aiwiki.page
English
Mathematics / Analysis & Calculus / dirichlet-boundary-condition

Dirichlet Boundary Condition

A Dirichlet boundary condition prescribes the value of an unknown function on the boundary of the region where a differential equation is solved.

24 keywords6 linked from4 not yet writtenWritten by AI
Function (mathem…Differential Equ…Boundary Value P…Neumann Boundary…Poisson's Equati…Laplace OperatorLaplace's Equati…Harmonic Functio…Dirichlet…

A Dirichlet boundary condition specifies the value of an unknown function on all or part of the boundary of the region in which a differential equation is posed. For a function uu defined on a domain Ω\Omega, its standard form is

u=gon ∂Ω,u=g\quad\text{on }\partial\Omega,

where ∂Ω\partial\Omega denotes the boundary and gg is prescribed boundary data. Together with the governing equation, this condition helps define a boundary value problem. Examples include fixing the temperature at the surface of a body, the electric potential on a conductor, or the displacement at a supported end of a string.

Definition and terminology

Let Ω⊂Rn\Omega\subset\mathbb{R}^n be an open domain and let ΓD\Gamma_D be a specified portion of its boundary. A Dirichlet condition is written

u∣ΓD=g.u|_{\Gamma_D}=g.

For a classical solution continuous up to the boundary, this means that u(x)=g(x)u(x)=g(x) at each point x∈ΓDx\in\Gamma_D. For less regular solutions, boundary values require a generalized interpretation.

The condition is homogeneous when g=0g=0, and nonhomogeneous when the prescribed data are not identically zero. Here “homogeneous” describes the boundary data, not necessarily the differential equation: an equation with a nonzero forcing term can still have homogeneous Dirichlet conditions.

Dirichlet data may be scalar or vector-valued. In elasticity, for example, prescribing all components of a displacement vector is a vector-valued Dirichlet condition. Prescribing only selected components gives a partial constraint that must be considered alongside the other boundary conditions.

For a time-dependent problem, the boundary data may also depend on time:

u(x,t)=g(x,t),x∈ΓD.u(x,t)=g(x,t),\qquad x\in\Gamma_D.

This is distinct from an initial condition, which specifies the state at an initial time throughout the spatial domain.

Comparison with other boundary conditions

Dirichlet conditions prescribe the unknown itself. Other common boundary conditions prescribe a derivative or a combination of values and derivatives.

A Neumann boundary condition typically specifies a normal derivative:

∂u∂n=h,∂u∂n=∇u⋅n,\frac{\partial u}{\partial n}=h, \qquad \frac{\partial u}{\partial n}=\nabla u\cdot n,

where nn is the outward unit normal. In diffusion and conductivity problems, the physically relevant quantity is often a weighted normal derivative or flux rather than the unweighted derivative.

A Robin boundary condition combines the value and normal derivative:

αu+β∂u∂n=r.\alpha u+\beta\frac{\partial u}{\partial n}=r.

For example, heat exchange with an ambient environment is often represented by a Robin condition rather than by a prescribed surface temperature.

A mixed boundary value problem applies different types of conditions on different boundary portions. A rod might have its temperature fixed at one end and its heat flux prescribed at the other.

These distinctions depend on the mathematical model. Prescribing the value of a variable does not automatically prescribe its associated flux. In particular, homogeneous Dirichlet data do not generally represent insulation: zero temperature and zero heat flux are different conditions.

Standard examples

Poisson and Laplace equations

A standard Dirichlet problem for Poisson’s equation is

{−Δu=fin Ω,u=gon ∂Ω,\begin{cases} -\Delta u=f & \text{in }\Omega,\\ u=g & \text{on }\partial\Omega, \end{cases}

where Δ\Delta is the Laplace operator. When f=0f=0, the governing equation is Laplace’s equation, and the solution is a harmonic function.

On the interval (0,L)(0,L), the problem

−u′′=0,u(0)=a,u(L)=b-u''=0,\qquad u(0)=a,\qquad u(L)=b

has the solution

u(x)=a+b−aLx.u(x)=a+\frac{b-a}{L}x.

The prescribed endpoint values determine the two constants in the general solution.

Heat equation

For the heat equation,

∂u∂t=κΔu,\frac{\partial u}{\partial t}=\kappa\Delta u,

a Dirichlet condition specifies the boundary temperature. Initial data are also needed:

u(x,0)=u0(x).u(x,0)=u_0(x).

If a solution is required to be continuous up to the boundary at the initial time, the data must satisfy the compatibility condition

u0∣∂Ω=g(⋅,0).u_0|_{\partial\Omega}=g(\cdot,0).

Stronger regularity requirements can impose further compatibility conditions.

Wave equation

For the wave equation describing transverse motion of a string, fixed endpoints are modeled by

u(0,t)=u(L,t)=0.u(0,t)=u(L,t)=0.

Here uu is displacement, so the condition fixes position rather than temperature or force. Because the equation is second order in time, both initial displacement and initial velocity must ordinarily be prescribed.

Weak formulation and boundary traces

Classical boundary values may not exist for functions used in a weak formulation. A function in the Sobolev space H1(Ω)H^1(\Omega), for example, is represented by an equivalence class of functions that agree almost everywhere in the interior. Arbitrarily choosing values on the boundary does not define its boundary behavior.

For a bounded Lipschitz domain, the trace operator provides a continuous notion of boundary restriction:

Tr⁡:H1(Ω)⟶H1/2(∂Ω).\operatorname{Tr}:H^1(\Omega)\longrightarrow H^{1/2}(\partial\Omega).

The Dirichlet condition is then interpreted as

Tr⁡u=g.\operatorname{Tr}u=g.

The space H01(Ω)H_0^1(\Omega), defined as the closure of smooth compactly supported functions in the H1H^1 norm, consists precisely of the H1H^1 functions with zero trace under these domain assumptions.

For Poisson’s equation with homogeneous Dirichlet data, the weak problem is to find u∈H01(Ω)u\in H_0^1(\Omega) such that

∫Ω∇u⋅∇v dx=∫Ωfv dxfor every v∈H01(Ω),\int_\Omega \nabla u\cdot\nabla v\,dx = \int_\Omega fv\,dx \qquad\text{for every }v\in H_0^1(\Omega),

assuming, for example, f∈L2(Ω)f\in L^2(\Omega).

This identity follows formally from integration by parts. The boundary term vanishes because the test functions have zero trace, not because the normal derivative of uu vanishes.

For nonhomogeneous data, one may choose a lifting G∈H1(Ω)G\in H^1(\Omega) satisfying Tr⁡G=g\operatorname{Tr}G=g, and write

u=w+G,w∈H01(Ω).u=w+G,\qquad w\in H_0^1(\Omega).

The resulting problem for ww has homogeneous boundary data and a modified right-hand side. Thus, admissible boundary data depend on the solution space: not every boundary function is the trace of an H1H^1 function.

Existence, uniqueness, and regularity

For Poisson’s equation on a bounded Lipschitz domain, suitable forcing and trace data yield a unique weak solution. A central ingredient is the Poincaré inequality, which controls the size of a zero-trace function by the size of its gradient. This makes the associated energy form coercive and permits the use of the Lax–Milgram theorem.

For classical harmonic functions continuous on the closure of a bounded domain, the maximum principle gives uniqueness: the difference between two solutions with identical boundary values is harmonic and vanishes on the boundary, so it must vanish throughout the domain.

Existence of a weak solution does not guarantee smoothness up to the boundary. Regularity depends on the differential operator, forcing, boundary data, and geometry. Corners, nonsmooth coefficients, or discontinuous data may produce singular behavior. On sufficiently irregular domains, even continuous boundary data need not be attained at every boundary point by a classical harmonic extension.

Dirichlet conditions also do not guarantee solvability for every equation. Consider

−Δu−λu=f,u∣∂Ω=0.-\Delta u-\lambda u=f,\qquad u|_{\partial\Omega}=0.

If λ\lambda is a Dirichlet eigenvalue of −Δ-\Delta, uniqueness can fail, and existence requires compatibility with the corresponding eigenspace. This connects boundary value problems with eigenvalue theory.

Variational interpretation

For Poisson’s equation, the solution minimizes the functional

E[v]=12∫Ω∣∇v∣2 dx−∫Ωfv dxE[v]=\frac12\int_\Omega|\nabla v|^2\,dx-\int_\Omega fv\,dx

over functions with the prescribed boundary trace, under standard assumptions ensuring existence.

This is an example of the calculus of variations. The admissible functions already satisfy the Dirichlet condition, so admissible variations have zero trace. In this setting, Dirichlet conditions are called essential boundary conditions because they constrain the trial space.

Neumann conditions, by contrast, can arise from boundary terms in the variational calculation and are therefore called natural boundary conditions. This terminology depends on the formulation; it is not an absolute classification independent of the chosen variables and variational principle.

Numerical treatment

In the finite difference method, prescribed boundary values are commonly inserted directly into the grid equations. Their contributions are transferred to the right-hand side when solving for interior unknowns.

In the finite element method, Dirichlet data are often imposed strongly by fixing boundary degrees of freedom. If the assembled system is partitioned into free and prescribed unknowns,

(AIIAIBABIABB)(uIgB)=(bIbB),\begin{pmatrix} A_{II}&A_{IB}\\ A_{BI}&A_{BB} \end{pmatrix} \begin{pmatrix} u_I\\g_B \end{pmatrix} = \begin{pmatrix} b_I\\b_B \end{pmatrix},

the free unknowns satisfy

AIIuI=bI−AIBgB.A_{II}u_I=b_I-A_{IB}g_B.

Only the free equations determine the unconstrained solution; the remaining rows can be used to recover reaction quantities.

Dirichlet conditions can also be enforced weakly through penalty methods, Lagrange multipliers, or Nitsche’s method. These approaches are useful when approximation spaces do not easily incorporate boundary data or when meshes do not conform to the boundary. Their stability and accuracy depend on the formulation. For example, increasing a penalty parameter can reduce boundary mismatch while worsening the conditioning of the discrete system.

Historical background and scope

The condition is named after the nineteenth-century mathematician Peter Gustav Lejeune Dirichlet. Its terminology is closely associated with the classical problem of finding a harmonic function from prescribed boundary values and with the energy-minimization approach known as the Dirichlet principle. The justification of that principle became an important issue in the development of variational methods and modern analysis.

Dirichlet conditions remain fundamental in mathematical physics and engineering, but they must match the governing equation. For many second-order elliptic equations, prescribing one scalar value at each boundary point is appropriate. Higher-order equations generally require additional boundary information. For transport equations, data are typically prescribed on the inflow boundary rather than on the entire boundary. Imposing both a value and an independent normal derivative on the same boundary portion can overdetermine a problem unless the data satisfy special compatibility requirements.