✓ Verified by Aevum Editorial Board

Classification of Partial Differential Equations

📅 Updated: November 2025 ⏱️ Read time: 14 min 🏷️ Tags: PDEs, Analysis, Applied Math

The classification of partial differential equations (PDEs) is a foundational framework in mathematical analysis that categorizes equations based on their intrinsic geometric and analytic properties. This classification determines the appropriate numerical methods, well-posedness conditions, and physical interpretations applicable to each equation.

Unlike ordinary differential equations (ODEs), PDEs describe phenomena evolving across multiple independent variables—typically space and time. Their classification originates from the analogy with quadratic curves in analytic geometry, extending naturally to higher dimensions via the spectral properties of the principal symbol.

1. Linearity & Order

Before geometric classification, PDEs are first categorized by their order (the highest derivative present) and linearity. A general PDE of order \(n\) in \(k\) independent variables \(x_1, \dots, x_k\) and unknown \(u\) takes the form:

F\left(x_1, \dots, x_k, u, \partial u, \dots, D^n u\right) = 0

Linearity is determined by the dependence on \(u\) and its derivatives:

Linear: \(u\) and all derivatives appear to the first power and are not multiplied together.
Quasilinear: Linear in the highest-order derivatives, but coefficients may depend on lower-order terms.
Semilinear: Linear in highest derivatives with coefficients depending only on independent variables.
Fully nonlinear: Nonlinear in the highest-order derivatives.

Note: The standard geometric classification (elliptic/parabolic/hyperbolic) applies strictly to second-order quasilinear PDEs. For nonlinear equations, classification becomes pointwise and may change across the domain.

2. Second-Order Classification

Consider a second-order PDE in two independent variables \(x\) and \(y\):

A u_{xx} + B u_{xy} + C u_{yy} + D u_x + E u_y + F u = G

where \(A, B, C, D, E, F, G\) are functions of \(x\) and \(y\). The classification depends entirely on the discriminant of the principal part (the highest-order terms):

\Delta = B^2 - 4AC

Type	Condition	Analogous Curve	Typical Boundary Conditions
Elliptic	\(\Delta < 0\)	Ellipse	Dirichlet / Neumann on closed boundary
Parabolic	\(\Delta = 0\)	Parabola	Initial + boundary conditions
Hyperbolic	\(\Delta > 0\)	Hyperbola	Initial data on non-characteristic curve

2.1 Elliptic Equations

Elliptic PDEs model steady-state phenomena and equilibrium distributions. They possess no real characteristic curves, implying that information propagates instantaneously throughout the domain. Solutions are typically smooth (analytic) within the interior.

\nabla^2 u = u_{xx} + u_{yy} = 0

Other examples include the Poisson equation, Helmholtz equation, and equations governing steady heat conduction, electrostatics, and incompressible fluid flow.

2.2 Parabolic Equations

Parabolic PDEs describe diffusion and dissipative processes. They feature a single family of real characteristics and exhibit infinite speed of propagation with strong smoothing effects over time.

u_t = \alpha \nabla^2 u \quad \text{or} \quad u_t - u_{xx} = 0

Here, \(\Delta = 0\) because the mixed derivative term is absent and the time derivative acts as a first-order operator in the spatial classification context. Solutions smooth out initial discontinuities immediately for \(t > 0\).

2.3 Hyperbolic Equations

Hyperbolic PDEs model wave propagation and conservation laws. They possess two distinct families of real characteristics along which information travels at finite speeds. Discontinuities can propagate along these characteristics.

u_{tt} = c^2 \nabla^2 u \quad \text{or} \quad u_{tt} - c^2 u_{xx} = 0

Here \(A = -c^2, B = 0, C = 1\), yielding \(\Delta = 0 - 4(-c^2)(1) = 4c^2 > 0\). Initial-value problems (Cauchy problems) are well-posed for hyperbolic systems.

3. Generalization to n Dimensions

For a second-order PDE in \(n\) variables \(x_1, \dots, x_n\), the principal part is given by the quadratic form:

\sum_{i,j=1}^{n} A_{ij}(x) \frac{\partial^2 u}{\partial x_i \partial x_j}

Let \(\mathbf{A}(x) = [A_{ij}]\) be the symmetric coefficient matrix. Classification is determined by the eigenvalues \(\lambda_1, \dots, \lambda_n\) of \(\mathbf{A}\):

Elliptic: All eigenvalues have the same sign (positive or negative definite).
Parabolic: One eigenvalue is zero, others share the same sign (positive/negative semidefinite).
Hyperbolic: One eigenvalue has opposite sign to the others (signature \((n-1,1)\) or \((1,n-1)\)).
Ultrahyperbolic: At least two positive and two negative eigenvalues.

This spectral classification underpins modern geometric analysis and the study of pseudo-differential operators.

4. Canonical Forms & Characteristic Curves

A powerful technique for solving PDEs involves a coordinate transformation \((x,y) \mapsto (\xi, \eta)\) that eliminates mixed derivatives, reducing the equation to a canonical form:

Elliptic: \(u_{\xi\xi} + u_{\eta\eta} + \dots = 0\)
Parabolic: \(u_{\eta\eta} + \dots = 0\) (or \(u_{\xi\xi} + \dots = 0\))
Hyperbolic: \(u_{\xi\eta} + \dots = 0\) or \(u_{\xi\xi} - u_{\eta\eta} + \dots = 0\)

The transformation is constructed using the characteristic equation:

A \left(\frac{dy}{dx}\right)^2 - B \frac{dy}{dx} + C = 0

Real, complex, or repeated roots correspond directly to hyperbolic, elliptic, or parabolic types. This geometric perspective unifies PDE classification with the theory of first-order ODE systems.

References & Further Reading

Evad, L. (1988). Partial Differential Equations. Birkhäuser. pp. 45–62.
Taylor, M. E. (2011). Partial Differential Equations I: Basic Theory. Springer. Section 2.1.
John, F. (1991). Partial Differential Equations (4th ed.). Springer. Chapter 1.
Evans, L. C. (2010). Partial Differential Equations (2nd ed.). AMS. Section 2.1.
Aevum Encyclopedia Editorial Board. (2025). "Geometric Classification in Higher Dimensions". Aevum Math. Rev. 12(3): 112–129.