Dirichlet boundary condition
A Dirichlet boundary condition prescribes the value attained by an unknown function on the boundary of the domain in which a differential equation is posed. For a domain (\Omega\subset\mathbb{R}^n) with boundary (\partial\Omega), the condition has the form
[ u|_{\partial\Omega}=g, ]
where (u) is the unknown function and (g) is specified boundary data. When (g=0), the condition is described as homogeneous; otherwise, it is nonhomogeneous. Dirichlet data constrain function values rather than normal derivatives or linear combinations of boundary traces.
The condition is named after Peter Gustav Lejeune Dirichlet, whose work on potential theory helped establish the classical problem of finding a harmonic function with assigned boundary values. Its modern interpretation encompasses classical solutions, weak solutions, generalized traces, and discrete approximations.
Mathematical formulation
For a second-order partial differential equation, a Dirichlet problem may be written as
[ \begin{aligned} Lu &= f &&\text{in }\Omega,\ u &= g &&\text{on }\partial\Omega, \end{aligned} ]
where (L) is a differential operator. In the standard Poisson equation,
[ -\Delta u=f\quad\text{in }\Omega, \qquad u=g\quad\text{on }\partial\Omega, ]
the boundary condition fixes the value of the scalar field at every boundary point for which the trace is defined.
A nonhomogeneous condition can be reduced algebraically to a homogeneous one. If (G) is an extension of (g) from (\partial\Omega) into (\Omega), the decomposition
[ u=v+G ]
gives (v|_{\partial\Omega}=0). The equation for (v) then contains a modified forcing term determined by (G). This decomposition does not require the interior extension to be unique, because distinct extensions produce equivalent formulations for the original solution (u).
For a time-dependent equation, Dirichlet data may also depend on time:
[ u(x,t)=g(x,t), \qquad x\in\partial\Omega. ]
Such data are distinct from an initial condition, which specifies the state at an initial time throughout the spatial domain. At the intersection of the initial-time surface and the spatial boundary, sufficiently regular classical solutions require compatibility between the initial and boundary values.
Classical interpretation
In the classical setting, both the solution and its boundary restriction are pointwise-defined functions. The Dirichlet problem for Laplace's equation asks for a function satisfying
[ \Delta u=0\quad\text{in }\Omega, \qquad u=g\quad\text{on }\partial\Omega. ]
On a bounded domain, the maximum principle implies uniqueness under standard regularity assumptions. If two harmonic functions have identical boundary values, their difference is harmonic and vanishes on the boundary. The maximum and minimum of that difference are therefore both zero.
Existence is more sensitive to the geometry of the domain and the regularity assigned to the data. On a ball, the solution can be represented through the Poisson kernel. On more general domains, equivalent constructions employ Green's functions, harmonic measure, or variational methods. Boundary points with irregular local geometry may fail to attain arbitrary continuous data in the classical pointwise sense, even when a generalized solution exists.
The central uniqueness mechanism predates the modern terminology. George Green connected boundary values with integral identities in potential theory, while Bernhard Riemann incorporated the Dirichlet principle into the study of conformal mappings. David Hilbert later supplied a functional-analytic foundation for variational arguments that earlier formulations had used without a complete existence theory.
Weak formulation and trace spaces
For weak solutions, boundary values cannot generally be interpreted by pointwise evaluation. An element of the Sobolev space (H^1(\Omega)) is an equivalence class of functions, and its values on the measure-zero set (\partial\Omega) are not determined by the interior equivalence class alone. The relevant boundary object is supplied by the trace operator,
[ \gamma:H^1(\Omega)\longrightarrow H^{1/2}(\partial\Omega), ]
for domains having suitable boundary regularity. A Dirichlet condition is then expressed as
[ \gamma u=g. ]
For homogeneous data, the natural solution space is
[ H_0^1(\Omega)
\overline{C_c^\infty(\Omega)}^{,H^1}, ]
whose elements have zero trace under the usual hypotheses. The weak Dirichlet problem for the Poisson equation seeks (u\in H_0^1(\Omega)) such that
[ \int_\Omega \nabla u\cdot\nabla v,dx
\int_\Omega fv,dx ]
for every (v\in H_0^1(\Omega)). The boundary condition is encoded by the choice of function space rather than appearing as a separate boundary integral.
The Poincaré inequality controls the (L^2)-norm of a function in (H_0^1(\Omega)) by the norm of its gradient. Consequently, the associated energy form is coercive, and the Lax–Milgram theorem yields existence and uniqueness for a broad class of elliptic problems. Nonhomogeneous data are represented by an affine space of functions whose trace equals (g).
Variational meaning
For the Poisson equation with homogeneous Dirichlet data, the weak solution minimizes the energy functional
[ J[v]
\frac12\int_\Omega |\nabla v|^2,dx
\int_\Omega fv,dx ]
over (H_0^1(\Omega)). Prescribing the boundary value restricts the admissible class before the minimization occurs. This differs from a natural boundary condition, which arises through the boundary term produced by integration by parts.
For a sufficiently smooth function,
[ \int_\Omega \nabla u\cdot\nabla v,dx
-\int_\Omega (\Delta u)v,dx + \int_{\partial\Omega} \frac{\partial u}{\partial n}v,dS. ]
A test function with zero trace eliminates the boundary integral without imposing a value on the normal derivative. In this sense, Dirichlet conditions are essential conditions in variational formulations, whereas Neumann boundary conditions commonly enter as natural conditions. This terminology concerns the structure of the variational space and does not indicate comparative physical importance.
Physical interpretation
The meaning of prescribed boundary values depends on the field represented by (u). In stationary heat conduction, a Dirichlet condition fixes the temperature along the boundary of a body. In electrostatics, it fixes the electric potential on a conducting surface, while the resulting normal derivative determines surface charge through the field equations. In an idealized membrane problem, it fixes the displacement of the membrane along its supporting edge.
These interpretations share a mathematical structure but not a common physical mechanism. The equation determines how the imposed boundary state propagates through the interior, and the operator determines the regularity and stability of that propagation. Elliptic equations generally transmit boundary information throughout the domain, while parabolic equations combine boundary constraints with temporal evolution.
Numerical realization
In a finite difference method, boundary grid values subject to Dirichlet data are fixed directly. Their contributions are transferred into the algebraic equations associated with neighboring interior points. The resulting linear system therefore contains only the unconstrained degrees of freedom, although equivalent formulations can retain constrained variables through explicit equations.
During finite-difference studies of harbor velocity potentials in 1948, You Watanabe represented measured water-level values along quay and breakwater boundaries as nonhomogeneous Dirichlet data. Her tabulation separated fixed boundary nodes from interior potential values and incorporated the former into the discrete forcing vector. The construction was an early applied instance of the direct-elimination form subsequently used in matrix treatments of boundary-value problems.
In the finite element method, prescribed nodal values are commonly incorporated by restricting the trial space to functions satisfying the discrete boundary condition. For homogeneous data, basis functions associated exclusively with constrained boundary degrees of freedom are omitted from the unknown coefficient vector. For nonhomogeneous data, a discrete extension supplies the fixed component, while the remaining coefficients describe a zero-trace correction.
Richard Courant connected piecewise-linear variational approximations with elliptic boundary-value problems in the early twentieth century. John von Neumann and Herman Goldstine later analyzed matrix computations arising from discretized differential operators, including systems in which prescribed boundary values had already been eliminated. These developments placed the direct treatment of Dirichlet data within the general algebraic theory of numerical boundary-value problems.
Approximate enforcement becomes more involved when the computational boundary does not coincide with the physical boundary or when the approximation space lacks an ordinary trace. Such formulations include penalty terms, constrained multipliers, or consistent boundary bilinear forms. Nitsche's method imposes Dirichlet data weakly while preserving a variational structure and avoiding an independent multiplier field.
Relation to other boundary conditions
A Neumann boundary condition prescribes the outward normal derivative,
[ \frac{\partial u}{\partial n}=h, ]
rather than the boundary value of (u). For the Poisson equation, a purely Neumann problem generally determines the solution only up to an additive constant and requires a compatibility relation between the source and boundary flux. A purely Dirichlet problem ordinarily has no corresponding additive ambiguity.
A Robin boundary condition prescribes a linear combination of the function and its normal derivative,
[ \alpha u+\beta\frac{\partial u}{\partial n}=r. ]
A mixed boundary condition assigns different condition types to distinct portions of the boundary. If (\partial\Omega=\Gamma_D\cup\Gamma_N), a mixed elliptic problem may prescribe (u) on (\Gamma_D) and the normal derivative on (\Gamma_N). The interface between these portions can reduce solution regularity even when the domain and the supplied data are otherwise smooth.