1 Definition
1.1 Basic formulation
Let \(f:D\to \mathbb{R}\) be a function defined on a domain \(D\). For a real number \(c\), the level set of \(f\) at height \(c\) is the set \[ \{x\in D: f(x)=c\}. \] It collects exactly those points whose function value equals the prescribed constant. The same construction applies when \(f\) maps into other ordered sets (with appropriate modifications), but the most common setting is real-valued functions.
1.2 Terminology and notation
1.2.1 Preimage interpretation
The level set can be viewed as a preimage: \[ f^{-1}(\{c\})=\{x\in D: f(x)=c\}. \] This viewpoint emphasizes that level sets depend on the inverse image structure of \(f\), and it extends naturally to other sets than singletons, such as intervals and rays.
1.2.2 Zero set as a special case
The zero set of \(f\) is the level set corresponding to \(c=0\): \[ \{x\in D: f(x)=0\}. \] Zero sets are especially prominent because many problems reduce to solving equations \(f(x)=0\). In geometry, they often describe implicit curves or surfaces.
1.3 Examples
- Univariate case. If \(f:\mathbb{R}\to\mathbb{R}\), then the level set at \(c\) is a subset of the real line consisting of all solutions to \(f(x)=c\). For polynomial functions, these are real roots of \(f(x)-c\).
- Quadratic form. For \(f(x,y)=x^2+y^2\) on \(\mathbb{R}^2\), the level set \(f=c\) is the circle \(x^2+y^2=c\) (when \(c\ge 0\)), illustrating how a single formula generates families of geometric objects.
- Linear function. If \(f(x,y)=ax+by\), then \(f=c\) is the line \(ax+by=c\). This shows that level sets can form affine subspaces, producing contour plots used in applications.
2 Types of level sets
2.1 Regular level sets
A level set \(f^{-1}(c)\) is called regular (informally) when \(c\) behaves well with respect to \(f\). In differential settings, a common definition uses the notion of regular values: if \(\nabla f\neq 0\) at every point of \(f^{-1}(c)\) (for a smooth \(f\)), then the level set is typically a smooth manifold of codimension \(1\). Such sets vary smoothly under perturbations of \(c\).
2.2 Singular level sets
A singular level set occurs when the level set contains points where regularity fails, such as locations where the gradient (or differential) vanishes. In that case, the level set may have corners, self-intersections, change of topology, or other non-smooth features. Singular behavior is central in catastrophe theory and in the study of critical points of scalar functions.
2.3 Sublevel sets and superlevel sets
Inequality versions of the level set are important in analysis and optimization. The sublevel set and superlevel set associated to \(c\) are \[ \{x\in D: f(x)\le c\}, \qquad \{x\in D: f(x)\ge c\}. \] These sets describe regions where the function lies below or above a threshold, producing “filled-in” contour regions rather than only the boundary where equality holds. Sublevel sets are particularly important for feasibility regions and for convexity-related arguments.
3 Geometric interpretation
3.1 Level curves
When \(f:\mathbb{R}^2\to\mathbb{R}\), each level set \(f=c\) is a level curve in the plane. Under sufficient smoothness and regularity, these curves are one-dimensional manifolds (e.g., smooth arcs or closed loops). In visualization, level curves form the contour lines of a scalar field.
3.2 Level surfaces
For \(f:\mathbb{R}^3\to\mathbb{R}\), the level set \(f=c\) is a level surface. With \(\nabla f\neq 0\) on the set, it locally resembles a smooth surface embedded in \(\mathbb{R}^3\). Such implicit surfaces include spheres, cylinders, and more general shapes defined without explicit parametrizations.
3.3 Higher-dimensional level sets
In \(\mathbb{R}^n\), the set \(f=c\) for a smooth scalar field is generically an \((n-1)\)-dimensional object (when regular). More generally, for maps \(F:\mathbb{R}^n\to\mathbb{R}^k\), the analogous “preimage of a value” construction produces sets of dimension roughly \(n-k\) under appropriate transversality assumptions. This links level sets to the broader geometry of preimages.
4 Properties
4.1 Nonemptiness and emptiness
A level set can be empty if the value \(c\) is not attained by \(f\) on \(D\). Conversely, it is nonempty whenever there exists at least one point \(x\in D\) with \(f(x)=c\). For continuous functions on compact domains, the attainable range is closed, which often makes the transition between emptiness and nonemptiness depend on the extrema of \(f\).
4.2 Connectedness
Level sets need not be connected. For instance, in the plane, a function like \(f(x,y)=(x^2-1)^2+y^2\) produces level curves that can split into multiple components depending on \(c\). Connectedness is influenced by the geometry of \(f\), the presence of critical points, and constraints on the domain.
4.3 Smoothness and regularity
4.3.1 Implicit function theorem
If \(f\) is continuously differentiable and \(\nabla f(x_0)\neq 0\) at a point \(x_0\in f^{-1}(c)\), then the implicit function theorem implies that near \(x_0\), the level set is a smooth hypersurface. Locally, one coordinate can be solved as a smooth function of the remaining coordinates, turning the level set into a graph in a neighborhood.
4.3.2 Regular value theorem
The regular value theorem states that if \(f\) is smooth and \(c\) is a regular value, then \(f^{-1}(c)\) is a smooth manifold whose dimension equals \(\dim(D)-1\) in the scalar-valued case. This theorem also underpins results like stability of the level set under smooth perturbations and provides a framework for counting components in certain contexts.
5 Level sets in analysis
5.1 Continuous functions
For merely continuous \(f\), the level set \(f^{-1}(\{c\})\) is closed in \(D\) (since \(\{c\}\) is closed in \(\mathbb{R}\) and preimages of closed sets under continuous maps are closed). However, closedness alone does not guarantee smoothness or manifold structure. The geometry may be fractal-like or highly irregular when \(f\) lacks differentiability.
5.2 Differentiable functions
When \(f\) is differentiable, the differential \(df\) determines local behavior of level sets. Regularity of \(c\) (in the sense that \(df\neq 0\) on the level set) yields manifold structure via standard theorems from differential calculus. If \(df=0\) somewhere on \(f^{-1}(c)\), singularities can appear, and finer tools are required to understand local topology.
5.3 Measure-theoretic aspects
From a measure-theoretic viewpoint, level sets often have special size properties. For example, in many smooth settings, level sets of a real-valued function in Euclidean space have Lebesgue measure zero unless the function is constant on a region. Integrating “along level sets” or “transverse to level sets” leads to results such as coarea-type formulas, which relate integrals over the domain to integrals over families of level sets.
6 Applications
6.1 Optimization
In constrained optimization, level sets model constraints and objective landscapes. Equality constraints \(f(x)=c\) describe feasible manifolds (when regularity holds), while sublevel sets \(f(x)\le c\) represent regions satisfying upper bounds, such as trust regions or error tolerances. For differentiable problems, the structure of level sets near optima provides information about gradients, curvature, and constraint qualification.
6.2 Differential equations
Level sets appear naturally in the study of partial differential equations, particularly those involving propagation or transformation of scalar quantities. For instance, in Hamilton–Jacobi type equations, solutions can often be interpreted through evolving level sets. In elliptic problems, zero sets of harmonic or potential functions serve as interfaces between regions of different sign, and qualitative theory uses level set geometry to describe nodal sets and regularity.
6.3 Physics and engineering
In physics, scalar fields such as temperature, pressure, potential, or refractive index are commonly analyzed through contours of constant value. Level sets then represent equipotential surfaces or iso-value curves used to interpret experimental data and theoretical models. In mechanics and materials science, interfaces between phases can be described by zero level sets of order parameters, enabling a compact mathematical representation of evolving boundaries.
6.4 Computer graphics and visualization
Computer graphics relies heavily on level sets for representing implicit geometry and for generating images. Contour plots produce visible level curves or surfaces from sampled data. Level set methods also provide a numerical framework for tracking evolving shapes by updating a scalar function whose zero level set represents the interface; this is used in applications like fluid simulation, topology change handling, and image segmentation.