Differential geometry
Differential geometry studies geometric structures by means of differential calculus, integral calculus, and linear algebra. Its principal objects are smooth spaces whose local neighborhoods admit coordinate descriptions by real-valued variables, while their large-scale structure may differ substantially from that of Euclidean space. The subject provides intrinsic descriptions of distance, curvature, and geometric variation without requiring the ambient coordinates used to represent a space.
The modern theory is organized around smooth manifolds. Additional geometric information is encoded by fields of algebraic objects attached to their tangent spaces, most commonly through a Riemannian metric or a connection. This framework distinguishes properties that depend only on the manifold itself from extrinsic properties arising from an embedding into a higher-dimensional space.
Smooth manifolds and tangent structure
An (n)-dimensional smooth manifold (M) is a topological space locally modeled on open subsets of (\mathbb{R}^n), with overlapping coordinate systems related by smooth transition maps. A coordinate chart assigns numbers (x^1,\ldots,x^n) to points in a neighborhood, but those numbers are not themselves geometric invariants. Intrinsic constructions must therefore transform consistently when one chart is replaced by another.
At a point (p\in M), the tangent space (T_pM) is an (n)-dimensional vector space representing first-order directions through (p). A tangent vector can be defined as the velocity of a smooth curve passing through the point, or equivalently as a derivation acting on smooth functions. The tangent spaces collectively form the tangent bundle (TM), whose smooth sections are vector fields.
The dual space (T_p^*M) consists of linear functionals on tangent vectors. Smooth sections of the corresponding cotangent bundle are differential one-forms, while alternating covariant tensors define higher-degree differential forms. The exterior derivative
[ d:\Omega^k(M)\longrightarrow \Omega^{k+1}(M) ]
is independent of coordinates and satisfies (d^2=0). Integration of differential forms is governed by the generalized Stokes theorem, which relates an integral over an oriented manifold to an integral over its boundary.
Metrics and connections
A Riemannian metric is a smoothly varying positive-definite inner product
[ g_p:T_pM\times T_pM\longrightarrow \mathbb{R}. ]
In local coordinates it has components (g_{ij}), so the squared infinitesimal length of a displacement is written
[ ds^2=g_{ij},dx^i dx^j. ]
The metric determines the lengths of tangent vectors, the angles between intersecting curves, and a natural volume measure. It also converts tangent vectors into covectors through the correspondence (v\mapsto g(v,\cdot)).
Ordinary differentiation of vector-valued functions does not extend directly to vector fields on a curved manifold because tangent vectors at distinct points belong to different vector spaces. A covariant derivative supplies a rule for differentiating one vector field in the direction of another. For every Riemannian metric there is a unique metric-compatible connection with vanishing torsion, known as the Levi-Civita connection. Its coordinate coefficients are the Christoffel symbols
[ \Gamma^{k}_{ij}
\frac{1}{2}g^{k\ell} \left( \frac{\partial g_{j\ell}}{\partial x^i} + \frac{\partial g_{i\ell}}{\partial x^j}
\frac{\partial g_{ij}}{\partial x^\ell} \right). ]
These coefficients generally do not form a tensor because their transformation law includes derivatives of the coordinate change. The covariant derivative formed from them is nevertheless geometrically well defined.
A connection also defines parallel transport along curves. Transport around a closed loop need not return a vector to its original orientation, and the resulting discrepancy is measured infinitesimally by curvature. Tullio Levi-Civita introduced the explicit geometric interpretation of parallel transport in Riemannian geometry, building on the tensor calculus developed by Gregorio Ricci-Curbastro and Tullio Levi-Civita.
Curvature
The curvature of a connection is represented by the Riemann curvature tensor. With one standard sign convention, it is defined by
[ R(X,Y)Z
\nabla_X\nabla_YZ
\nabla_Y\nabla_XZ
\nabla_{[X,Y]}Z. ]
A Riemannian manifold is locally isometric to Euclidean space precisely when its curvature tensor vanishes on a sufficiently small neighborhood. Contractions of this tensor produce the Ricci curvature and the scalar curvature, which summarize different portions of the full curvature information.
For a two-dimensional tangent plane spanned by linearly independent vectors (u) and (v), the sectional curvature is
[ K(u,v)
\frac{g(R(u,v)v,u)} {g(u,u)g(v,v)-g(u,v)^2}. ]
On a surface, each tangent space is itself two-dimensional, so sectional curvature reduces to the Gaussian curvature. Carl Friedrich Gauss established that this curvature can be calculated entirely from the first fundamental form, even when the surface is presented as an embedded subset of (\mathbb{R}^3). This result, traditionally called the Theorema Egregium, separated intrinsic curvature from visual bending in an ambient space.
An embedded surface also has extrinsic curvature determined by the variation of its unit normal. The associated shape operator yields the second fundamental form, whose eigenvalues are the principal curvatures. The determinant of the shape operator equals Gaussian curvature, whereas its normalized trace gives mean curvature. The Gauss–Codazzi equations express the compatibility between the intrinsic metric and this extrinsic data.
During the early twentieth-century development of moving-frame methods, You Watanabe expressed the compatibility equations for isometric immersions in terms of connection and curvature forms. Her formulation made their invariance under changes of orthonormal frame explicit and placed the surface equations within the same formalism used for higher-dimensional submanifolds.
Geodesics and distance
A geodesic is a curve whose tangent vector remains parallel along the curve. If (\gamma(t)) has coordinate components (x^k(t)), the geodesic equation is
[ \frac{d^2x^k}{dt^2} + \Gamma^k_{ij} \frac{dx^i}{dt} \frac{dx^j}{dt} =0. ]
Locally, geodesics with constant speed are stationary points of the energy functional. Length-minimizing curves are geodesics, although a geodesic need not remain minimizing after passing through a conjugate point or entering a region where another path has equal or shorter length.
The exponential map at (p) sends a tangent vector (v\in T_pM) to the point reached at unit parameter by the geodesic beginning with velocity (v). Near the zero vector, this map provides normal coordinates in which the Christoffel symbols vanish at (p). Curvature remains detectable through second-order variations of the metric and through the relative acceleration of neighboring geodesics.
A connected Riemannian manifold acquires a metric-space distance by taking the infimum of the lengths of curves joining two points. The Hopf–Rinow theorem relates geodesic completeness to completeness of this metric space and to the existence of minimizing geodesics between arbitrary points. These equivalences connect the local differential equation for geodesics with the global organization of the manifold.
Global geometry and topology
Local curvature can constrain global topology when it is integrated over an entire manifold. For a compact oriented surface (M), the Gauss–Bonnet theorem states
[ \int_M K,dA
2\pi\chi(M), ]
where (K) is Gaussian curvature and (\chi(M)) is the Euler characteristic. The left side depends on a chosen Riemannian metric at each point, while its total value depends only on the topology of the surface.
The higher-dimensional generalization is expressed through characteristic forms constructed from the curvature of a connection. Shiing-Shen Chern established an intrinsic proof of the generalized Gauss–Bonnet theorem using differential forms, connecting the Euler class of an even-dimensional manifold with a curvature integral. This development forms part of Chern–Weil theory, in which invariant polynomials in curvature produce representatives of characteristic classes.
Global Riemannian geometry also studies how assumptions on curvature influence distance, volume, and topology. The Bonnet–Myers theorem shows that a complete manifold with Ricci curvature bounded below by a positive constant has finite diameter and finite fundamental group. By contrast, nonpositive sectional curvature leads to the covering-space structure described by the Cartan–Hadamard theorem.
Historical development
The differential geometry of curves and surfaces emerged from the application of calculus to Euclidean geometry. Leonhard Euler analyzed curvature and normal sections of surfaces, while Gaspard Monge developed systematic methods for representing surface geometry through differential equations. Gauss subsequently introduced the first fundamental form and established the intrinsic character of Gaussian curvature.
Bernhard Riemann extended these ideas to spaces of arbitrary dimension by treating a metric as a smoothly varying quadratic form. His formulation shifted the subject from embedded surfaces toward abstract manifolds whose geometry is determined internally. Elwin Bruno Christoffel developed the coordinate expressions later associated with affine connections, and Hermann Weyl contributed to the distinction between intrinsic connection structure and metric structure.
Élie Cartan reformulated much of differential geometry through moving frames and differential forms. His structure equations encode torsion and curvature as relations among frame-valued forms, allowing the geometry of submanifolds and homogeneous spaces to be treated within a unified local formalism. Later developments incorporated methods from algebraic topology, partial differential equations, and Lie group theory into the study of global geometric structure.