Geometry
Differential geometry
Uses calculus to study curvature, geodesics and other local properties of curves and surfaces.
IntuitionIntuition: measuring curvature with calculus
A flat sheet of paper can be rolled into a cylinder without stretching, but it can never be wrapped smoothly onto a ball without tearing or wrinkling it. Differential geometry explains this difference with a single number attached to every point of a curve or surface: its curvature. By applying calculus — derivatives and integrals — to a parametrized curve or surface, we can measure exactly how much it bends, twists, and departs from being flat, entirely from measurements you could make while standing on the surface itself.
SchoolFrom plane curves to the Frenet–Serret frame
Definition: Curvature of a curve
For a smooth curve in space, the curvature measures how fast the unit tangent vector turns per unit of arc length. A straight line has everywhere; a circle of radius has constant curvature .
Formula lets you compute curvature directly from any parametrization, without first reparametrizing by arc length. At every point of the curve we can also build a moving frame of three orthonormal vectors: the unit tangent , the principal normal (pointing toward the center of the curve's bend), and the binormal . How this frame rotates as it travels along the curve is recorded by the curvature and the torsion , which measures how far the curve twists out of its osculating plane.
UndergraduateThe two fundamental forms and Gaussian curvature
Definition: First fundamental form
For a parametrized surface , the first fundamental form , with , , , tells you how to measure lengths, angles and areas using only the coordinates — it is entirely intrinsic, meaning an ant living on the surface could measure it without ever leaving the surface.
Definition: Second fundamental form
The second fundamental form , built from , , where is the unit normal, measures how fast the surface pulls away from its own tangent plane — it is extrinsic, since it depends on how the surface sits in space.
Combining the two forms gives the single most important local invariant of a surface, the Gaussian curvature . When the surface curves the same way in every direction (like a sphere); when it curves oppositely in different directions (like a saddle); when the surface is developable and can be unrolled flat (like a cylinder or cone).
| Surface | Sign of | Local shape |
|---|---|---|
| Sphere of radius | Bowl-shaped in every direction; the surface curves away from the tangent plane on the same side everywhere. | |
| Plane or cylinder | Developable: can be flattened without stretching. | |
| Saddle or hyperboloid | Curves upward in one direction and downward in the perpendicular direction; the surface crosses its tangent plane. |
UndergraduateKey theorems: Theorema Egregium and Gauss–Bonnet
The Gaussian curvature of a surface can be computed entirely from the first fundamental form and its derivatives — it does not depend on the second fundamental form or on how the surface is embedded in space. Consequently, is preserved by any isometry (a map that preserves the first fundamental form, hence all lengths and angles measured on the surface).
Why is it true?
This is surprising because was originally defined using the second fundamental form, i.e. using how the surface bends within the ambient space. Gauss's theorem says curvature is secretly intrinsic: a two-dimensional being confined to the surface, unable to see the surrounding space, could still compute by measuring lengths and angles alone.
Proof
Work in coordinates where at the point of interest (an orthogonal parametrization, which always exists locally). Differentiating the identities and and using , expresses in terms of the second derivatives projected onto .
Because form a basis of space at each point, every second derivative such as can be written as a combination of , and , with the tangential coefficients being the Christoffel symbols — functions built only from and their derivatives with respect to and , obtained by solving the linear system that comes from differentiating .
Substituting these expressions into and comparing with from the normal components, then simplifying with the compatibility (Gauss) equations, produces Brioschi's formula: expressed purely as a rational function of and their first and second partial derivatives with respect to and , with no reference to left in the final expression.
Since an isometry between two surfaces is, by definition, a map that carries the first fundamental form of one surface exactly onto the first fundamental form of the other (the same as functions of the parameters), and Brioschi's formula computes from alone, the two surfaces must have equal Gaussian curvature at corresponding points. This proves the theorem.
For a compact, orientable surface without boundary, , where is the Euler characteristic, for any triangulation of (also equal to for a surface of genus ).
Why is it true?
The theorem links a purely local, geometric quantity (curvature, which can change from point to point) to a purely global, topological quantity (the Euler characteristic, which only depends on how the surface is connected, not on its shape). No matter how you bend, stretch or dent a surface without tearing it, the total curvature stays the same.
Proof
Triangulate using geodesic triangles (triangles whose sides are shortest paths on the surface), obtaining vertices, edges and faces, related by .
The local Gauss–Bonnet formula for a single geodesic triangle with interior angles states : the amount the angle sum exceeds the Euclidean value equals exactly the integral of curvature over that triangle, a fact obtained by applying Stokes' theorem to the rotation of a tangent vector parallel-transported around the triangle's boundary.
Summing over all triangles gives : the triangles meeting at each of the vertices sweep out exactly one full turn there, so all their angles together sum to , while each of the triangles contributes a .
Each triangle has edges and each edge is shared by exactly triangles, so . Substituting into gives , which matches the previous step exactly. Hence , proving the theorem.
UndergraduateReal-World Applications and Worked Examples
Differential geometry is the mathematical backbone of any technology that must represent, measure or move along curved shapes. Cartography lives with the Theorema Egregium every day: because the sphere has and the plane has , no flat map of the Earth can show all distances and angles correctly at once, forcing mapmakers to choose which distortion — area, angle, or distance — to accept. Computer graphics and CAD systems compute Gaussian and mean curvature at every vertex of a 3D mesh to decide where to add detail, smooth a surface, or detect a manufacturing defect. GPS navigation and orbital mechanics compute geodesics — the straightest possible paths — on the curved reference ellipsoid of the Earth rather than on a flat map. Architects use surfaces of negative Gaussian curvature, such as hyperbolic paraboloid roofs, because they are doubly ruled and can be built from straight beams while still carrying loads efficiently.
Example: Curvature and torsion of a helix
A wire is bent into the helix . Find its curvature and torsion .
Solution
Differentiate the position vector: , , and . Note that , a constant, independent of .
Compute the cross product , which has length . By the curvature formula, .
For torsion, use . The dot product is constant, and , so .
Both and are constant along the whole curve, which is exactly why a helix looks the same at every point: it is (up to a rigid motion) the unique curve with constant nonzero curvature and torsion.
Example: Gaussian curvature of a sphere and a Gauss–Bonnet check
Using the parametrization of a sphere of radius , verify that everywhere, then check the Gauss–Bonnet theorem for the whole sphere.
Solution
Compute the tangent vectors and , giving first fundamental form coefficients , , .
The unit normal is (pointing inward), and computing the second derivatives against gives second fundamental form coefficients , , .
By the Gaussian curvature formula, , confirming that every point of a sphere of radius has the same curvature.
For Gauss–Bonnet, the area element is , so . Since a sphere has Euler characteristic , the theorem predicts , which matches exactly.
At a point on a surface, the first and second fundamental form coefficients are and . What is the Gaussian curvature at that point?
Why can no single flat map of the Earth preserve every distance and angle at once?
A sphere has Euler characteristic . According to the Gauss–Bonnet theorem, what is ?
In the Frenet–Serret frame of a space curve, which vector points toward the center of the curve's instantaneous bend, perpendicular to the direction of motion?
References
- Manfredo P. do Carmo (2016). Differential Geometry of Curves and Surfaces
- Kristopher Tapp (2016). Differential Geometry of Curves and Surfaces