Loading field map
How can one study the geometry of a surface that is too rough to be described by the smooth tools of differential geometry? This question lies at the heart of geometric measure theory (GMT), a field that emerged from a concrete variational problem: the Plateau problem, which asks for the surface of least area spanning a given boundary curve. Classical calculus of variations could handle smooth surfaces, but the minimal surface for a given boundary might develop corners, edges, or even fractal-like singularities. To make progress, mathematicians needed a new language—one that could assign a precise size to irregular sets, perform calculus on them, and ultimately find surfaces that minimize area even when they are not smooth. The four frameworks that built this language are Hausdorff measure and dimension, rectifiable sets and the area/coarea formulas, currents, and varifolds.
The first step was to find a way to measure the size of a set that might not be a smooth manifold. Felix Hausdorff, in 1918, introduced a family of measures that generalize the familiar notions of length, area, and volume to sets of any dimension—including fractional dimensions. The idea is simple: cover a set by small balls, sum the diameters raised to a power s, and take the limit as the balls shrink. The resulting s-dimensional Hausdorff measure, denoted ℋ^s, assigns a number to every subset of Euclidean space. For a smooth curve, ℋ^1 gives its length; for a smooth surface, ℋ^2 gives its area. But ℋ^s can also be nonzero for non-integer s, capturing the size of fractals. The Hausdorff dimension of a set is the critical exponent where the measure jumps from infinity to zero. This framework provided the first rigorous way to talk about the size of irregular sets, but it did not yet supply the tools needed to do calculus on them. Hausdorff measure alone cannot differentiate functions or integrate over a surface in a way that respects its geometry.
The second framework, developed in the mid-20th century by Besicovitch, Federer, and others, identified a class of sets that are irregular enough to include singularities but still regular enough to support calculus. A set is called rectifiable if it can be covered, up to a set of measure zero, by a countable union of Lipschitz images of Euclidean space. In practical terms, a rectifiable set looks like a smooth manifold almost everywhere: at ℋ^k-almost every point, it has an approximate tangent plane. This property allows one to define integrals over the set and to differentiate functions with respect to the set's geometry. The two central computational tools are the area formula and the coarea formula. The area formula generalizes the change-of-variables formula from multivariable calculus: it tells how the ℋ^k measure of a set changes under a Lipschitz map. The coarea formula is its dual, relating integrals over the domain to integrals over the level sets of a map. Together, these formulas made it possible to perform integration on rectifiable sets and to relate geometric quantities to analytic ones. Rectifiable sets provided the right notion of a "surface" for GMT, but they still lacked a way to handle boundaries and to solve variational problems like the Plateau problem in full generality.
The third framework, introduced by de Rham and then developed by Federer and Fleming in the 1960s, solved the Plateau problem by embedding surfaces into a linear space. A current is a continuous linear functional on the space of smooth differential forms with compact support. This definition may seem abstract, but it directly generalizes the idea of integrating a differential form over a surface: every oriented rectifiable set with integer multiplicity defines a current by integration. The key innovation is that currents form a chain complex with a boundary operator ∂, defined by duality with the exterior derivative: ∂T(ω) = T(dω). This boundary operator makes currents into a homology theory, and it allows one to formulate the Plateau problem as a search for a current of minimal mass (the total variation of the current) with a given boundary. Federer and Fleming proved that a solution always exists in the class of integral currents—those with integer multiplicity and rectifiable support—using a compactness theorem that relies on the weak topology on currents. This was a landmark result: it showed that a minimal surface exists for any rectifiable boundary, even if the surface must develop singularities. Currents are inherently oriented, because differential forms are oriented objects. This orientation is a strength when studying oriented minimal surfaces, but it becomes a limitation when the physical problem involves unoriented surfaces, such as soap films that can collapse onto themselves.
The fourth framework, varifolds, was introduced by Almgren and later developed by Allard in the 1970s to handle unoriented surfaces and geometric evolution problems. A varifold is a Radon measure on the product of Euclidean space with the Grassmannian of unoriented k-planes. Intuitively, a varifold records, for each point, not just the presence of a surface but also the distribution of tangent planes at that point. This extra information allows varifolds to represent surfaces that are not rectifiable in the usual sense, and it provides a natural setting for studying mean curvature flow and other geometric evolution equations. Unlike currents, varifolds do not have a boundary operator, so they are less suited to the Plateau problem. But they excel where orientation is irrelevant or where surfaces can change topology. Allard's compactness theorem for varifolds, analogous to Federer–Fleming for currents, guarantees that sequences of varifolds with bounded mass and bounded first variation have convergent subsequences. This compactness is the foundation for proving existence of solutions to geometric flows. Varifolds also support a notion of generalized mean curvature, which can be defined even when the surface is not smooth, making them a powerful tool for regularity theory.
Today, currents and varifolds coexist as complementary frameworks, each with its own domain of strength. Currents remain the primary tool for problems that involve orientation and boundaries: they are the natural setting for the classical Plateau problem, for calibrated geometry, and for many questions in the calculus of variations where the surface must have a prescribed boundary. Varifolds dominate in geometric flows, such as mean curvature flow, where surfaces evolve without regard to orientation and may develop singularities. The two frameworks agree on the core idea that a surface can be represented as a measure-theoretic object, but they disagree on what data that object should carry: currents store orientation and a boundary operator, while varifolds store unoriented tangent plane distributions. This disagreement is not a conflict but a division of labor. Both frameworks rely on the earlier foundations of Hausdorff measure and rectifiable sets: currents and varifolds are defined over rectifiable sets, and their mass is measured by Hausdorff measure. The area and coarea formulas, originally developed for rectifiable sets, are essential tools for estimating the mass of currents and varifolds and for proving regularity results. The leading open questions in GMT today often involve understanding the singularities that arise in minimal surfaces and geometric flows, and both currents and varifolds contribute to this effort—currents through the regularity theory of integral currents (Almgren's big regularity paper) and varifolds through Allard's regularity theorem. The field remains deeply connected to the calculus of variations, which supplies the variational principles, and to partial differential equations, which describe the behavior of minimal surfaces and mean curvature flow. A student entering GMT today will find a mature but still active discipline, where the four frameworks form a layered toolkit: Hausdorff measure provides the ruler, rectifiable sets provide the canvas, currents provide the oriented brush, and varifolds provide the unoriented one.