--- title: Least Total Curvature Solutions url: https://www.emergentmind.com/topics/least-total-curvature-solutions type: topic --- # Least Total Curvature Solutions Searching arXiv for relevant papers on "least total curvature solutions" and adjacent minimal total curvature literature. Least total curvature solutions are extremal objects selected either by minimizing a curvature functional or by attaining the equality case in a sharp lower-bound theorem for curvature. In current research usage, the term spans several distinct but structurally related settings: steady incompressible Euler flows in a strip obtained from a total-curvature variational principle, compact and complete surfaces whose total absolute curvature is smallest under topological or ambient constraints, and discrete planar graphs whose positive total curvature exhibits a first nonzero gap [2507.11837]. ## 1. Extremality notions and curvature functionals The phrase is not attached to a single universal curvature quantity. In the Euclidean surface setting of Han and Khuri, the total absolute curvature of a compact orientable surface \(S\subset \mathbb{R}^3\) is \[ T(S)=\int_S |K|\,dA, \] and the class \(T\) of “tight” surfaces is defined by \[ \int_{S^+}K\,dA=4\pi, \] equivalently, \[ T(S)=2\pi(4-\chi(S)). \] In this formulation, “minimal total absolute curvature” means attainment of the Chern–Lashof lower bound [1104.0474]. For closed immersed surfaces \(\Sigma\subset M^3\) in a Cartan–Hadamard manifold, the curvature quantity is the total absolute extrinsic curvature \[ T(\Sigma):=\int_\Sigma |GK|\,dA, \] where \(GK\) is the Gauss–Kronecker curvature. The least-total-curvature theorem states \(T(\Sigma)\ge 4\pi\), with equality characterized by convexity and ambient flatness on the enclosed body [2604.25024]. In the equiaffine setting, the relevant invariant is Koike’s total absolute curvature \[ \tau_S(f)=\frac{1}{\mathrm{vol}(\mathbb S^{n+r-1})}\int_B |G(\eta)|\,\omega_B, \] which is also the average number of critical points of height functions. Here the least value is \(\tau_S(f)=2\), and the equality case is equivalent to the image being a convex hypersurface in an affine \((n+1)\)-plane [2604.00384]. The strip-Euler problem uses a different functional. For a stream function \(\psi\), the total curvature of the flow is \[ J[\psi] = \int_{\Omega_{\infty}} \frac{\bigl|\nabla^{\perp}\psi\cdot D^{2}\psi\bigr|^{2}}{|\nabla\psi|^{2}} \,dx, \] interpreted as \(0\) on stagnation. In that setting, “least total curvature solutions” are minimizers of \(J\) under boundary and asymptotic constraints, and the Euler–Lagrange condition is the steady Euler vorticity transport equation \(v\cdot\nabla\omega=0\) [2507.11837]. A discrete analogue appears for planar graphs with nonnegative combinatorial curvature. The vertex curvature is \[ k(x)=1-\frac{\deg(x)}2+\sum_{\sigma\ni x}\frac1{\deg(\sigma)}, \] and the total curvature is \(K(G)=\sum_{x\in V}k(x)\). The least positive value is not \(0\) but \(1/12\), producing a gap phenomenon rather than a continuous minimization principle [1709.05309]. ## 2. Gaussian-curvature minimization with prescribed contour A direct variational surface problem is formulated in “Minimal Gaussian Curvature Surface” [2101.06673]. The objective is to find surfaces in \(\mathbb R^3\) that are as close as possible to being flat, span a given contour, and have that contour as a geodesic on the sought surface. The functional being minimized is the total Gaussian curvature squared. The paper states that, after a change of coordinates, the curvature of the optimal surface is controlled by a PDE reducible to the biharmonic equation with an easy-to-define Dirichlet boundary condition and Neumann boundary condition zero. It then states a system of PDEs for the function whose graph is the optimal surface [2101.06673]. This formulation differs from equality-case theories such as Chern–Lashof or Cartan–Hadamard rigidity. Here the extremal object is selected by minimizing a curvature-square energy under a boundary-geodesic constraint, rather than by classifying surfaces that saturate a topological lower bound. A plausible implication is that the topic “least total curvature” naturally includes both integral-square functionals and absolute-curvature minimization, depending on the geometric constraint class. ## 3. Steady Euler flows in a strip The most literal use of the expression “least total curvature solutions” in recent PDE literature concerns the steady incompressible Euler system in the infinite strip \[ \Omega_\infty=\{(x_1,x_2)\in\mathbb R^2:00\), together with appropriate decay as \(x_1\to\pm\infty\). The least-total-curvature problem is \[ J_{\min}(L)=\inf_{\psi\in\mathcal A}J[\psi]. \] A formal first variation yields \[ \delta J[\psi](\eta)=0 \quad\Longleftrightarrow\quad v\cdot\nabla\omega=0, \] so any minimizer is a steady Euler flow [2507.11837]. The existence theory proceeds through an auxiliary semilinear Dirichlet problem \[ \begin{cases} -\Delta u=f(u), & x\in\Omega_\infty,\\ u(x_1,0)=0,\quad u(x_1,1)=c, & x_1\in\mathbb R, \end{cases} \] with \(v=(-u_{x_2},u_{x_1})\). The nonlinearity is chosen so that the one-dimensional ODE \[ -\psi''(x_2)=f(\psi(x_2)),\qquad \psi(0)=0,\ \psi(1)=c, \] has exactly two global-energy minimizers \(\underline\psi<\overline\psi\), with no others in between. Minimization on truncated cylinders \(\Omega_n=(-n,n)\times(0,1)\), monotonicity \(u_{n,x_1}>0\), an alignment argument, compactness, and a Hamiltonian identity then produce a heteroclinic limit \[ u(x_1,x_2)\to \underline\psi(x_2)\quad(x_1\to-\infty),\qquad u(x_1,x_2)\to \overline\psi(x_2)\quad(x_1\to+\infty). \] Theorem A asserts the existence of \(f\in C^\infty\), \(c>0\), and a bounded smooth solution of the semilinear problem with \(u_{x_1}>0\), where \(\underline\psi\) is strictly increasing in \(x_2\) and \(\overline\psi\) has exactly one change of monotonicity; the associated \(v=(-u_{x_2},u_{x_1})\) is a least total curvature Euler flow of a new type. The same scheme also yields a strictly increasing positive solution of \(-\Delta u=f(u)\) in the strip with a non-convex superlevel set, giving a negative answer to a generalized problem of Hamel–Nadirashvili–Sire in the unbounded strip [2507.11837]. ## 4. Tight compact surfaces and rigidity in \(\mathbb R^3\) For compact orientable immersed surfaces in \(\mathbb R^3\), minimal total absolute curvature is encoded by the tight class \(T\). Han and Khuri assume \(S\in T\), define \(S^-=\{p:K(p)<0\}\), and study rigidity of isometric immersions when the Gaussian curvature changes sign in a controlled way. Their first theorem assumes \(S\) is \(C^{a+2}\), \(K\) vanishes to finite odd order along \(\partial S^-\), and every closed asymptotic curve \(\gamma\subset S^-\) satisfies \[ \oint_\gamma k_g\,k_n\,|K|^{-1/2}\,ds\neq 0. \] Under these hypotheses, any other \(C^{a+2}\) isometric immersion differs from \(S\) by a Euclidean motion. The second theorem assumes \(S\) is \(C^4\), \(H\neq 0\) along \(\partial S^-\), \(K\) changes sign monotonically across \(\partial S^-\), and imposes the same nondegeneracy integral along each closed asymptotic curve; rigidity again follows [1104.0474]. The geometry of the proof decomposes the surface into a convex part and a negative-curvature region. Kuiper’s theorem and Pogorelov rigidity control the closure of the positive region \(S'=\operatorname{cl}\{K>0\}\), whose boundary curves lie in planes. Under the finite-order vanishing and nondegeneracy assumptions, each connected component of \(S^-\) is topologically an annulus. The Gauss–Codazzi system for two isometric immersions is then rewritten as a weakly hyperbolic \(2\times 2\) system for the differences in second fundamental form coefficients, with local uniqueness near \(\partial S^-\) and near closed asymptotic curves, followed by propagation along asymptotic characteristics [1104.0474]. The equality case is also topologically restrictive. Since \(\chi(S)\ge 0\) by Chern–Lashof, the only orientable tight surfaces are \(S^2\) and \(T^2\). Spherical tight surfaces are exactly the convex ones, and the standard ring torus of revolution is the canonical genus-\(1\) example. The paper’s conclusion is that least-total-absolute-curvature solutions in genus \(0\) or \(1\) exhibit rigidity phenomena analogous to those of strictly convex surfaces, provided the sign change of \(K\) and the asymptotic foliation satisfy the stated finite-order and nondegeneracy conditions [1104.0474]. ## 5. Least admissible total curvature for complete minimal and stationary surfaces A different use of least-curvature language arises for complete minimal surfaces with fixed end structure. For a complete conformal minimal immersion \(f:M\to\mathbb R^3\) of finite total curvature, the Gauss map \(g\) is meromorphic on the compactification, and the total absolute curvature is \[ \tau(M)=\int_M |K|\,dA = 4\pi\,\deg(g). \] If the genus is \(\gamma\) and the number of ends is \(n\), then \(\deg(g)\ge \gamma+n-1\). For \(n=2\), this yields \(\deg(g)\ge \gamma+1\), with equality if and only if each end is embedded; in that case one recovers the plane or the catenoid. For non-catenoidal two-ended surfaces one has \[ \deg(g)\ge \gamma+2. \] The paper constructs families of complete, non-catenoidal, two-ended minimal surfaces realizing this lower bound for genus \(\gamma=1\) and for even \(\gamma\), so that \[ \tau(M)=4\pi(\gamma+2). \] It further proves a uniqueness theorem from symmetry: any such surface with \(4(\gamma+1)\) symmetries must coincide, up to rigid motion, with the explicit examples [1509.06021]. In Lorentzian surface theory, an analogous classification holds for complete spacelike stationary surfaces in \(\mathbb R^4_1\). Using the Weierstrass representation with meromorphic functions \(\phi,\psi\) and holomorphic \(1\)-form \(\mathrm dh\), the total Gaussian curvature of an algebraic surface is quantized. When \[ -\int_M K\,dM=4\pi, \] the generalized Jorge–Meeks formula forces \[ r+\sum \tilde d_j+2g=4. \] A case-by-case analysis shows that the only complete immersed algebraic examples are the generalized Enneper surfaces, with genus \(0\) and one regular end of multiplicity \(3\), and the generalized catenoids, with genus \(0\) and two regular ends each of multiplicity \(1\). For non-orientable algebraic stationary surfaces with oriented double cover of genus \(g\), a finer analysis yields the sharp lower estimate \[ -\int_M K\,dM\ge 2\pi(g+3). \] No non-algebraic example with \(-\int K\,dM=4\pi\) is known, and the conjecture is that none exists [1210.8254]. These two theories share a common pattern: the least admissible curvature is not absolute in the ambient category, but relative to topological and asymptotic constraints. In \(\mathbb R^3\), the catenoid occupies the embedded equality case, while the non-catenoidal problem begins at the next quantized level. In \(\mathbb R^4_1\), the oriented algebraic equality case at \(4\pi\) is exhausted by the generalized Enneper and generalized catenoid families. This suggests that symmetry, end multiplicity, and period conditions serve as the principal rigidity mechanisms in finite-total-curvature classification. ## 6. Equality theorems beyond Euclidean surface theory In Cartan–Hadamard \(3\)-manifolds, Ghomi, Hoisington, Raffaelli, and Stavroulakis prove a direct analogue of the Euclidean Chern–Lashof theorem. If \(\Sigma\subset M^3\) is a closed \(C^\infty\) immersed surface, then \[ T(\Sigma)\ge 4\pi. \] Equality \(T(\Sigma)=4\pi\) occurs if and only if \(\Sigma\) bounds a compact convex body \(C\subset M\) on which the ambient sectional curvature vanishes identically, so \(C\) is isometric to a Euclidean convex body. The proof reduces to the convex hull boundary \(\Gamma_0=\partial\operatorname{conv}(\Gamma)\), uses Gauss–Bonnet on \(\Gamma_0\), derives trivial holonomy from flatness along \(\Gamma_0\), constructs a \(C^1\) isometric immersion \(f:\Gamma_0\to\mathbb R^3\) preserving total curvature of curves, proves convexity of \(f(\Gamma_0)\) via Pogorelov’s bounded-extrinsic-curvature theory, and extends \(f\) to the whole convex hull using a Schur-type comparison theorem and a Kirszbraun–Lang–Schroeder argument. The result generalizes the Euclidean equality case and solves a problem posed by Gromov in 1985 [2604.25024]. In equiaffine differential geometry, Koike’s total absolute curvature has a parallel equality theory. For an equiaffine immersion \(f:(M,\nabla,\theta)\to(\mathbb A^{n+r},\tilde\nabla,\omega)\), the total absolute curvature is \[ \tau_S(f) = \frac1{\mathrm{vol}(\mathbb S^{n+r-1})} \int_B |G(\eta)|\,\omega_B = \frac1{\mathrm{vol}(\mathbb S^{n+r-1})} \int_{S\setminus \nu(\mathrm{Crit}\,\nu)} \#\{\text{critical points of }h_\phi\}\,\omega_S. \] Koike’s inequality gives \[ \tau_S(f)\ge \sum_{k=0}^n b_k(M;\mathbb F), \] and if \(\tau_S(f)<3\) then \(M\approx S^n\). The equality-case theorem states that \[ \tau_S(f)=2 \] if and only if the image \(f(M)\) is a convex hypersurface lying in some affine \((n+1)\)-plane. The proof first reduces codimension by showing that otherwise one could produce a height function with at least three critical points; after iterating this reduction, the hypersurface case is handled by proving that \(\tau_S(f)=2\) is equivalent to every tangent hyperplane being supporting [2604.00384]. Both results recast least total curvature as a rigidity statement. In Cartan–Hadamard geometry, equality singles out convex spheres in flat regions of a nonpositively curved ambient space. In the equiaffine category, equality singles out convex hypersurfaces in codimension one. A plausible implication is that convexity is the dominant geometric signature of least-total-curvature configurations whenever the curvature functional is interpreted as an averaged or absolute extrinsic measure. ## 7. Discrete first-gap phenomena for total curvature For infinite planar graphs with nonnegative combinatorial curvature, least positive total curvature appears as a discrete gap problem rather than a continuous variational one. A semiplanar graph \(G=(V,E,F)\) gives rise to a polygonal surface \(S(G)\) by replacing each face of degree \(d\) with a regular Euclidean \(d\)-gon of side length \(1\) and gluing along corresponding edges. The combinatorial curvature at a vertex \(x\) is \[ k(x)=1-\frac{\deg(x)}2+\sum_{\sigma\ni x}\frac1{\deg(\sigma)}, \] equivalently \(k(x)=(2\pi-\theta_x)/2\pi\), and the total curvature is \[ K(G)=\sum_{x\in V}k(x). \] A discrete Cohn–Vossen theorem gives \(0\le K(G)\le 1\) for infinite planar graphs with nonnegative curvature [1709.05309]. The main theorem determines the first positive gap: \[ T_1:=\inf\{\,K(G)>0:\ G\text{ infinite planar},\ k(x)\ge 0\}=\frac1{12}. \] Moreover, equality \(K(G)=1/12\) occurs if and only if the polygonal surface \(S(G)\) is isometric either to a Euclidean cone with apex angle \[ \theta=2\arcsin\Bigl(\tfrac1{12}\Bigr), \] in which case there is exactly one vertex of positive curvature \(1/12\), or to a frustum over a hendecagon base, in which case there are exactly \(11\) boundary vertices each of curvature \(1/132\), so that \(11\cdot(1/132)=1/12\) [1709.05309]. The proof combines reduction to faces of degree at most \(42\), a complete analysis of vertex patterns with \(k(x)>0\), local clustering arguments showing that any vertex with \(0