---
title: Parabolic Representation Pair Variety
url: https://www.emergentmind.com/topics/parabolic-representation-pair-variety
type: topic
---

# Parabolic Representation Pair Variety

A parabolic representation pair variety, in the complex-algebraic sense developed for logarithmic pairs and parabolic Higgs bundles, is the geometric object obtained by attaching a representation into \(PU(n,1)\) to a logarithmic pair \((X,D)\) via a stable parabolic Higgs bundle whose parabolic data are concentrated along the compact part of the boundary. For a smooth complex projective variety \(X\) with \(D=D^p+D^c\) a simple normal crossing divisor, the construction passes through the root stack \(X[\sqrt[p_i]{D_i^c}]\), turns the equality case of a parabolic Bogomolov--Gieseker inequality into a flat trace-free adjoint harmonic bundle and a principal \(PU(n,1)\)-variation, and produces a period map to the complex unit ball. The local behavior of this period map records branching prescribed by the rational parabolic weights, while the converse direction reconstructs positivity of the log canonical class and parabolic polystability from a branched complex-hyperbolic structure of the same type [2606.06778].

## 1. Logarithmic pair, parabolic data, and the canonical Higgs bundle

The basic input is a smooth complex projective variety \(X\) of dimension \(n\) together with a simple normal crossing divisor
\[
D=D^p+D^c.
\]
The divisor \(D^p\) is the cusp part: it is removed from the open subset \(X^o:=X\setminus D^p\), and it carries logarithmic poles but no parabolic filtration. The divisor \(D^c\) is the compact orbifold part: each irreducible component \(D_i^c\) carries a single rational parabolic weight
\[
\alpha_i=\frac{q_i}{p_i}\in[0,1),\qquad \gcd(p_i,q_i)=1.
\]

The underlying system of Hodge bundles is
\[
E:=\Omega_X^1(\log D^p)\oplus\mathcal O_X,
\qquad
E^{1,0}=\Omega_X^1(\log D^p),\quad E^{0,1}=\mathcal O_X,
\]
equipped with the tautological Higgs field
\[
\theta:E\to E\otimes\Omega_X^1(\log D),\qquad \theta(a,b)=(0,a).
\]
The associated filtered parabolic Higgs bundle is
\[
E_*=(\Omega_X^1(\log D^p)\oplus\mathcal O_X)_*.
\]

Its parabolic structure is highly concentrated: it is trivial along \(D^p\), while along each \(D_i^c\) the only non-trivial step occurs on the conormal line
\[
N^*_{D_i^c/X}\subset \Omega_X^1|_{D_i^c}\subset E|_{D_i^c},
\]
with weight \(q_i/p_i\); all other directions have weight \(0\). Thus the only compact parabolic weights occur on conormal lines to \(D^c\). This concentration of weights on conormal directions governs the local branching behavior of the representation and period map attached to the pair \((X,D)\) [2606.06778].

## 2. Root stacks and the parabolic Bogomolov--Gieseker equality

To package the rational weights intrinsically, one passes to the root stack
\[
\mathfrak X:=X[\sqrt[p_i]{D_i^c}]_{i\in I},
\]
with coarse moduli morphism \(\pi:\mathfrak X\to X\). The stack \(\mathfrak X\) is smooth and carries finite stabilizers \(\mu_{p_i}\) along the stacky divisors \(\mathfrak D_i^c\) lying over \(D_i^c\). On the open stack
\[
\mathfrak X^o:=\mathfrak X\setminus \pi^{-1}(D^p),
\]
the parabolic Higgs bundle \(E_*\) corresponds, by the parabolic--root-stack dictionary, to a logarithmic orbifold Higgs bundle \((\mathcal E,\theta_{\mathcal E})\) with \(\mathcal E^{0,1}=\mathcal O_{\mathfrak X}\).

In a local root coordinate
\[
z_i=w_i^{p_i},
\]
the conormal line is represented by the equivariant frame
\[
\eta_i:=w_i^{-q_i}dz_i=p_i\,w_i^{p_i-q_i-1}dw_i.
\]
This shows that \(\mathcal E^{1,0}\) differs from \(\Omega_{\mathfrak X}^1(\log D^p)\) unless \(q_i=p_i-1\). More precisely, the morphism
\[
\iota:\mathcal E^{1,0}\to \Omega_{\mathfrak X}^1(\log D^p)
\]
is generically an isomorphism but degenerates along \(\mathfrak D_i^c\), with determinant vanishing of order \(p_i-q_i-1\). A common misconception is that the orbifold Higgs bundle is always the logarithmic cotangent bundle on the root stack; it is exactly so only in the standard orbifold case \(q_i=p_i-1\).

The parabolic Bogomolov--Gieseker inequality is naturally formulated on \(\mathfrak X\). Writing \(r:=\operatorname{rk}(E)=n+1\), one has
\[
\left(\frac{\operatorname{par}_1(E_*)^2}{2(n+1)}-\operatorname{par}_2(E_*)\right)\cdot c_1(L)^{n-2}\ge 0
\]
for any ample line bundle \(L\). The equality case is equivalent to
\[
\operatorname{orb}_2(\operatorname{End}_0(\mathcal E))\cdot c_1(L)^{n-2}=0,
\]
using
\[
\operatorname{orb}_2(\operatorname{End}F)=2r\cdot \operatorname{orb}_2(F)-\operatorname{orb}_1(F)^2
\]
for any orbifold vector bundle \(F\) of rank \(r\). The working hypothesis is that \((E_*,\theta)\) is \(L\)-polystable and satisfies the equality
\[
\left(\frac{\operatorname{par}_1(E_*)^2}{2(n+1)}-\operatorname{par}_2(E_*)\right)\cdot c_1(L)^{n-2}=0,
\]
and it is precisely this equality that triggers the representation-theoretic uniformization package [2606.06778].

## 3. Equality case: flat adjoint bundle, principal \(PU(n,1)\)-variation, and period map

Under the equality hypothesis, the adjoint Higgs bundle
\[
(\operatorname{End}_0(\mathcal E),\theta_{\operatorname{End}_0})
\]
is polystable with trivial characteristic numbers and therefore carries an adapted tame pluri-harmonic metric. Its Hitchin--Simpson connection is flat. This produces a semisimple local system with real projective form \(PU(n,1)\).

The resulting structure is encoded as a principal \(PU(n,1)\)-variation of Hodge structure on \(\mathfrak X^o\). In the formulation used here, such a variation consists of a principal \(G=PGL(V)\)-bundle with Hodge reduction to \(K\subset G\), where
\[
V=V^{1,0}\oplus V^{0,1},\qquad \dim V^{1,0}=n,\quad \dim V^{0,1}=1,
\]
together with a Higgs/Kodaira--Spencer field
\[
\tau:T_{\mathfrak X}(-\log D^p)\to P\times_K \mathfrak g^{-1,1}\simeq \operatorname{Hom}(\mathcal E^{1,0},\mathcal E^{0,1}).
\]
The real form \(G_0=PU(n,1)\) with maximal compact \(K_0=U(n)\) has symmetric space
\[
G_0/K_0\simeq \mathbb B^n.
\]

Consequently one obtains a representation
\[
\rho:\pi_1^{\operatorname{orb}}(\mathfrak X^o)\to PU(n,1)
\]
and a \(\rho\)-equivariant holomorphic period map
\[
\mathcal P:\widetilde{\mathfrak X^o}\to PU(n,1)/U(n)\simeq \mathbb B^n.
\]
Its differential is exactly \(\tau\). Because \(\tau\) is dual to \(\iota\), the differential is generically an isomorphism and degenerates only along \(\mathfrak D^c\). This identifies the parabolic representation pair variety, in this setting, with a canonical representation-theoretic object attached to the logarithmic pair through the equality locus of the parabolic Chern inequality [2606.06778].

## 4. Local normal form, branching, and the distinction between orbifold and branched uniformization

The local geometry of the period map is controlled by the same root coordinates that encode the parabolic weights. In the chart \(z_i=w_i^{p_i}\), one has
\[
\iota(\eta_i)=p_i\,w_i^{p_i-q_i-1}dw_i.
\]
Since \(\tau\) is dual to \(\iota\), the normal component of \(d\mathcal P\) is, up to a unit,
\[
w_i^{p_i-q_i-1}dw_i.
\]
Integrating gives the normal form
\[
\xi_i\circ \mathcal P = u_i(w)\,w_i^{p_i-q_i},
\qquad u_i(0)\neq 0.
\]

This exhibits a ramification index \(p_i-q_i\) along the lift of \(\mathfrak D_i^c\). Passing to coarse spaces, the induced map
\[
X^o\to \mathbb B^n/\Gamma
\]
has branch index \(p_i-q_i\) along \(D_i^c\). The general equality case therefore yields a branched complex-hyperbolic structure.

The unramified orbifold ball uniformization appears only in the standard case
\[
q_i=p_i-1\quad\text{for all }i.
\]
Then \(\mathcal E^{1,0}=\Omega_{\mathfrak X}^1(\log D^p)\), so \(\tau\) is everywhere an isomorphism on \(\mathfrak X^o\). The period map becomes an unramified orbifold local isometry; its pull-back metric is complete at the cusp boundary and has finite volume; since \(\mathbb B^n\) is complete and simply connected, \(\mathcal P\) identifies the orbifold universal cover with \(\mathbb B^n\), and \(\rho\) has discrete lattice image \(\Gamma\subset PU(n,1)\). In that case \((\mathfrak X,D^p)\) is the orbifold toroidal compactification of \(\mathbb B^n/\Gamma\).

Outside the standard case, the descended metric is of cone type of order \(q_i/p_i\) along \(D_i^c\). The resulting structure is not an unramified orbifold ball quotient but a branched complex-hyperbolic structure. This sharply distinguishes the orbifold equality case from the general parabolic equality case [2606.06778].

## 5. Converse direction: mixed Poincaré--cone currents, positivity, and parabolic stability

The converse statement starts from geometry rather than stability. Suppose \((X,D^p,D^c,\{q_i/p_i\})\) carries a branched complex-hyperbolic structure of type
\[
\Delta:=D^p+\sum_i \frac{q_i}{p_i}D_i^c,
\]
meaning that on the root stack \(\mathfrak X\) the representable universal cover admits a \(\rho\)-equivariant map
\[
\mathcal P:M\to \mathbb B^n
\]
with local normal form
\[
\xi_i\circ \mathcal P=u_i(w)w_i^{m_i},\qquad m_i=p_i-q_i,
\]
and cusp Poincaré growth along \(D^p\). Let \(\omega_{\operatorname{br}}\) be the descended pull-back of the Bergman metric. Then the closed positive current
\[
\Theta:=\frac{n+1}{2\pi}\,\omega_{\operatorname{br}}
\]
extends across \(|D|\) and represents
\[
[\Theta]=c_1(K_X+\Delta)=c_1\!\left(K_X+D^p+\sum_i \frac{q_i}{p_i}D_i^c\right).
\]

Near \(D^p\), \(\Theta\) has at most Poincaré cusp growth. Near \(D_i^c\), it is mutually bounded with
\[
|x_i|^{-2(q_i/p_i)}\,i\,dx_i\wedge d\bar x_i
\]
in adapted coordinates. This mixed Poincaré--cone growth implies zero Lelong numbers along \(D\), and by regularization the class \(c_1(K_X+\Delta)\) is automatically big and nef.

A class \(\alpha\in H^{1,1}(X,\mathbb R)\) is called \(\Delta\)-admissible if it is big and nef and contains a closed positive current \(T\) that is a smooth Kähler form on \(X^\circ\) and is dominated, in the sense of currents, by a fixed mixed Poincaré--cone model. For every \(\Delta\)-admissible class \(\alpha\), the Hodge metric \(h\) obtained from \(\omega_{\operatorname{br}}\) is adapted and acceptable for \(E_*\), and its Hitchin--Simpson curvature is trace-free flat on \(X^\circ\). The parabolic Chern--Weil identities then yield parabolic \(\mu_\alpha\)-polystability and the parabolic Chern equality
\[
\frac{\operatorname{par}_1(E_*)^2}{2(n+1)}-\operatorname{par}_2(E_*)=0,
\]
equivalently
\[
2\,\operatorname{par}_2(E_*^{1,0})-\frac{n}{n+1}\operatorname{par}_1(E_*^{1,0})^2=0.
\]
In particular, \(E_*\) is parabolic \(\mu_{c_1(K_X+\Delta)}\)-polystable.

The representation itself is controlled by the boundary decomposition. Along \(D^p\), monodromy is parabolic (unipotent), and the cusp divisor is forced to be smooth, with disjoint components. Along each \(D_i^c\), the local orbifold stabilizer has order \(p_i\), and the normal character has order \(p_i\) with ramification index \(p_i-q_i\). Conormal lines and their weights therefore encode both the local branching and the orbifold data of the canonical \(PU(n,1)\)-representation [2606.06778].

## 6. Other established meanings of the term

The expression “parabolic representation pair variety” is not confined to the complex-hyperbolic uniformization framework. In the literature it appears in several mathematically distinct settings, and these usages should be kept separate.

In the theory of symmetric pairs and double flag varieties, the phrase is used for the variety
\[
G/P_1\times G/P_2
\]
attached to parabolic subgroups \(P_1,P_2\subset G\). For a simple group \(G\), there exist \(P_1,P_2\) with
\[
P_1\cap P_2=Q\subset K,\qquad P_1P_2\ \text{dense in}\ G
\]
if and only if \((G,K)\) is a Hermitian symmetric pair, and the closed \(K\)-orbit through \((eP_1,eP_2)\) is \(K/Q\) inside the open dense \(G\)-orbit [1204.1118].

For punctured surfaces and Betti moduli, the term denotes the closed subvariety
\[
\mathfrak{PRP}(r,\mathbf d)\subset R(\Gamma,G)\times \mathrm{FL}(r,\mathbf d)
\]
consisting of pairs \((\rho,\mathbf P)\) with prescribed parabolic type at punctures. In that setting the Zariski tangent space at \((\rho,\mathbf P)\) has complex dimension \(r^2(2g+n-1)\), the analytic germs are controlled by DGLAs of parabolic logarithmic flat bundles, and the weighted moduli space is constructed by GIT, with a Kobayashi--Hitchin-type theorem for polystable parabolic representation pairs [2509.20791].

In the representation theory of \(GL_n(F)\), a related “pair variety” is the subvariety
\[
\mathcal Y_{m,m'}=\mathcal D'_{m,m'}\cap \mathcal D''_{m,m'}
\subset \mathcal X(V\oplus V'),
\]
used to formulate the geometric conditions \(SG(m,m')\) and \(IG(m,m')\) that are conjecturally related to subrepresentation-additivity and irreducibility of normalized parabolic induction [1911.04281].

In complex hyperbolic free-group theory, the parabolic locus for representations of the three-punctured sphere in \(PU(2,1)\) has real dimension \(5\). It is parametrized by balanced tetrahedra \((\theta,\phi,\psi)\) together with unit eigenphases \((\lambda_A,\lambda_B)\), and contains a one-parameter family where
\[
A,\ B,\ AB,\ AB^{-1},\ AB^2,\ A^2B,\ [A,B]
\]
are all parabolic [1312.3795].

These parallel usages share the motif of coupling parabolic boundary data with a representation-theoretic or geometric object, but they differ fundamentally in ambient category, deformation theory, and intended applications. In the sense arising from orbifold uniformization, the term refers specifically to the canonical \(PU(n,1)\)-representation attached to a logarithmic pair through the equality case of the parabolic Bogomolov--Gieseker inequality, together with its period map and local branching structure [2606.06778].

Source: https://www.emergentmind.com/topics/parabolic-representation-pair-variety