---
title: 'Chen–Gackstatter Torus: Minimal Surface Theory'
url: https://www.emergentmind.com/topics/chen-gackstetter-torus
type: topic
---

# Chen–Gackstatter Torus: Minimal Surface Theory

Searching arXiv for relevant papers on the Chen–Gackstatter torus and related deformations in \(\mathbb{R}^4_1\) and \(\mathbb{R}^4\).
The Chen–Gackstatter torus is the classical complete minimal surface of genus one with a single Enneper-type end and total Gaussian curvature $-\int K\,dM=8\pi$, often described as a “torus with one end.” In the literature surveyed here, the term also encompasses higher-codimension analogues and deformations, especially stationary surfaces in Minkowski $4$-space $\mathbb{R}^4_1$ and minimal surfaces in Euclidean $\mathbb{R}^4$, where the genus-one, one-end, total-curvature-$8\pi$ configuration persists but the ambient geometry introduces new Gauss-map data, new period conditions, and a more intricate embeddedness theory [1212.6802], [2507.12914].

## 1. Classical object in $\mathbb{R}^3$

The classical Chen–Gackstatter surface in $\mathbb{R}^3$ is the first example of a complete minimal surface of genus $1$ with a single Enneper-type end. Its topology is that of a torus punctured at one point, and its total Gaussian curvature is
\[
-\int K\,dM=8\pi,
\]
which is the least possible for genus one in the framework discussed in the cited work [1212.6802].

In the algebraic formulation used for its Lorentzian deformation theory, the underlying Riemann surface is the square torus
\[
y^2=x(x-1)(x+1),
\]
equivalently the elliptic curve with parameter $\lambda=-1$. The Weierstrass data are
\[
G=\frac{1}{\rho}\cdot\frac{x}{y}, \qquad \mathrm{d}h=\mathrm{d}x,
\]
with
\[
\rho=\frac{\sqrt{6}\Gamma(\frac{3}{4})}{\Gamma(\frac{1}{4})}\approx 0.8279.
\]
The parameter $\rho$ is chosen so that the horizontal period condition is satisfied, while $\mathrm{d}h$ is exact, so the vertical periods vanish [1212.6802].

The end is of Enneper type with multiplicity $3$, and consequently the surface is not embedded in $\mathbb{R}^3$. The square torus model carries a $D_4$ symmetry group, described in the source as two planar reflections and two $180^\circ$ rotations, and the classical surface has two isolated self-intersection points along the $X_3$-axis [1212.6802]. A later Euclidean $\mathbb{R}^4$ treatment also states that the classical Chen–Gackstatter torus is the unique complete minimal torus in $\mathbb{R}^3$ of genus one with a single end and total curvature $-8\pi$, parametrized by the square torus $(\tau=i)$, complete and proper, and non-embedded [2507.12914].

## 2. Stationary-surface framework in Minkowski $\mathbb{R}^4_1$

In $\mathbb{R}^4_1$, endowed with Lorentzian metric of signature $(-,+,+,+)$,
\[
\langle x,x\rangle=-x_0^2+x_1^2+x_2^2+x_3^2,
\]
the relevant generalization is the class of stationary surfaces: spacelike immersions with zero mean curvature [1212.6802]. The codimension-two setting differs essentially from the $\mathbb{R}^3$ theory because it involves two Gauss maps and correspondingly richer period constraints.

The Weierstrass-type representation used in this setting is formulated in terms of meromorphic functions $\phi,\psi$ and a holomorphic $1$-form $\mathrm{d}h$ on a Riemann surface $M$. If $\phi\neq \overline{\psi}$ on $M$, the poles of $\phi$ and $\psi$ do not coincide, the zeros of $\mathrm{d}h$ coincide with the poles of $\phi$ or $\psi$ with the same order, and the horizontal and vertical period conditions
\[
\oint_\gamma \phi \mathrm{d}h=-\overline{\oint_\gamma \psi \mathrm{d}h}, \qquad 
\mathrm{Re}\oint_\gamma \mathrm{d}h=\mathrm{Re}\oint_\gamma \phi\psi \mathrm{d}h=0
\]
hold along every closed path $\gamma$, then
\[
X=2\mathrm{Re} \int \Big(\phi+\psi, -i(\phi-\psi), 1-\phi\psi,1+\phi\psi\Big)\mathrm{d}h
\]
defines a stationary immersion $X:M\to \mathbb{R}^4_1$, with induced metric
\[
\mathrm{d}s^2=|\phi-\bar\psi|^2\cdot|\mathrm{d}h|^2.
\]
Conversely, any stationary surface arises in this way [1212.6802].

For algebraic stationary surfaces with regular ends, the total curvature is governed by a Jorge–Meeks-type formula:
\[
-\int_{M} K\, dM =4\pi \deg\phi =4\pi \deg\psi
=2\pi\big(2g-2+r+\sum_{j=1}^r d_j\big).
\]
In the genus-one, one-end case with pole order $d+1=4$ at the end, so that $d=3$, this yields
\[
-\int K\,dM=8\pi.
\]
That normalization is used throughout the deformation theory of Chen–Gackstatter-type stationary surfaces [1212.6802].

This formalism makes clear that the Chen–Gackstatter torus in $\mathbb{R}^3$ is not merely a single isolated example but the prototype for a genus-one, one-end, finite-total-curvature configuration whose core invariants can survive in Lorentzian codimension two. A plausible implication is that the one-end $8\pi$ regime functions as a rigidity threshold: sufficiently constrained data lead back to the classical model, while relaxed ambient geometry permits nontrivial deformations.

## 3. Lorentz deformations and Chen–Gackstatter-type surfaces in $\mathbb{R}^4_1$

A central result is that the classical $\mathbb{R}^3$ Chen–Gackstatter surface admits explicit Lorentz deformations into $\mathbb{R}^4_1$ [1212.6802]. If a minimal surface in $\mathbb{R}^3$ has complex differential
\[
X_z\,dz=(\Theta_1,\Theta_2,\Theta_3),
\]
its Lorentz deformation is defined by
\[
\widetilde{X}_z dz=(\Theta_1,\Theta_2,a\Theta_3,b\Theta_3)
\]
where $a,b\in\mathbb{C}$ satisfy $a^2-b^2=1$. Writing
\[
a=\frac{\zeta+\zeta^{-1}}{2}, \qquad b=\frac{\zeta-\zeta^{-1}}{2}, \qquad \zeta\in \mathbb{C}\setminus i\mathbb{R},
\]
and expressing the original $\mathbb{R}^3$ minimal surface by Weierstrass data $\{G,\mathrm{d}h\}$ with $\psi=-1/G$, the deformed data are
\[
\phi=\frac{G}{\zeta}, \qquad \psi=\frac{-1}{\zeta G}, \qquad \widetilde{\mathrm{d}h}=\zeta \mathrm{d}h.
\]
When $\zeta=e^{i\theta}$,
\[
\widetilde{X}_\theta=(X_1,X_2,\cos\theta X_3,-\sin\theta X_3^*),
\]
where $X_3^*$ is the harmonic conjugate of $X_3$ [1212.6802].

Applied to the Chen–Gackstatter surface, this produces a real $2$-parameter family of complete stationary surfaces in $\mathbb{R}^4_1$, each retaining genus one, a unique regular end, and total curvature $-\int K=8\pi$. Up to congruence, the family reduces to an $S^1$-family of non-congruent deformations parameterized by $\operatorname{Arg}(\zeta)$, because deformations with parameters related by $\zeta_2/t=\zeta_1$ for real $t>0$ are congruent in $\mathbb{R}^4_1$ [1212.6802].

These deformations preserve the same $D_4$ symmetry group as the classical example. They also preserve the two self-intersection points of the original $\mathbb{R}^3$ surface, so in general the deformed surfaces still intersect themselves at exactly two points. At the level of the end, however, the Lorentz deformation changes the local geometry more substantially: the Enneper end becomes embedded near the end in $\mathbb{R}^4_1$, even though the whole surface remains non-embedded [1212.6802]. This contrast isolates one of the main geometric novelties of the Lorentzian setting: local end embeddedness need not imply global embeddedness, and the topological obstruction imposed by the Enneper end in $\mathbb{R}^3$ can be partially removed without eliminating all self-intersections.

A converse statement is equally significant. If $\phi\psi$ is constant, then the stationary surface is precisely a Lorentz deformation of a minimal surface in $\mathbb{R}^3$ [1212.6802]. When $\phi\psi\equiv c\in\mathbb{C}\setminus\mathbb{R}$, the surface is not contained in any $3$-dimensional affine subspace. When $\phi\psi$ is negative real, the surface is congruent to a minimal surface in $\mathbb{R}^3$ [1212.6802]. This criterion provides the precise bridge between the Lorentzian genus-one theory and the classical minimal-surface model.

## 4. Divisor structure, period calculus, and the real $4$-parameter family

For Chen–Gackstatter-type stationary surfaces
\[
X:T^2-\{P\}\to \mathbb{R}^4_1
\]
with one regular end and total curvature $-\int K=8\pi$, the source torus is represented as
\[
T_\lambda=\{[x,y,1]\in \mathbb{C}P^2\mid y^2=x(x-1)(x-\lambda)\}, \qquad \lambda\in \hat{\mathbb{C}\setminus\{0,1,\infty\}.
\]
The period analysis uses the holomorphic form
\[
\omega=dz=\frac{dx}{y}
\]
and
\[
\Phi=\frac{x\,dx}{y},
\]
with periods $\omega_j=\oint_{\gamma_j}\omega$ and $\Phi_j=\oint_{\gamma_j}\Phi$ on generators $\gamma_1,\gamma_2$ of $H_1(T^2,\mathbb{Z})$. The Legendre relation
\[
\omega_1\Phi_2-\omega_2\Phi_1=8\pi i
\]
and the reduction
\[
x^2\,\frac{dx}{y}\approx \frac{2(\lambda+1)}{3}\Phi-\frac{\lambda}{3}\omega
\]
supply the key algebraic control on the period problem [1212.6802].

Under the normalization $\phi(P)=0$ and $\psi(P)=\infty$, there are two possible divisor configurations [1212.6802]:

| Case | Divisors |
|---|---|
| Case 1 | $(\phi)=P+C_1-V_1-V_2$, $(\psi)=-P-C_2+N_1+N_2$, $(\mathrm{d}h)=-3P+V_1+V_2+C_2$ |
| Case 2 | $(\phi)=2P-V_1-V_2$, $(\psi)=-2P+C_1+C_2$, $(\mathrm{d}h)=-2P+V_1+V_2$ |

In Case 1, the data can be written in terms of six complex parameters $v=(b,c,l,m,y_0,\lambda)$:
\[
\phi=\frac{x-x_0}{b(y-y_0)+c(x-x_0)}, \qquad
\psi=\frac{l(y+y_0)+m(x-x_0)}{x-x_0},
\]
\[
\mathrm{d}h=[b(y-y_0)+c(x-x_0)]\mathrm{d}z \thickapprox cx\mathrm{d}z-(cx_0+by_0)\mathrm{d}z,
\]
\[
\phi\psi \mathrm{d}h=[l(y+y_0)+m(x-x_0)]\mathrm{d}z \thickapprox mx\mathrm{d}z+(-mx_0+ly_0)\mathrm{d}z,
\]
\[
\phi\mathrm{d}h=(x-x_0)\mathrm{d}z.
\]
The period conditions reduce to
\[
\mathrm{Re}[c\Phi_i-(cx_0+by_0)\omega_i]=0,
\]
\[
\mathrm{Re}[m\Phi_i-(mx_0-ly_0)\omega_i]=0,
\]
\[
-\overline{\Phi_i}+\overline{x_0}\overline{\omega_i}=A\Phi_i+B\omega_i,
\]
where
\[
A=bl\left[x_0-\frac{\lambda+1}{3}\right]+cm,\qquad
B=(cl-bm)y_0+bl\left[x_0^2-(\lambda+1)x_0+\frac{2\lambda}{3}\right]-cmx_0.
\]
Regularity requires
\[
|x-x_0|^2 \neq \overline{[b(y-y_0)+c(x-x_0)]}[l(y+y_0)+m(x-x_0)]
\]
away from the designated divisor points [1212.6802].

The special subcase $x_0=0$ yields a sharp characterization. The period conditions simplify to
\[
\mathrm{Re}[c\Phi_i]=0, \qquad \mathrm{Re}[m\Phi_i]=0,
\]
\[
-\overline{\Phi_i}=
\left[mc-\frac{bl}{3}(\lambda+1)\right]\Phi_i+\frac{2bl}{3}\lambda\omega_i.
\]
By Weber’s theorem, one obtains $\lambda=-1$, so the torus is square, and $\phi\psi=l/b$ is constant. By the converse of the Lorentz deformation theorem, the surface is therefore a Lorentz deformation of the classical Chen–Gackstatter surface [1212.6802].

The same paper proves a more general local-existence result. In Case 1, the period conditions define $8$ real equations in $12$ real variables. Writing the period mapping
\[
\mathcal{P}:\mathbb{R}^{12}\to\mathbb{R}^8,
\]
the classical Chen–Gackstatter point corresponds to
\[
v^*=(b,c,l,m,y_0,\lambda)=(\rho,0,-\rho,0,0,-1),
\]
with $\mathcal{P}(v^*)=\vec{0}$. The Jacobian $D\mathcal{P}(v^*)$ has rank $8$, and hence, by the preimage theorem, there exists a smooth real $4$-dimensional family of nearby solutions [1212.6802]. The resulting theorem states:

\[
\text{There exists a (real) 4-parameter family of deformations of the Chen-Gackstatter surface}
\]
\[
\text{which still has genus one, a unique end, with total curvature }8\pi,\text{ and the Gauss maps }\phi,\psi\text{ both have order }1\text{ at the end.}
\]

Regularity $\phi\neq \overline{\psi}$ persists under small deformation by uniform continuity of the stereographic distance on $S^2$ between $(\phi,\psi)$ [1212.6802]. This establishes an explicit distinction between the globally described Lorentz-deformation family and a larger local moduli space detected through the period map.

## 5. Symmetry, uniqueness, and classification constraints

A major classification theorem states that among genus-one Chen–Gackstatter surfaces in $\mathbb{R}^4_1$ with a unique end and total curvature $-\int K\,dM=8\pi$, the deformations obtained from the explicit Lorentz deformation construction are the only ones whose symmetry group $G$ satisfies $|G|>4$ [1212.6802].

The symmetry analysis proceeds through the conformal type of the torus. If $|G|>4$, the torus must be conformally equivalent either to the square torus $(\tau=i,\lambda=-1)$ with automorphism group $D_4$ of order $8$, or to the equilateral torus $(\tau=e^{\pi i/3}, \lambda=\epsilon,\ \epsilon^3=-1)$ with automorphism group $D_6$ of order $12$ [1212.6802]. A key lemma states that in both divisor cases the divisors of $\phi\,dh$ and $\psi\,dh$ are preserved by any symmetry, while the divisors of $dh$ and $\phi\psi\,dh$ are preserved or interchanged. This uses the ambient block-diagonal action $O(2)\times O(1,1)$ preserving tangent and normal limit planes at the end [1212.6802].

The case-by-case consequences are decisive. In Case 1 on the square torus, $D_4$ symmetry forces $C_1=C_2$ at the center of the fundamental square, so $\phi\,dh$ has a double zero at $x_0=0$, and the earlier uniqueness proposition applies; the surface must be a Lorentz deformation of the classical Chen–Gackstatter surface. In Case 1 on the equilateral torus, a $\mathbb{Z}_3$ symmetry would force $V_1=V_2=C_2$ and $N_1=N_2=C_1$, contradicting regularity because $\phi$ and $\psi$ would then have common zeros or poles. In Case 2 on the square torus, there are no solutions. In Case 2 on the equilateral torus, $\mathbb{Z}_3$ symmetry forces the zeros and poles to be located at triangle centers so that $\phi\psi$ is constant; by the converse Lorentz-deformation theorem this would come from a genus-one minimal torus in $\mathbb{R}^3$ with one end on the equilateral torus, excluded by López’s uniqueness and classification as quoted in the source [1212.6802].

The combined effect is a rigidity principle for highly symmetric examples: symmetry larger than order $4$ forces the Minkowski-space theory back to the square-torus Lorentz deformations of the classical $\mathbb{R}^3$ object. This suggests that the broader $4$-parameter family exists in a substantially less symmetric regime.

## 6. Embeddedness, self-intersections, and multiplicity issues

Embeddedness is among the most delicate points in the Chen–Gackstatter theory. In the classical $\mathbb{R}^3$ case, the Enneper-type end of multiplicity $3$ forces non-embeddedness, and the surface has two isolated self-intersection points [1212.6802]. In $\mathbb{R}^4_1$, the situation changes locally but not completely globally.

For the explicit Lorentz-deformed family in $\mathbb{R}^4_1$, the end becomes embedded near infinity, which the source describes as the Lorentz deformation “untangling” the Enneper end, but the surface still self-intersects at the two points inherited from the classical example [1212.6802]. For the larger $4$-parameter family of Case 1, the embeddedness problem remains open [1212.6802].

Case 2 admits a definitive obstruction. If a complete regular algebraic stationary surface
\[
X:T^2-\{P\}\to \mathbb{R}^4_1
\]
with total curvature $-\int K=8\pi$ has Gauss maps of order $2$ at the end, then it is not embedded [1212.6802]. The proof uses the involution
\[
I:[x,y,1]\mapsto [x,-y,1]
\]
on the elliptic curve, which satisfies
\[
I^*(X_z\,dz)=-X_z\,dz.
\]
Hence $X$ is centrally symmetric with center at $(0,0,0,0)$. The three Weierstrass points $(x,y)=(0,0),(1,0),(\lambda,0)$ are fixed by $I$, and their images coincide at the symmetry center, producing a triple point [1212.6802].

The paper further notes a sharp contrast with the Euclidean $\mathbb{R}^3$ monotonicity framework. In $\mathbb{R}^3$, a multiplicity inequality derived from the monotonicity formula bounds point multiplicity by the sum of end multiplicities and excludes triple points in the genus-one, one-end class. In $\mathbb{R}^4_1$, because the metric is indefinite, no direct analogue is available. The authors therefore formulate a conjectural multiplicity inequality: if a complete algebraic stationary surface has multiplicity $m$ at a point and regular ends of multiplicities $d_j$, then
\[
m<\sum_{j=1}^r d_j.
\]
This remains conjectural in the cited work [1212.6802].

A comparison given in the same source underscores that the embeddedness obstruction is specific to the Lorentzian stationary setting rather than to genus one or one-endedness alone. In Euclidean $\mathbb{R}^4$, embedded minimal tori with one end and total curvature $-\int K=8\pi$ are readily available, for example
\[
T^2_{\lambda}-\{P\}=\{(x,y)\mid y^2=x(x-1)(x-\lambda)\}\subset \mathbb{C}^2=\mathbb{R}^4,
\]
which are embedded with a unique end of multiplicity $3$ and curvature $-8\pi$ by Jorge–Meeks [1212.6802].

## 7. Euclidean $\mathbb{R}^4$ generalizations and later developments

A recent Euclidean $\mathbb{R}^4$ development reformulates the genus-one, one-end, total-curvature-$8\pi$ problem in terms of minimal immersions
\[
F(z)=\big(e(z)+\overline{f(z)},\, g(z)+\overline{h(z)}\big)
\]
with holomorphic data satisfying
\[
e'(z)f'(z)+g'(z)h'(z)=0,
\]
where the derivatives are meromorphic on the torus with a common pole at the puncture [2507.12914]. In this framework, complete proper non-holomorphic minimal immersions of a punctured torus into $\mathbb{R}^4$ with one end and total curvature $-8\pi$ are sought through a system of $10$ quadratic or linear equations in $11$ real variables [2507.12914].

For genus-one one-end examples, the Euclidean theory uses two holomorphic Gauss maps $\gamma_\pm:\hat{\Sigma}\to \mathbb{C}P^1$ with degrees $d_\pm$. The total curvature and normal curvature satisfy
\[
-\int_{\Sigma}K^T=2\pi(d_++d_-), \qquad
-\int_{\Sigma}K^N=2\pi(d_+-d_-).
\]
In the examples constructed in that paper, total curvature $-8\pi$ implies $d_++d_-=4$, and the surfaces are not complex for any isometric complex structure on $\mathbb{R}^4$, so both degrees are positive; in fact,
\[
d_+=d_-=2, \qquad \int_\Sigma K^N=0
\]
[2507.12914].

The square torus case is completely solved there. Let $\wp$ be the Weierstrass function on $T_i$ and let
\[
A^2=\frac{3\pi}{2g_2}.
\]
For each $\lambda\in \mathbb{C}\setminus\{0\}$, the map
\[
z\longmapsto
\Big(
A^2\int_{z_0}^z (4\wp^2(w)-g_2)\,dw-\overline{\int_{z_0}^z \wp(w)\,dw},
\;
\lambda A\,\wp(z)+\overline{A\lambda}\,\overline{\wp(z)}
\Big)
\]
is minimal and proper, has one end of order $N=3$, and total curvature $-8\pi$ [2507.12914]. It specializes to the classical Chen–Gackstatter torus precisely when $|\lambda|=1$, in which case the image lies in a $3$-dimensional Euclidean subspace [2507.12914]. The same source states that these are the only non-holomorphic proper minimal immersions of the square torus with one end and total curvature $-8\pi$.

That Euclidean $\mathbb{R}^4$ family is described as generalizing the Chen–Gackstatter torus in $\mathbb{R}^3$, and the comparison with the Lorentzian theory is instructive. In $\mathbb{R}^4_1$, the explicit deformations preserve the $D_4$ symmetry and retain two self-intersection points while locally embedding the end [1212.6802]. In Euclidean $\mathbb{R}^4$, the square-torus family also recovers the classical $\mathbb{R}^3$ torus when restricted to $|\lambda|=1$, but for $|\lambda|\neq 1$ the immersion is genuinely $4$-dimensional [2507.12914]. A plausible implication is that the Chen–Gackstatter configuration acts as a common organizing center for several adjacent moduli problems: minimal surfaces in $\mathbb{R}^3$, stationary surfaces in $\mathbb{R}^4_1$, and minimal surfaces in $\mathbb{R}^4$.

The Euclidean $\mathbb{R}^4$ paper also introduces braid, link, and writhe at infinity to analyze embeddedness. For the $4$-dimensional Chen–Gackstatter family on the square torus, if $|\lambda|\neq 1$, the end is asymptotic after coordinate change to
\[
z\mapsto (z^3+o(|z|^3),\, z^2+o(|z|^2)),
\]
the knot at infinity is a $(3,2)$-torus knot, and the braid at infinity has writhe $w_\infty=\pm 4$. Since
\[
\frac{1}{2\pi}\int_\Sigma K^N=w_\infty(\Sigma)-2D_\Sigma=0,
\]
the surfaces are not embedded; if only transverse double points occur then their algebraic count is $D_\Sigma=\pm 2$ [2507.12914]. This resonates with the non-embeddedness results and open embeddedness questions already present in the Lorentzian theory.

Taken together, the cited works place the Chen–Gackstatter torus at the center of a broad genus-one finite-total-curvature classification program. The classical square-torus example in $\mathbb{R}^3$ supplies the prototype. In $\mathbb{R}^4_1$, it generates a real $2$-parameter explicit Lorentz family and a local real $4$-parameter deformation family, while large symmetry forces uniqueness back to that explicit model [1212.6802]. In Euclidean $\mathbb{R}^4$, it gives rise to a unique square-torus family of non-holomorphic proper minimal immersions and to further rectangular-torus constructions, while embeddedness remains governed by new invariants at infinity [2507.12914].

Source: https://www.emergentmind.com/topics/chen-gackstetter-torus