(\pi_1(\Sigma_{g,n}),\mathbb R_{>0})\approx H^1(\Sigma_{g,n};\mathbb R_{>0}),
\]
let \(L_\chi\to \Sigma_{g,n}\) be the associated flat complex line bundle, and define a twisted holomorphic \(1\)-form to be a meromorphic section
\[
\omega\in H^0(X,K_X\otimes L_\chi)
\]
with poles and zeros only at the marked points. Equivalently, on the universal cover \(\widetilde X\),
\[
\gamma^*\omega=\chi(\gamma)^{-1}\omega
\qquad
\text{for every } \gamma\in \pi_1(\Sigma_{g,n}).
\]
If \(\chi\equiv 1\), this reduces to an ordinary holomorphic \(1\)-form [2507.10685].

These definitions are compatible in spirit but not identical in notation. In particular, the sign convention in the punctured-sphere model uses \(d+\omega\wedge\cdot\), whereas the compact-surface formulation uses \(d-\alpha\wedge\cdot\).

## 2. Local structure on the punctured projective line

For the twisted de Rham complex on \(U\), the local behavior of a twisted holomorphic \(1\)-form is controlled by the residues of the twisting form. Near \(t=z_j\), one has an expansion
\[
\eta=\sum_{k=-N}^\infty c_k\,(t-z_j)^{k-1}\,dt.
\]
Substituting into \(d\eta+\omega\wedge\eta=0\) gives, for each \(k\), the recurrence
\[
(k+\alpha_j)\,c_k+(\text{lower-order terms in }c_{<k})=0.
\]
In particular, the exponent of the pole is shifted by \(\alpha_j\) [1812.09791].

A convenient \(\mathbb C\)-basis of \(\Omega^0(U)\) is
\[
\{t^a\mid a\ge 0\}\cup \{(t-z_i)^{-a}\mid a>0,\ i=1,\dots,n\},
\]
and a convenient basis of \(\Omega^1(U)\) is
\[
\{t^a\,dt\mid a\ge 0\}\cup \left\{\frac{dt}{(t-z_i)^a}\mid a>0\right\}.
\]
Writing \(\kappa=1\) and \(\alpha_j=-m_j/\kappa\) as in the paper’s conventions, the action of \(d_\alpha\) on these basis elements is explicit:
\[
\kappa\,d_\alpha\bigl((t-z_i)^{-a}\bigr)
=
-(m_i+a\kappa)\frac{dt}{(t-z_i)^{a+1}}
+\sum_{k=1}^a\sum_{j\ne i} m_j\frac{dt}{(z_j-z_i)^k}(t-z_i)^{-a-1+k}
-\sum_{j\ne i} m_j\frac{dt}{(z_j-z_i)^a}\frac1{t-z_j},
\]
and
\[
\kappa\,d_\alpha(t^a)
=
\left(a\kappa-\sum_{j=1}^n m_j\right)t^{a-1}dt
-\sum_{k=1}^{a-1}\sum_{j=1}^n m_j z_j^k\,t^{a-1-k}dt
-\sum_{j=1}^n m_j z_j^a\,\frac{dt}{t-z_j}.
\]

These formulas make the twisting effect concrete: the ordinary derivative is replaced by a differential that couples the local pole structure at one puncture to all other punctures. This is the mechanism behind the later appearance of resonance and extra cohomological relations.

## 3. Logarithmic subcomplex, resonance, and representation-theoretic reflection

Inside \(\Omega^\bullet(U)\), the logarithmic subcomplex is
\[
\Omega^0_{\log}=\mathbb C\cdot 1,
\qquad
\Omega^1_{\log}=\operatorname{Span}\left\{\frac{dt}{t-z_1},\dots,\frac{dt}{t-z_n}\right\},
\]
and its differential sends \(1\mapsto \omega\). If no resonance occurs, the inclusion of the logarithmic subcomplex is a quasi-isomorphism, and
\[
H^0(\Omega^\bullet,d_\alpha)=0,
\qquad
\dim H^1(\Omega^\bullet,d_\alpha)=n-1.
\]
A convenient basis of \(H^1\) is given by the logarithmic forms
\[
\omega_j=\frac{dt}{t-z_j},\qquad j=1,\dots,n,
\]
subject to the single cohomological relation
\[
\sum_{j=1}^n m_j\,\omega_j=0.
\]
Hence \(\dim H^1=n-1\) in the generic case [1812.09791].

The resonance parameters are those for which new relations appear among the \(\omega_j\). Setting
\[
m_{n+1}=\sum_{j=1}^n m_j-2,
\]
the three types of resonance are:
\[
\text{(a)}\quad m_i+a\kappa=0 \text{ for some } i\in\{1,\dots,n\},\ a\in \mathbb Z_{>0},
\]
\[
\text{(b)}\quad m_{n+1}+2-a\kappa=0 \text{ for some } a\in \mathbb Z_{>0},
\]
\[
\text{(c)}\quad \kappa=0.
\]
Under type (a) or (b), one finds an extra linear relation among the \(\omega_j\) of degree \(a\). The summary records, for example, that if \(m_{n+1}+2-\kappa=0\), then
\[
\sum z_j\,m_j\,\omega_j\equiv 0 \quad \text{in } H^1,
\]
and if \(m_{n+1}+2-2\kappa=0\), then
\[
\sum z_j^2\,m_j\,\omega_j-\kappa^{-1}\Bigl(\sum z_j\,m_j\Bigr)\Bigl(\sum z_j\,m_j\,\omega_j\Bigr)\equiv 0.
\]
At each resonance the dimension of \(H^1\) drops by one more; if no two resonance conditions coincide, then
\[
\dim H^1(\Omega^\bullet,d_\alpha)=n-1-(\#\text{ of independent resonances among (a),(b),(c)}).
\]

The same paper considers a second complex: the chain complex of the Lie algebra of \(\mathfrak{sl}_2\)-valued algebraic functions on the same complement, with coefficients in a tensor product of contragradient Verma modules over \(\hat{\mathfrak{sl}}_2\). Following a construction suggested by Schechtman and Varchenko, a monomorphism from the twisted de Rham complex into this chain complex is established, and under this monomorphism the existence of singular vectors in the Verma modules, specifically the Malikov–Feigin–Fuchs singular vectors, is reflected in relations between cohomology classes of the de Rham complex [1812.09791]. This suggests that twisted holomorphic \(1\)-forms on the punctured sphere encode both analytic and representation-theoretic resonance.

## 4. Flat bundles, logarithmic poles, and residues on complex surfaces

On a complex surface \(X\) with an effective normal-crossing divisor \(D\), the relevant sheaf is
\[
\Omega_X^1(\log D)\subset \Omega_X^1(D),
\]
whose sections are \(1\)-forms on \(X\setminus D\) with at most simple poles on \(D\) and whose exterior derivative also has at most simple poles there. Equivalently, a local section of \(\Omega_X^1(\log D)\otimes L\) near a point of \(D\) may be written
\[
\theta=\theta_0+\sum_{j=1}^k g_j\,\frac{df_j}{f_j}\otimes s
\quad\in\quad \Omega^1(U\setminus D)\otimes L,
\]
where \(f_j\) are local defining equations of the irreducible components of \(D\), \(g_j\) are holomorphic, \(\theta_0\) is holomorphic, and \(s\) is a local flat frame of \(L\). Such a \(\theta\) satisfies
\[
\overline\partial_{-\alpha}\theta=0
\quad\text{on }X\setminus D
\]
[2204.02122].

By standard residue theory, each component \(D_j\subset D\) carries a constant residue \(\operatorname{Res}_{D_j}(\theta)\in \mathbb C\), and in the sense of currents one has
\[
\partial_\alpha[\theta\wedge\cdot]
=
2\pi i\sum_j \operatorname{Res}_{D_j}(\theta)\,\delta_{D_j},
\]
where \(\delta_{D_j}\) is the current of integration along \(D_j\). A twisted logarithmic \(1\)-form is said to be of positive type if each residue is real and strictly positive. The same framework yields twisted Dolbeault cohomology groups
\[
H^q\bigl(X,\Omega^p\otimes L\bigr),
\]
and the paper emphasizes that all twisted cohomologies enjoy Serre duality via currents.

In this setting, twisted holomorphic \(1\)-forms are not restricted to the pole-free case. The theory explicitly includes logarithmic poles along \(D\), residues as numerical invariants, and current-theoretic identities that translate geometric positivity into cohomological information.

## 5. Class VII surfaces, Lee classes, and explicit geometric consequences

Apostolov–Dloussky study minimal non-Kähler surfaces \(S\) in Kodaira’s class VII, where \(b_1(S)=1\) and the Kodaira dimension is \(-\infty\). One of the main results is a classification of when such a surface admits a non-zero twisted logarithmic \(1\)-form
\[
0\ne \theta\in H^0\bigl(S,\Omega_S^1(\log D)\otimes L_a\bigr),
\qquad
a=[\alpha]\in H^1_{\mathrm{dR}}(S,\mathbb R).
\]
If \(D=\varnothing\), Lemma 6.11 shows that \(S\) must be a Hopf surface, an Inoue–Bombieri surface, or an Enoki surface. If \(D\ne\varnothing\), Proposition 6.10 and the lemmas of Section 6.2.2 imply that each connected component of \(D\) must contain a cycle of rational curves and that there are at most two connected components of \(D\). If \(D\) has exactly two connected components one recovers the Inoue–Hirzebruch surfaces, while if \(D\) is a single connected cycle one gets an intermediate Kato surface. In all cases, the only possibilities for the twisting parameter \(a\) are finitely many real values, and exactly one of them satisfies \(a<0\) [2204.02122].

The same paper ties these forms to the set \(T(S)\) of Lee classes of locally conformally symplectic forms taming the complex structure:
\[
T(S)=\left\{\lambda\in H^1_{\mathrm{dR}}(S,\mathbb R)\,\middle|\,
\exists\text{ LCS form }\omega\text{ taming }J,\ d\omega=\alpha\wedge\omega
\right\}.
\]
Under the hypotheses of Theorem 1.6, including that the foliation defined by \(\theta\) has only real negative characteristic numbers at any nondegenerate singularity on \(D\), the class \(a\) bounds the Lee classes of taming LCS forms and exactly one of the following holds: \(a=0\) and \(T(S)=(-\infty,0)\), in which case \(S\) is a Hopf or Enoki surface; \(a<0\) and \(T(S)\subset(-\infty,a)\), in which case \(S\) is a hyperbolic Kato surface of intermediate type; \(a<0\) and \(T(S)=\{a\}\), in which case \(S\) is an Inoue–Bombieri surface; or \(a<0\) and \(T(S)\subset(a,0)\).

Upper and lower bounds on \(T(S)\subset(-\infty,0)\) are characterized by automorphic plurisubharmonic functions on the minimal \(\mathbb Z\)-cover \(\widetilde S\to S\). Specifically, \(T(S)\subset(-\infty,b)\) for some \(b<0\) if and only if \(\widetilde S\) admits a negative, strictly PSH function \(u<0\) satisfying
\[
u\circ \gamma=C_b\,u,
\qquad
C_b>1,
\]
equivalently a weakly negative degree-zero current \(\tau\le 0\) with \(d_\beta d_{-\beta}\tau\ge 0\). Likewise, \(T(S)\subset(b,0)\) for some \(b<0\) if and only if \(\widetilde S\) admits a nonnegative, strictly PSH function \(v\ge 0\) with the same automorphy.

The explicit hyperbolic Kato examples make the correspondence concrete. For the intermediate Kato surface defined by
\[
F(z_1,z_2)=\bigl(\lambda z_1+P(z_2),\lambda^k z_2\bigr),
\qquad
|\lambda|>1,\ k\ge 2,
\]
one sets
\[
v(z_1,z_2)=-\log\bigl(-\log|z_2|\bigr),
\qquad
F^*v=v-\log k,
\]
and hence
\[
u=-e^{-v}=-\frac1{\log|z_2|}
\]
is strictly PSH on \(\widetilde S\) with \(u\circ \gamma=ku\). Consequently,
\[
T(S)=(-\infty,b),
\qquad
b=-\log k<0,
\]
and the logarithmic \(1\)-form
\[
\theta=\frac{dz_2}{z_2}\otimes s
\]
lies in \(H^0(S,\Omega^1(\log D)\otimes L_b)\), with \(D=\{z_2=0\}\) and \(\operatorname{Res}_D(\theta)=1>0\). For an Inoue–Hirzebruch surface with contraction determined by an integer matrix having eigenvalues \(0<|\lambda_2|<1<\lambda_1\), one constructs \(u=G(z)\) with \(u\circ\gamma=\lambda_1u\), obtains
\[
T(S)=(-\infty,b),
\qquad
b=-\log\lambda_1<0,
\]
and a twisted logarithmic form
\[
\theta=\xi\,\frac{dz_1}{z_1}+\eta\,\frac{dz_2}{z_2}
\in H^0\bigl(S,\Omega^1(\log D)\otimes L_b\bigr),
\qquad
D=\{z_1z_2=0\}.
\]

An application recorded in Theorem 4.6 is a new obstruction to bi-Hermitian structures: on a Kato surface of intermediate type one may have
\[
H^0\bigl(S,K_S\otimes L_b\bigr)=0
\quad (\text{so } b\notin T(S)),
\]
yet no bi-Hermitian metric exists. In particular, the necessary index-\(1\) NAC condition of Apostolov–Hitchin–Goto does not suffice in the intermediate Kato case.

## 6. Moduli spaces, dilation surfaces, and invariant measures

Apisa–Salter identify twisted holomorphic \(1\)-forms on Riemann surfaces with dilation surfaces with scaling. If \(C=\{c_1,\dots,c_n\}\) are marked points and \(\chi\in \operatorname{Hom}(\pi_1(\Sigma_{g,n}),\mathbb R_{>0})\), then a twisted holomorphic \(1\)-form determines charts on \(X\setminus C\) whose transition maps lie in
\[
\mathrm{Dil}=\{z\mapsto az+b\mid a\in \mathbb R_{>0},\ b\in \mathbb C\}.
\]
The local holonomy around \(c_i\) is \(\exp(-2\pi b_i)\in \mathbb R_{>0}\), the cone angle is \(2\pi a_i\), and these are packaged into complex cone angles
\[
\kappa_i=a_i+i\,b_i\in \mathbb Z\oplus i\mathbb R.
\]
The stratum of twisted \(1\)-forms with fixed signature \(\kappa=(\kappa_1,\dots,\kappa_n)\) is denoted
\[
\Omega^{tw}\mathcal M_{g,n}(\kappa).
\]
Its real dimension is
\[
\dim_{\mathbb R}\Omega^{tw}\mathcal M_{g,n}(\kappa)=4g-2+2n,
\]
exactly as in the classical translation-surface case [2507.10685].

The period map is defined on the universal cover, or on a suitable framing cover. Choosing a basepoint \(p\in \Sigma_{g,n}\) among the integral zeros and a lift \(\widetilde p\), one associates to \((X,\chi,\omega,\widetilde p)\) the map
\[
\mathrm{Per}(\omega):\pi_1(\Sigma_{g,n},p)\to \mathrm{Dil},
\]
with
\[
\mathrm{Per}(\omega)(\gamma)(z)=\chi(\gamma)z+\lambda(\gamma),
\qquad
\lambda(\gamma)=\int_{\widetilde\gamma}\omega.
\]
Here \(\lambda\in Z^1(\pi_1;\mathbb C_\chi)\) is a twisted cocycle for the \(\pi_1\)-module \(\mathbb C_\chi\), and the resulting map
\[
\mathrm{Period}:\widetilde{\Omega^{tw}\mathcal M_{g,n}(\kappa)}\longrightarrow \operatorname{Hom}^\circ(\pi_1,Dil)
\]
is a local diffeomorphism away from trivial loci. With chosen generators \((a_1,b_1,\dots,a_g,b_g)\), one gets local coordinates
\[
(x_i=\chi(a_i),\,y_i=\chi(b_i),\,\alpha_i=\lambda(a_i),\,\beta_i=\lambda(b_i))
\]
subject to a single twisted cocycle relation.

A central technical input is the calculation
\[
H^1\bigl(\operatorname{Mod}_{g,n+1};\,H_1(\Sigma_{g,n};A)\bigr)\cong A,
\]
for \(A=\mathbb Z\) or a characteristic-\(0\) field, generated by the change-of-winding-number cocycle. The cocycle is described using a global nonvanishing vector field \(\xi_0\) with zero winding around all but the last puncture:
\[
C(f)([\alpha])=
W_{\xi_0}\bigl(f(\alpha)\bigr)-W_{\xi_0}(\alpha).
\]
Passing to an \(N\)-framed mapping class subgroup makes this cocycle a coboundary, so its pullback vanishes. This is the cohomological mechanism behind the existence of invariant volume forms on suitable finite covers.

The measure-theoretic outcome is a Masur–Veech analogue. Over the base of characters \(\chi\in H^1(\Sigma_{g,n};\mathbb R_{>0})^\circ\) one has the Lebesgue-class Haar measure \(d\chi\), while over each \(\chi\) one has the twisted cocycle space \(Z^1_\chi(\pi_1;\mathbb R)\) of real dimension \(2g-1\). Lemma 4.5 shows that the determinant line of this bundle is canonically trivial, rationally, yielding a nowhere-vanishing fiberwise Lebesgue measure \(|\omega_\chi|\). The product
\[
d\mathrm{Vol}=d\chi\otimes |\omega_\chi|
\]
defines an \(SL(2,\mathbb R)\)-invariant Lebesgue-class measure on \(\operatorname{Hom}^\circ(\pi_1,Dil)\), and pulling back by the period map gives such a measure on the appropriate cover of \(\Omega^{tw}\mathcal M_{g,n}(\kappa)\).

The main existence theorem states that if \(\Omega^{tw}\mathcal M_{g,n}(\kappa)\) has at least one integral zero, then after passing to an explicit finite cover defined by an \(N\)-framed mapping class subgroup, there exists a full-support \(SL(2,\mathbb R)\)-invariant Borel measure of Lebesgue class on the complement of the translation-and-homothety locus. For \(g\ge 3\), the associated invariant section of the top exterior power exists if and only if the subgroup is contained in an \(N\)-framed mapping class group; in that case the measure is unique up to overall constant and is ergodic. A further corollary gives a necessary and sufficient condition for the unscaled dilation stratum to admit an \(SL(2,\mathbb R)\)-invariant Lebesgue class measure: the existence of a measurable area function that is \(SL(2,\mathbb R)\)-invariant and homogeneous of degree \(2\) under diagonal scaling [2507.10685].

Taken together, these results show that twisted holomorphic \(1\)-forms form a mathematically coherent class across several domains. On punctured curves they govern twisted de Rham cohomology and resonance; on class VII surfaces they control residues, Lee-class bounds, and geometric obstructions; and on higher-genus moduli spaces they organize strata of dilation surfaces with period coordinates and invariant measures.

Source: https://www.emergentmind.com/topics/twisted-holomorphic-1-forms