---
title: Left-Invariant Gauss Map in Lie Groups
url: https://www.emergentmind.com/topics/left-invariant-gauss-map
type: topic
---

# Left-Invariant Gauss Map in Lie Groups

Searching arXiv for recent and foundational papers on left-invariant Gauss maps in Lie groups and related harmonicity results.
A left-invariant Gauss map is the Gauss map naturally attached to an immersed oriented surface in a Lie group by transporting the unit normal from each point back to the identity via left translation and then viewing the result in the unit sphere of the Lie algebra. In this form, it generalizes the classical Euclidean Gauss map and encodes extrinsic geometry in a way adapted to the ambient group structure. In the literature, the same object also appears under the name “generalized Gauss map,” especially in settings where harmonicity, integrability, and stability are studied for surfaces in Lie groups [2406.09927], [1506.03743].

## 1. Definition and Lie-algebraic formulation

Let \(M\) be a Riemannian Lie group with a left-invariant metric, and let \(\{E_1(q),E_2(q),E_3(q)\}\) be a left-invariant orthonormal frame, with \(E_i(q)=qE_i\). If \(\varphi:\Sigma\to M\) is an immersion of an orientable surface and \(N\) is a unit normal vector field, the Gauss map is defined by left-translating \(N(p)\in T_{\varphi(p)}M\) to the identity. In the notation of the Lie algebra \(\mathfrak m\), this is

\[
N^e(p)=\sum_{i=1}^3\langle N(p),E_i(\varphi(p))\rangle E_i \equiv \varphi(p)^{-1}N(p),
\]

which gives a map

\[
N^e:\Sigma\to S^2\subset \mathfrak m.
\]

This is the basic left-invariant Gauss map construction in a 3-dimensional Lie group with left-invariant metric [1506.03743].

In the bi-invariant setting considered for constant-mean-curvature surfaces in a 3-dimensional Lie group \(\mathbb G^3\), the same construction is written as

\[
\mathcal N:\Sigma \to \mathbb S^2 \subset \mathfrak g,\qquad
\mathcal N(p)=d(L_{p^{-1}})_p(\eta(p)),
\]

where \(\eta\) is the global unit normal and \(\mathfrak g\cong T_e\mathbb G^3\) is the Lie algebra. Because the metric is left-invariant, the translated normal remains unit length, so the image lies in \(\mathbb S^2\subset\mathfrak g\). In this terminology, \(\mathcal N\) is exactly the left-translation-based Gauss map, and the designation “generalized Gauss map” emphasizes that it extends the usual Euclidean normal map to Lie groups with bi-invariant metrics [2406.09927].

A stereographic parameterization is often used. If \(N^e\equiv (N_1,N_2,N_3)\), then

\[
g=\frac{N_1+iN_2}{1-N_3},
\]

and conversely

\[
N^e=\frac{1}{1+|g|^2}\bigl(g+\overline g,\,-i(g-\overline g),\,-1+|g|^2\bigr).
\]

This complex representation is central in integrability formulas and harmonicity criteria [1506.03743].

## 2. Differential geometry and the invariant shape operator

The left-invariant Gauss map differs from the Euclidean Gauss map at the level of its differential. For an immersed surface in a Lie group, Ripoll’s formula expresses the derivative of the Gauss map as

\[
dL_p \circ d\mathcal N_p = -(A_p+\alpha_p),
\]

where \(A\) is the ordinary shape operator and \(\alpha\) is the invariant shape operator defined by

\[
\alpha_p(X)=\nabla_X \tilde\eta,
\]

with \(\tilde\eta\) the left-invariant extension of the normal field. Thus the derivative of the Gauss map is controlled not only by the second fundamental form but also by a correction term coming from the ambient Lie-group geometry [2406.09927].

This correction term disappears in the commutative case. If \(\mathbb G\) is commutative, then \(\alpha\equiv 0\), and the generalized Gauss map is the ordinary Gauss map. In noncommutative bi-invariant groups, \(\alpha\) may be nonzero and encodes ambient group geometry [2406.09927].

In the 3-dimensional bi-invariant case, the matrix \(\alpha_B=(a_{ij})\) satisfies

\[
a_{ij}=\langle \nabla_{W_i}W_{n+1},W_j\rangle,
\]

and \(\alpha_B\) is antisymmetric, so that

\[
\langle \alpha, A\rangle =0.
\]

The paper also records the 3-dimensional classification

\[
\alpha = 0,\qquad
\alpha=\begin{pmatrix}0&1\\-1&0\end{pmatrix},\qquad
\alpha=\begin{pmatrix}0&-1\\1&0\end{pmatrix}.
\]

These formulas isolate precisely how noncommutativity modifies the Gauss-map differential relative to the Euclidean model [2406.09927].

The same structural identity appears explicitly in the Heisenberg group. For an orientable hypersurface \(S\) in a Lie group \(G\) with left-invariant metric, if \(\gamma=dL_p^{-1}(\nu)\) is the Gauss map, then

\[
dL_p\circ d\gamma_p(v)=-(A_\nu(v)+\alpha_\nu(v)).
\]

In that setting, the formula is used to study rank, minimality, and umbilicity for surfaces in \(\mathcal H_3\) [1106.0779].

## 3. Harmonicity, constant mean curvature, and index theory

A central theorem in the bi-invariant setting is the Lie-group analogue of the Ruh–Vilms theorem:

\[
\Sigma \text{ has constant mean curvature } \iff \mathcal N \text{ is harmonic}.
\]

Accordingly, for a constant-mean-curvature surface in a 3-dimensional Lie group with bi-invariant metric, the left-invariant Gauss map becomes a harmonic map into the unit sphere in the Lie algebra [2406.09927].

The variational theory of this harmonic Gauss map supports quantitative index estimates. For tangent vector fields identified with sections of \(\mathcal N^*T\mathbb S^2\) by

\[
\bar\xi(p)=d(L_{p^{-1}})_p(\xi(p)),
\]

the second variation of the energy is expressed as

\[
D^2_\xi \mathcal E
= \int_\Sigma \langle \Delta \xi -K_\Sigma \xi,\xi\rangle
- \int_\Sigma F(\xi)\, d\Sigma,
\]

with

\[
F(\xi) = (\|A\|^2+c)\|\xi\|^2-\|A\xi\|^2+2\langle A\xi,\alpha\xi\rangle,
\qquad
c=
\begin{cases}
0,& \mathbb G^3 \text{ Abelian},\\
1,& \text{otherwise}.
\end{cases}
\]

After simplification, the key formula becomes

\[
D^2_\xi \mathcal E
= -(4H^2+c)\int_\Sigma \|\xi\|^2\,d\Sigma
+ \int_\Sigma \bigl(2H\langle A\xi,\xi\rangle-\langle A\xi,\alpha\xi\rangle\bigr)\,d\Sigma.
\]

For harmonic \(\xi\), this is the test formula used in the index argument [2406.09927].

The main theorem for closed surfaces states that if \(\Sigma\) is a closed constant-mean-curvature surface in a 3-dimensional Lie group with bi-invariant metric, and in the Abelian case \(H\neq 0\), then

\[
\operatorname{Ind}_{\mathcal E}(\mathcal N)\ge g(\Sigma).
\]

Thus the energy index of the harmonic left-invariant Gauss map is bounded below by the topological genus. For complete noncompact constant-mean-curvature surfaces, the paper proves the analogue

\[
\operatorname{Ind}_{\mathcal E}(\mathcal N)\ge \frac{1}{2}\dim \mathcal H^1(\Sigma),
\]

where \(\mathcal H^1(\Sigma)\) is the space of \(L^2\)-harmonic \(1\)-forms, and \(\dim \mathcal H^1(\Sigma)\ge g(\Sigma)\) when \(\Sigma\) has \(g(\Sigma)\) handles [2406.09927].

An application in \(\mathbb S^3\) is that if a closed constant-mean-curvature surface has stable generalized Gauss map, then it must be a geodesic sphere [2406.09927].

## 4. Integrability and reconstruction from Gauss data

The left-invariant Gauss map is not merely an extrinsic invariant; it can be part of a reconstruction theory for surfaces in Lie groups. For surfaces with positive extrinsic curvature \(K>0\), one chooses a conformal coordinate \(z\) for the second fundamental form \(II\), writes

\[
II=2\rho\,|dz|^2,
\]

and obtains the Weingarten-based reconstruction formula

\[
\partial_z=\frac{i}{\sqrt K}\,N\times_\varphi \nabla_{\partial_z}N.
\]

This identity is the basic mechanism by which the immersion can be recovered from the Gauss map and the extrinsic curvature, provided the compatibility equations hold [1506.03743].

The corresponding integrability condition is

\[
N\times_{\varphi}\left(\nabla_{\partial_z}\left(\frac{1}{\sqrt K}\,\nabla_{\partial_z}N\right)
+\nabla_{\partial_{\overline z}}\left(\frac{1}{\sqrt K}\,\nabla_{\partial_z}N\right)\right)=0.
\]

When the immersion is written in a left-invariant frame as

\[
\varphi^{-1}\varphi_z=a_1E_1+a_2E_2+a_3E_3,
\]

the coefficients \(a_i\) are determined by the Gauss map \(g\), the curvature \(K\), and the Christoffel symbols of the left-invariant frame. In this formulation the immersion is determined from Gauss data up to left translations [1506.03743].

The corresponding existence theorem states that if \(M\) is a Riemannian Lie group with left-invariant orthonormal frame, \(\Sigma\) is simply connected, a map \(N^e:\Sigma\to S^2\subset\mathfrak m\) and a positive function \(K\) are given, the coefficients \(a_1,a_2,a_3\) solving the reconstruction system exist, and the integrability condition holds, then there exists a unique immersion \(\varphi:\Sigma\to M\), up to left translations, such that \(N^e\) is its Gauss map and \(K\) is its extrinsic curvature [1506.03743].

For unimodular Lie groups, the Lie algebra brackets are written in an orthonormal basis as

\[
[E_2,E_3]=c_1E_1,\qquad [E_3,E_1]=c_2E_2,\qquad [E_1,E_2]=c_3E_3,
\]

and the Levi-Civita connection is encoded by constants \(\mu_1,\mu_2,\mu_3\). This includes \(\mathbb R^3\), \(\mathbb S^3\), Berger spheres, \(\widetilde{PSL}_2(\mathbb R,\tau)\), \(\mathrm{Nil}_3(\tau)\), and \(\mathrm{Sol}_3\). In that setting, the compatibility equations can be rewritten as a scalar PDE for \(g\) and \(K\) [1506.03743].

A particularly rigid specialization occurs in \(\mathbb S^3\): a surface immersed in \(\mathbb S^3\) has positive constant extrinsic curvature if and only if its Gauss map is harmonic into the Riemann sphere. The harmonicity equation is

\[
g_{z\overline z}-2\frac{\overline g}{1+|g|^2}\,g_z g_{\overline z}=0.
\]

The same paper also establishes correspondences between simply connected surfaces in \(\mathbb S^3\) and in \(\mathbb R^3\) with the same Gauss map and the same conformal structure on \(II\), subject to the stated curvature restrictions [1506.03743].

## 5. Explicit models: the Heisenberg group and rank phenomena

In the Heisenberg group \(\mathcal H_3\), the left-invariant Gauss map becomes fully explicit. The group is equipped with the left-invariant orthonormal frame

\[
E_1 = \partial_x-\frac{y}{2}\partial_z,\qquad
E_2 = \partial_y+\frac{x}{2}\partial_z,\qquad
E_3=\partial_z,
\]

and metric

\[
ds^2 = dx^2+dy^2+\left(dz+\frac{y\,dx-x\,dy}{2}\right)^2.
\]

For a graph \(X(x,y)=(x,y,f(x,y))\), one sets

\[
p=f_x+\frac{y}{2},\qquad q=f_y-\frac{x}{2},\qquad
w=\sqrt{1+p^2+q^2},
\]

so the unit normal is

\[
\nu = \frac{1}{w}\left(-p\,E_1-q\,E_2+E_3\right).
\]

The Gauss map is then

\[
\gamma(x,y)=\frac{1}{w}(-p,-q,1)\in S^2\subset\mathfrak h_3.
\]

Here the left-invariant definition is not auxiliary: it is the coordinate system in which the normal becomes a Lie-algebra-valued object amenable to algebraic analysis [1106.0779].

Several classification results in \(\mathcal H_3\) are formulated in terms of this Gauss map. First, the vertical plane is the unique connected surface in \(\mathcal H_3\) with constant Gauss map. Second, there are no totally umbilical surfaces in \(\mathcal H_3\). Third, for a graph \(z=f(x,y)\), minimality is equivalent to

\[
\operatorname{tr}(dL_p\circ d\gamma_p)=0,
\]

which is another form of the minimal surface equation

\[
(1+q^2)f_{xx}-2pq\,f_{xy}+(1+p^2)f_{yy}=0.
\]

Thus the Gauss map converts the mean-curvature condition into an algebraic statement about the trace of its differential [1106.0779].

The Heisenberg paper also organizes minimal graphs by the rank of the Gauss map. Rank \(0\) corresponds to constant Gauss map, hence to vertical planes. Rank \(1\) is characterized by

\[
f_{xx}f_{yy}-f_{xy}^2+4=0,
\]

and leads to a ruled-family classification. The minimal graphs in \(\mathcal H_3\) whose Gauss map has rank \(1\) are explicitly described by

\[
f(x,y)=2k\,y-\;x+\;k\ln\!\bigl(y+\sqrt{1+y^2}\bigr)+\sqrt{1+y^2},
\qquad k\in\mathbb R.
\]

In this example, the left-invariant Gauss map functions simultaneously as a differential invariant, a rigidity criterion, and a classification device [1106.0779].

## 6. Scope, terminology, and related but distinct Gauss maps

The expression “Gauss map” in homogeneous 3-manifold geometry does not always mean a left-invariant Gauss map. A useful boundary of the concept is provided by the work on \(\widetilde{\mathrm{PSL}_2(\mathbb R)}\). There, a canonical Gauss map is defined for surfaces in the universal cover of \(\mathrm{PSL}_2(\mathbb R)\) endowed with a left-invariant Riemannian metric having a 4-dimensional isometry group, but the paper states explicitly that this Gauss map is not related to the Lie group structure [1305.1491].

In that construction, the map is built from a distinguished frame adapted to the fibration over \(H^2(\kappa)\), not by translating the normal to the identity. The defining theorem produces a unique map \(\Pi:UE\to \bar{\mathbb C}\) with natural equivariance under ambient isometries, and for a surface with unit normal \(N\) the Gauss map is \(g=\Pi\circ N\). For nowhere vertical critical constant-mean-curvature immersions, this Gauss map is harmonic into the hyperbolic disk \(D\), satisfying

\[
(1-|g|^2)g_{z\bar z}+2\bar g\,g_z\,\bar g_{\bar z}=0.
\]

The paper also derives a Weierstrass-type representation formula from such harmonic maps [1305.1491]. This is closely related to the general theory of harmonic Gauss maps, but it is not a left-invariant Gauss map in the Lie-algebraic sense.

An even clearer contrast occurs for free boundary minimal surfaces in the Euclidean unit ball. There the Gauss map is the standard normal map

\[
\nu:\Sigma\to \mathbb S^2,
\]

and the main theorem concerns the Jacobi–Steklov operator: if the span of the components \(\{\nu_1,\nu_2,\nu_3\}\) is an eigen-subspace, then the surface is rotationally symmetric. The paper expressly does not discuss a left-invariant Gauss map or any Lie-group invariant analogue [1711.05682].

These distinctions matter terminologically. A left-invariant Gauss map is specifically the normal field transported to the identity by left translation and viewed in the Lie algebra. Harmonic Gauss maps in other homogeneous spaces may share formal features—equivariance, harmonicity, representation formulas, or spectral rigidity—but they need not be left-invariant in this sense.

Source: https://www.emergentmind.com/topics/left-invariant-gauss-map