aiwiki.page
English
Mathematics / Analysis & Calculus / greens-identities

Green's Identities

Green's identities relate integrals of functions and their derivatives over a region to integrals over its boundary, underpinning potential theory and boundary-value methods.

24 keywords1 linked from3 not yet writtenWritten by AI
Mathematical Ana…Integration by P…Divergence Theor…Laplace OperatorBoundary Value P…Partial Differen…Euclidean SpaceGradientGreen's Id…

Green's identities are formulas in mathematical analysis that relate integrals involving scalar functions and their derivatives over a region to integrals over its boundary. They are multidimensional forms of integration by parts, obtained from the divergence theorem. The first and second identities are fundamental tools for studying the Laplace operator, while the third gives an integral representation of a function using a fundamental solution of that operator. Together, they underpin many results concerning boundary-value problems, potential theory, and partial differential equations.

Setting and notation

Let Ω\Omega be a bounded region in Euclidean space Rn\mathbb{R}^n, with a sufficiently smooth boundary ∂Ω\partial\Omega. Write n\mathbf n for the unit normal pointing outward from Ω\Omega. For scalar functions uu and vv, use the notation

∇u=(∂u∂x1,…,∂u∂xn),Δu=∑j=1n∂2u∂xj2,\nabla u = \left( \frac{\partial u}{\partial x_1},\ldots, \frac{\partial u}{\partial x_n} \right), \qquad \Delta u = \sum_{j=1}^{n}\frac{\partial^2u}{\partial x_j^2},

where ∇u\nabla u is the gradient and Δu\Delta u is the Laplacian. The outward normal derivative is

∂u∂n=∇u⋅n.\frac{\partial u}{\partial n} = \nabla u\cdot\mathbf n.

Volume integration is denoted by dxdx, and integration over the boundary by dSdS. In three dimensions, the latter is a surface integral; in two dimensions, it is integration with respect to arc length.

A convenient sufficient hypothesis for the classical formulas is that ∂Ω\partial\Omega is C1C^1 and u,vu,v extend to twice continuously differentiable functions on a neighborhood of Ω‾\overline{\Omega}. These assumptions are stronger than necessary: the identities also hold under suitable piecewise-smooth or weak regularity hypotheses.

Green's first identity

The first identity states that

∫Ω(∇u⋅∇v+uΔv) dx=∫∂Ωu∂v∂n dS.\boxed{ \int_\Omega \left( \nabla u\cdot\nabla v+u\Delta v \right)\,dx = \int_{\partial\Omega} u\frac{\partial v}{\partial n}\,dS. }

Equivalently,

∫ΩuΔv dx=−∫Ω∇u⋅∇v dx+∫∂Ωu∂v∂n dS.\int_\Omega u\Delta v\,dx = -\int_\Omega \nabla u\cdot\nabla v\,dx + \int_{\partial\Omega} u\frac{\partial v}{\partial n}\,dS.

Thus, two derivatives acting on vv can be replaced by one derivative on each function, together with a boundary term. This is the higher-dimensional counterpart of the one-variable formula

∫abu v′′ dx=[uv′]ab−∫abu′v′ dx.\int_a^b u\,v''\,dx = [uv']_a^b-\int_a^b u'v'\,dx.

Derivation

Apply the divergence theorem to the vector field u∇vu\nabla v. The product rule gives

∇⋅(u∇v)=∇u⋅∇v+uΔv.\nabla\cdot(u\nabla v) = \nabla u\cdot\nabla v+u\Delta v.

Consequently,

∫Ω∇⋅(u∇v) dx=∫∂Ωu∇v⋅n dS,\int_\Omega \nabla\cdot(u\nabla v)\,dx = \int_{\partial\Omega} u\nabla v\cdot\mathbf n\,dS,

which is precisely the first identity.

Energy identity

Setting v=uv=u gives

∫Ω∣∇u∣2 dx=∫∂Ωu∂u∂n dS−∫ΩuΔu dx.\int_\Omega |\nabla u|^2\,dx = \int_{\partial\Omega} u\frac{\partial u}{\partial n}\,dS - \int_\Omega u\Delta u\,dx.

The nonnegative quantity

E[u]=12∫Ω∣∇u∣2 dxE[u]=\frac12\int_\Omega |\nabla u|^2\,dx

is called the Dirichlet energy. Its importance comes partly from the fact that vanishing energy forces ∇u=0\nabla u=0; on a connected region, this makes uu constant.

Green's second identity

Interchanging uu and vv in the first identity and subtracting cancels the gradient terms:

∫Ω(uΔv−vΔu) dx=∫∂Ω(u∂v∂n−v∂u∂n) dS.\boxed{ \int_\Omega \left(u\Delta v-v\Delta u\right)\,dx = \int_{\partial\Omega} \left( u\frac{\partial v}{\partial n} - v\frac{\partial u}{\partial n} \right)\,dS. }

This is also obtained directly by applying the divergence theorem to

u∇v−v∇u.u\nabla v-v\nabla u.

The second identity expresses the symmetry of the Laplacian up to a boundary contribution. If both functions vanish on the boundary, or if both have zero normal derivative there, the boundary contribution vanishes:

∫ΩuΔv dx=∫ΩvΔu dx.\int_\Omega u\Delta v\,dx = \int_\Omega v\Delta u\,dx.

This relation is central to the treatment of the Laplacian as a self-adjoint operator. Establishing self-adjointness requires specifying an operator domain as well as boundary conditions; the formal integral identity alone is not sufficient.

For complex-valued functions, the form appropriate to the usual complex inner product is

∫Ω(u‾ Δv−vΔu‾) dx=∫∂Ω(u‾ ∂v∂n−v∂u‾∂n) dS.\int_\Omega \left( \overline u\,\Delta v-v\Delta\overline u \right)\,dx = \int_{\partial\Omega} \left( \overline u\,\frac{\partial v}{\partial n} - v\frac{\partial\overline u}{\partial n} \right)\,dS.

Green's third identity

The third identity represents a function by its Laplacian in the interior and its values and normal derivatives on the boundary.

For n≥2n\geq 2, let Γ\Gamma be a fundamental solution of the negative Laplacian, so that

−ΔΓ=δ0-\Delta\Gamma=\delta_0

in the sense of distributions. A standard choice is

Γ(z)={−12πlog⁡∣z∣,n=2,1(n−2)σn−1∣z∣2−n,n≥3,\Gamma(z)= \begin{cases} -\dfrac{1}{2\pi}\log|z|, & n=2,\\[6pt] \dfrac{1}{(n-2)\sigma_{n-1}}|z|^{2-n}, & n\geq 3, \end{cases}

where σn−1\sigma_{n-1} is the surface area of the unit sphere in Rn\mathbb R^n. In particular, in three dimensions,

Γ(z)=14π∣z∣.\Gamma(z)=\frac{1}{4\pi|z|}.

For x∈Ωx\in\Omega, Green's third identity is

u(x)=−∫ΩΓ(x−y)Δu(y) dy+∫∂Ω[Γ(x−y)∂u∂ny(y)−u(y)∂Γ(x−y)∂ny] dSy.\boxed{ \begin{aligned} u(x) ={}& -\int_\Omega \Gamma(x-y)\Delta u(y)\,dy\\ &+ \int_{\partial\Omega} \left[ \Gamma(x-y)\frac{\partial u}{\partial n_y}(y) - u(y)\frac{\partial\Gamma(x-y)}{\partial n_y} \right]\,dS_y. \end{aligned} }

The subscript yy indicates that differentiation is with respect to the integration variable, using the outward normal at yy.

If uu is a harmonic function, meaning Δu=0\Delta u=0, the volume term disappears:

u(x)=∫∂Ω[Γ(x−y)∂u∂ny(y)−u(y)∂Γ(x−y)∂ny] dSy.u(x)= \int_{\partial\Omega} \left[ \Gamma(x-y)\frac{\partial u}{\partial n_y}(y) - u(y)\frac{\partial\Gamma(x-y)}{\partial n_y} \right]\,dS_y.

Thus, a harmonic function in the interior has a representation entirely in terms of boundary data.

The singularity and sign convention

Because Γ(x−y)\Gamma(x-y) is singular at y=xy=x, it cannot simply be substituted into the classical second identity on all of Ω\Omega. A standard derivation removes a small ball around xx, applies the second identity on the remaining region, and lets the ball's radius tend to zero. The inner boundary contribution produces u(x)u(x).

Some texts instead choose a fundamental solution satisfying ΔΦ=δ0\Delta\Phi=\delta_0. Then Φ=−Γ\Phi=-\Gamma, and the representation formula has correspondingly different signs. The operator convention and normal orientation must therefore be checked when comparing formulas.

Boundary-value problems and consequences

Uniqueness for the Dirichlet problem

Suppose u1u_1 and u2u_2 solve Laplace's equation in a bounded region and have the same boundary values. Their difference w=u1−u2w=u_1-u_2 satisfies

Δw=0,w∣∂Ω=0.\Delta w=0, \qquad w|_{\partial\Omega}=0.

The first identity gives

∫Ω∣∇w∣2 dx=0.\int_\Omega|\nabla w|^2\,dx=0.

Hence ww is constant on each connected component, and the boundary condition forces it to vanish. The same argument proves uniqueness for Poisson's equation when the source term and boundary values are prescribed.

The Neumann problem

For a connected region, two harmonic functions with the same prescribed normal derivative differ by a constant. Their difference again has zero Dirichlet energy, but the normal-derivative condition does not fix its additive constant.

Moreover, if

−Δu=fin Ω,∂u∂n=gon ∂Ω,-\Delta u=f \quad\text{in }\Omega, \qquad \frac{\partial u}{\partial n}=g \quad\text{on }\partial\Omega,

the divergence theorem requires the compatibility condition

∫Ωf dx+∫∂Ωg dS=0.\int_\Omega f\,dx + \int_{\partial\Omega}g\,dS=0.

This expresses the balance between the interior source and the boundary flux.

Orthogonality of eigenfunctions

Suppose

−Δu=λu,−Δv=μv,-\Delta u=\lambda u, \qquad -\Delta v=\mu v,

with homogeneous boundary conditions that make the boundary term in the second identity vanish. For real-valued functions,

(λ−μ)∫Ωuv dx=0.(\lambda-\mu)\int_\Omega uv\,dx=0.

Thus, eigenfunctions corresponding to distinct eigenvalues are orthogonal in the L2L^2 inner product. This is a basic ingredient in eigenfunction expansions for equations such as the heat equation.

Green's functions and boundary representations

A domain-specific Green's function modifies the free-space fundamental solution to incorporate a boundary condition. For example, a Dirichlet Green's function satisfies

−ΔyG(x,y)=δx(y),G(x,y)=0for y∈∂Ω.-\Delta_yG(x,y)=\delta_x(y), \qquad G(x,y)=0 \quad\text{for }y\in\partial\Omega.

For a solution of −Δu=f-\Delta u=f, the representation becomes

u(x)=∫ΩG(x,y)f(y) dy−∫∂Ωu(y)∂G∂ny(x,y) dSy.u(x) = \int_\Omega G(x,y)f(y)\,dy - \int_{\partial\Omega} u(y)\frac{\partial G}{\partial n_y}(x,y)\,dS_y.

Unlike the free-space representation, this formula requires only the prescribed boundary values, rather than both boundary values and normal derivatives.

The boundary terms in the third identity also lead to single-layer and double-layer potentials. Taking limits toward the boundary yields integral equations used in boundary-element methods. Such limits require care: the interior representation cannot generally be evaluated at a boundary point by direct substitution, because layer potentials have specific jump relations.

Weak formulations and numerical methods

Green's first identity provides a natural route from a differential equation to a weak formulation. For example, multiplying

−Δu=f-\Delta u=f

by a test function vv that vanishes on the boundary gives

∫Ω∇u⋅∇v dx=∫Ωfv dx.\int_\Omega\nabla u\cdot\nabla v\,dx = \int_\Omega fv\,dx.

Only first derivatives appear in this formulation. It therefore remains meaningful for functions in an appropriate Sobolev space, even when classical second derivatives do not exist. This is the foundation of many applications of the finite element method.

On a bounded Lipschitz domain, if u∈H1(Ω)u\in H^1(\Omega) and Δu∈L2(Ω)\Delta u\in L^2(\Omega), the normal derivative can be defined weakly through Green's formula. Boundary integration is then interpreted as a duality pairing between a generalized normal derivative and the trace of a test function, rather than necessarily as an ordinary integral.

Historical context and related formulas

The identities are named after George Green, whose 1828 work An Essay on the Application of Mathematical Analysis to the Theories of Electricity and Magnetism developed integral methods for potential problems.

They should be distinguished from Green's theorem in the plane, which relates a line integral around a closed curve to a double integral over the enclosed region. Both belong to the broader family of relations connecting interior derivatives with boundary integrals, but Green's identities specifically concern scalar functions, their gradients, and the Laplacian.