Riemannian geometry
Riemannian geometry is the study of smooth manifolds equipped with a metric tensor — a positive-definite inner product on the tangent space at each point that permits the measurement of lengths, angles, volumes, and curvature. Developed by Bernhard Riemann in his 1854 habilitation lecture On the Hypotheses Which Lie at the Foundations of Geometry, it generalizes Euclidean geometry to curved spaces and provides the mathematical foundation for general relativity, though its scope extends far beyond physics.
In Riemannian geometry, the metric gμν replaces the flat-space notion of distance. Given a curve on the manifold, its length is computed by integrating the metric along the curve. The geometry is intrinsic: it can be determined entirely by measurements within the space, without reference to any embedding in a higher-dimensional ambient space. This was Riemann's revolutionary insight — that the geometry of space is a property of space itself, not of how space sits inside something larger.
Curvature and the Fundamental Theorem
The central object of Riemannian geometry is the Riemann curvature tensor Rρσμν, which measures the failure of vectors to return to their original orientation after parallel transport around an infinitesimal loop. The full tensor contains twenty independent components in four dimensions, but these can be contracted to produce the Ricci tensor Rμν and the scalar curvature R — measures of curvature that appear directly in the Einstein field equations.
The fundamental theorem of Riemannian geometry states that on any Riemannian manifold there exists a unique torsion-free, metric-compatible connection — the Levi-Civita connection. This connection defines parallel transport and gives meaning to the idea of a straightest path, or geodesic, on a curved manifold. The theorem is remarkable: it says that the metric alone determines how vectors are carried from point to point, with no additional structure required.
Riemannian vs. Semi-Riemannian Geometry
In general relativity, spacetime is not Riemannian but semi-Riemannian (or pseudo-Riemannian): the metric has signature (−, +, +, +) rather than being positive definite. This permits the distinction between timelike, spacelike, and null directions, and it allows the metric to serve as a causal structure as well as a geometric one. The mathematical apparatus — connections, curvature, geodesics — extends naturally to the semi-Riemannian case, but the physical interpretation is radically different. In Riemannian geometry, geodesics are shortest paths. In semi-Riemannian geometry, timelike geodesics are longest paths (maximizing proper time), a fact that underlies the twin paradox and gravitational time dilation.
Connections to Other Fields
Riemannian geometry has become essential in machine learning, where manifold learning algorithms assume that high-dimensional data lies on or near a low-dimensional Riemannian manifold. The diffusion map and Laplacian eigenmap methods exploit the metric structure of data to find meaningful low-dimensional representations. In information geometry, statistical manifolds are equipped with the Fisher information metric, and the resulting Riemannian structure encodes the geometry of inference.
More recently, the theory of optimal transport — particularly the work of Cédric Villani on synthetic Ricci curvature — has shown that Ricci curvature lower bounds can be defined on arbitrary metric measure spaces without any differentiable structure. This suggests that the core concepts of Riemannian geometry are more robust than their smooth-manifold formulation, and that curvature is fundamentally a property of metric spaces, not just of manifolds with charts and atlases.
Riemannian geometry is often taught as the mathematics of curved spaces, but this description understates its significance. It is the mathematics of measurement itself — the structure that makes it possible to ask, at any point in any space, 'What is the distance to there?' The metric is not a decoration on a manifold; it is the manifold's identity. A smooth manifold without a metric is a bare topological skeleton. The metric is the flesh. To study Riemannian geometry is to study what it means for a space to have shape.