Skip to content

Energy Green Function

The energy Green function is the coordinate-space kernel of a resolvent boundary value. With the outgoing convention,

G+(x,x′;E)=⟨x∣1E−H+i0∣x′⟩.G^+(x,x';E) = \left\langle x\left\rvert \frac{1}{E-H+i0} \right\lvert x'\right\rangle.

It answers an energy-domain source problem:

(E−Hx)G+(x,x′;E)=δ(x−x′),(E-H_x)G^+(x,x';E) = \delta(x-x'),

together with the spatial and asymptotic boundary conditions encoded by HH and +i0+i0.

This page is the canonical home for the coordinate kernel, its differential equation, and the free outgoing examples. The abstract operator-valued inverse belongs to Resolvent Operator, while the general discrete-plus-continuum expansion belongs to Spectral Representation of Green Functions.

For a self-adjoint Hamiltonian HH, the resolvent is

R(z)=(zI−H)−1R(z)=(zI-H)^{-1}

when zz lies outside the spectrum. In a coordinate basis,

G(x,x′;z)=⟨x∣R(z)∣x′⟩.G(x,x';z) = \langle x\rvert R(z)\lvert x'\rangle.

For a real energy in the continuous spectrum, the inverse does not exist as an ordinary bounded operator at z=Ez=E. One instead takes a boundary value:

G±(x,x′;E)=lim⁡ϵ→0+⟨x∣1E−H±iϵ∣x′⟩.G^\pm(x,x';E) = \lim_{\epsilon\to0^+} \left\langle x\left\rvert \frac{1}{E-H\pm i\epsilon} \right\lvert x'\right\rangle.

The superscripts have physical content:

KernelResolvent boundary valueTypical asymptotic condition
G+G^+(E−H+i0)−1(E-H+i0)^{-1}outgoing
G−G^-(E−H−i0)−1(E-H-i0)^{-1}incoming

For a time-reversal-invariant real Hamiltonian and real EE where both boundary values are defined,

G−(x,x′;E)=G+(x′,x;E)∗.G^-(x,x';E) = G^+(x',x;E)^*.

The energy Green function is not a time-evolution kernel. It has dimensions of inverse energy times the coordinate delta distribution. In dd Cartesian dimensions,

[G]=1energy lengthd.[G]=\frac{1}{\text{energy}\,\text{length}^d}.

The operator identity

(zI−H)R(z)=I(zI-H)R(z)=I

implies, after taking coordinate matrix elements,

(z−Hx)G(x,x′;z)=δ(x−x′).(z-H_x)G(x,x';z) = \delta(x-x').

For a one-dimensional Schrödinger Hamiltonian

H=−ℏ22md2dx2+V(x),H = -\frac{\hbar^2}{2m}\frac{d^2}{dx^2} +V(x),

the Green function satisfies

[E−V(x)+ℏ22md2dx2]G+(x,x′;E)=δ(x−x′).\left[ E-V(x) +\frac{\hbar^2}{2m}\frac{d^2}{dx^2} \right] G^+(x,x';E) = \delta(x-x').

Away from x=x′x=x', it solves the homogeneous stationary Schrödinger equation. At the source point, GG is continuous for an ordinary finite potential, while its first derivative has a fixed jump. Integrating across x′x' gives

∂G∂x∣x=x′+−∂G∂x∣x=x′−=2mℏ2.\left. \frac{\partial G}{\partial x} \right|_{x=x'^+} - \left. \frac{\partial G}{\partial x} \right|_{x=x'^-} = \frac{2m}{\hbar^2}.

This jump condition fixes the overall normalization after the left and right homogeneous solutions have been chosen.

If

(E−H)ψ(x)=f(x),(E-H)\psi(x)=f(x),

then a solution with the boundary prescription carried by GG is

ψ(x)=∫dx′ G(x,x′;E)f(x′).\psi(x) = \int dx'\, G(x,x';E)f(x').

The integration measure and domain must match the coordinate basis. On a curved or radial configuration space, the delta function and integral carry the corresponding measure factors.

An inverse differential equation does not select a unique Green function until boundary conditions are supplied. The energy Green function must satisfy:

  • the operator-domain conditions of HH in the coordinate variable xx;
  • the adjoint conditions in x′x';
  • the source normalization at x=x′x=x';
  • an asymptotic prescription for continuum energies.

For a hard-wall interval, GG vanishes at each Dirichlet endpoint. For a radial problem, regularity at the origin and the radial measure matter. For a bound-state problem below threshold, the kernel decays at spatial infinity. For positive-energy scattering on the full line or in three dimensions, G+G^+ selects outgoing waves and G−G^- selects incoming waves.

In three dimensions, the outgoing Sommerfeld condition has the schematic form

lim⁡r→∞r(∂∂r−ik)G+(r,r′;E)=0,\lim_{r\to\infty} r\left( \frac{\partial}{\partial r}-ik \right) G^+(\mathbf r,\mathbf r';E) =0,

with r′\mathbf r' fixed and

E=ℏ2k22m.E=\frac{\hbar^2k^2}{2m}.

The incoming kernel uses the opposite sign. Spatial boundaries and the i0i0 prescription solve different parts of the specification: the former define the Hamiltonian domain, while the latter select a continuum boundary value at infinity.

The notation

1E−H±i0\frac{1}{E-H\pm i0}

means

lim⁡ϵ→0+1E−H±iϵ.\lim_{\epsilon\to0^+} \frac{1}{E-H\pm i\epsilon}.

The scalar distribution identity

1x±i0=PV⁡1x∓iπδ(x)\frac{1}{x\pm i0} = \operatorname{PV}\frac{1}{x} \mp i\pi\delta(x)

lifts through the spectral theorem to

R+(E)−R−(E)=−2πi δ(E−H).R^+(E)-R^-(E) = -2\pi i\,\delta(E-H).

Consequently,

−1πIm⁡G+(x,x;E)-\frac{1}{\pi} \operatorname{Im}G^+(x,x;E)

is the local spectral density under the stated convention. The trace gives the total density of states when that trace is well defined or appropriately regularized. Those formulas are developed in Green Functions and Density of States.

For finite ϵ=η>0\epsilon=\eta\gt0,

Gη(E)=1E−H+iηG_\eta(E) = \frac{1}{E-H+i\eta}

is a smoothened resolvent. Numerically, η\eta broadens delta-function spectral lines into Lorentzian peaks. It is useful, but it is not identical to the exact i0i0 limit.

For a time-independent Hamiltonian,

1E−H+i0=−iℏ∫0∞dt ei(E+i0)t/ℏe−iHt/ℏ.\begin{aligned} \frac{1}{E-H+i0} &= -\frac{i}{\hbar} \int_0^\infty dt\, e^{i(E+i0)t/\hbar} e^{-iHt/\hbar}. \end{aligned}

Thus G+(E)G^+(E) is the energy transform of a retarded time-domain evolution kernel in this convention. The step-function normalization and advanced counterpart belong to Retarded and Advanced Green Functions.

Let

H0=−ℏ22md2dx2,E=ℏ2k22m,k>0.H_0 = -\frac{\hbar^2}{2m}\frac{d^2}{dx^2}, \qquad E=\frac{\hbar^2k^2}{2m}, \qquad k\gt0.

The outgoing Green function solves

(d2dx2+k2)G0+(x,x′;E)=2mℏ2δ(x−x′).\left( \frac{d^2}{dx^2}+k^2 \right) G_0^+(x,x';E) = \frac{2m}{\hbar^2}\delta(x-x').

For x≠x′x\ne x', outgoing behavior requires a wave traveling away from the source point on each side. The solution is

G0+(x,x′;E)=−imℏ2keik∣x−x′∣.G_0^+(x,x';E) = -\frac{im}{\hbar^2k} e^{ik\lvert x-x'\rvert}.

Its derivative jump is

∂xG0+∣x=x′+−∂xG0+∣x=x′−=2mℏ2,\begin{aligned} &\left. \partial_xG_0^+ \right|_{x=x'^+} - \left. \partial_xG_0^+ \right|_{x=x'^-}\\ &\qquad= \frac{2m}{\hbar^2}, \end{aligned}

which verifies the delta-function normalization. The incoming kernel is the complex conjugate:

G0−(x,x′;E)=imℏ2ke−ik∣x−x′∣.G_0^-(x,x';E) = \frac{im}{\hbar^2k} e^{-ik\lvert x-x'\rvert}.

The absolute-value displacement makes the physical meaning transparent: for x>x′x\gt x', G0+G_0^+ carries eikxe^{ikx} to the right; for x<x′x\lt x', it carries e−ikxe^{-ikx} to the left. Both pieces move away from the source.

The momentum representation is

G0+(r,r′;E)=∫d3p(2πℏ)3eip⋅(r−r′)/ℏE−p2/(2m)+i0.\begin{aligned} G_0^+(\mathbf r,\mathbf r';E) &= \int\frac{d^3p}{(2\pi\hbar)^3} \frac{e^{i\mathbf p\cdot(\mathbf r-\mathbf r')/\hbar}} {E-\mathbf p^2/(2m)+i0}. \end{aligned}

Let

R=∣r−r′∣,k=2mEℏ.R=\lvert\mathbf r-\mathbf r'\rvert, \qquad k=\frac{\sqrt{2mE}}{\hbar}.

For E>0E\gt0, contour integration or the outgoing Helmholtz fundamental solution gives

G0+(r,r′;E)=−m2πℏ2eikRR.G_0^+(\mathbf r,\mathbf r';E) = -\frac{m}{2\pi\hbar^2} \frac{e^{ikR}}{R}.

The distributional identity

(∇2+k2)eikRR=−4πδ(3)(r−r′)(\nabla^2+k^2) \frac{e^{ikR}}{R} = -4\pi\delta^{(3)} (\mathbf r-\mathbf r')

checks the normalization. At large rr with r′\mathbf r' fixed,

G0+∼eikrr×an angular phase,G_0^+ \sim \frac{e^{ikr}}{r} \times \text{an angular phase},

so +i0+i0 has selected an outgoing spherical wave.

For E<0E\lt0, define

κ=−2mEℏ.\kappa=\frac{\sqrt{-2mE}}{\hbar}.

Then the decaying free resolvent is

G0(r,r′;E)=−m2πℏ2e−κRR.G_0(\mathbf r,\mathbf r';E) = -\frac{m}{2\pi\hbar^2} \frac{e^{-\kappa R}}{R}.

Below the continuum threshold there is no outgoing-versus-incoming oscillatory ambiguity. The physically appropriate spatial condition is decay.

Suppose HH has normalized isolated bound states

H∣n⟩=En∣n⟩.H\lvert n\rangle=E_n\lvert n\rangle.

Their contribution to the coordinate Green function is

G(x,x′;z)⊃∑nψn(x)ψn∗(x′)z−En.G(x,x';z) \supset \sum_n \frac{\psi_n(x)\psi_n^*(x')}{z-E_n}.

Near a nondegenerate isolated eigenvalue,

G(x,x′;z)=ψn(x)ψn∗(x′)z−En+Gregular(x,x′;z).G(x,x';z) = \frac{\psi_n(x)\psi_n^*(x')}{z-E_n} + G_{\mathrm{regular}}(x,x';z).

Therefore

Res⁡z=EnG(x,x′;z)=ψn(x)ψn∗(x′).\operatorname*{Res}_{z=E_n} G(x,x';z) = \psi_n(x)\psi_n^*(x').

For a degenerate eigenspace, the residue is the full projector kernel

⟨x∣Pn∣x′⟩.\langle x\rvert P_n\lvert x'\rangle.

Poles reveal bound-state energies, while residues reveal their wavefunction or projector content. Continuum thresholds instead produce boundary-value discontinuities and branch-cut structure. Resonance poles require analytic continuation away from the physical sheet and should not be confused with normalizable bound-state poles.

The full organization belongs to Spectral Representation of Green Functions and Bound States and Scattering Poles.

Let

H=H0+V,H=H_0+V,

and consider a free incoming state ∣k⟩\lvert\mathbf k\rangle at energy EE. The outgoing Lippmann–Schwinger equation is

∣ψk(+)⟩=∣k⟩+1E−H0+i0V∣ψk(+)⟩.\lvert\psi_{\mathbf k}^{(+)}\rangle = \lvert\mathbf k\rangle + \frac{1}{E-H_0+i0} V \lvert\psi_{\mathbf k}^{(+)}\rangle.

In coordinate space,

ψk(+)(r)=eik⋅r+∫d3r′ G0+(r,r′;E)×V(r′)ψk(+)(r′).\begin{aligned} \psi_{\mathbf k}^{(+)}(\mathbf r) &= e^{i\mathbf k\cdot\mathbf r} + \int d^3r'\, G_0^+(\mathbf r,\mathbf r';E)\\ &\qquad\times V(\mathbf r') \psi_{\mathbf k}^{(+)}(\mathbf r'). \end{aligned}

The factor eik∣r−r′∣/∣r−r′∣e^{ik\lvert\mathbf r-\mathbf r'\rvert}/\lvert\mathbf r-\mathbf r'\rvert produces the outgoing spherical part of the asymptotic wave. Replacing +i0+i0 by −i0-i0 selects incoming asymptotics instead.

This page owns the coordinate-kernel definition, source normalization, and free examples. Green Function for Scattering owns the scattering-specific passage from a continuum boundary value through far-field factorization to the amplitude. Lippmann–Schwinger Equation Preview develops the operator bridge from resolvents to formal scattering states, while Lippmann–Schwinger Equation owns the exact state equation.

  • Calling (E−H)−1(E-H)^{-1} well defined on the continuous spectrum without a boundary value or regularization.
  • Omitting the coordinate delta function in the defining differential equation.
  • Solving the homogeneous equation on either side of the source but missing the derivative jump.
  • Treating +i0+i0 as decorative notation rather than the outgoing boundary prescription.
  • Reversing the outgoing and incoming signs after changing Fourier conventions.
  • Applying a full-line or free-space Green function when walls, topology, or radial domains require different spatial conditions.
  • Confusing the energy Green function with the time-domain propagator kernel.
  • Squaring GG and interpreting it as a normalized transition probability.
  • Reading a finite numerical broadening η\eta as the exact infinitesimal limit.
  • Treating every singularity as a normalizable bound state and ignoring thresholds, cuts, and resonances.
  • Dropping degenerate-state projectors from pole residues.
  • Using the outgoing free Green function without matching the plane-wave normalization in a scattering calculation.
  • E. N. Economou, Green’s Functions in Quantum Physics, 3rd ed., Springer, 2006.
  • R. G. Newton, Scattering Theory of Waves and Particles, 2nd ed., Springer, 1982.
  • J. R. Taylor, Scattering Theory: The Quantum Theory of Nonrelativistic Collisions, Wiley, 1972.
  • M. Reed and B. Simon, Methods of Modern Mathematical Physics III: Scattering Theory, Academic Press, 1979.
  • A. L. Fetter and J. D. Walecka, Quantum Theory of Many-Particle Systems, Dover, 2003.
  • J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020.
  1. Derive the differential equation for G(x,x′;z)G(x,x';z) from the resolvent identity.
Solution

Start with

(zI−H)R(z)=I.(zI-H)R(z)=I.

Insert coordinate states:

⟨x∣(zI−H)R(z)∣x′⟩=⟨x∣x′⟩.\langle x\rvert (zI-H)R(z) \lvert x'\rangle = \langle x\vert x'\rangle.

The operator HH acts on the xx variable, so

(z−Hx)⟨x∣R(z)∣x′⟩=δ(x−x′).(z-H_x) \langle x\rvert R(z)\lvert x'\rangle = \delta(x-x').

Using the definition of the coordinate kernel gives

(z−Hx)G(x,x′;z)=δ(x−x′).(z-H_x)G(x,x';z) = \delta(x-x').
  1. Determine the coefficient of the outgoing one-dimensional free Green function from its derivative jump.
Solution

Use the outgoing ansatz

G0+(x,x′;E)=Aeik∣x−x′∣.G_0^+(x,x';E) = A e^{ik\lvert x-x'\rvert}.

For x>x′x\gt x',

∂xG0+=ikAeik(x−x′),\partial_xG_0^+=ikA e^{ik(x-x')},

while for x<x′x\lt x',

∂xG0+=−ikAeik(x′−x).\partial_xG_0^+=-ikA e^{ik(x'-x)}.

At x=x′x=x', the jump is

2ikA.2ikA.

The source equation requires

2ikA=2mℏ2,2ikA=\frac{2m}{\hbar^2},

so

A=miℏ2k=−imℏ2k.A = \frac{m}{i\hbar^2k} = -\frac{im}{\hbar^2k}.
  1. Verify the normalization of the three-dimensional free outgoing Green function.
Solution

For

G0+(R;E)=−m2πℏ2eikRR,G_0^+(R;E) = -\frac{m}{2\pi\hbar^2} \frac{e^{ikR}}{R},

use

(∇2+k2)eikRR=−4πδ(3)(R).(\nabla^2+k^2) \frac{e^{ikR}}{R} = -4\pi\delta^{(3)}(\mathbf R).

Because

E−H0=ℏ22m(∇2+k2),E-H_0 = \frac{\hbar^2}{2m} (\nabla^2+k^2),

one obtains

(E−H0)G0+=ℏ22m(−m2πℏ2)(−4π)δ(3)=δ(3).\begin{aligned} (E-H_0)G_0^+ &= \frac{\hbar^2}{2m} \left( -\frac{m}{2\pi\hbar^2} \right) (-4\pi)\delta^{(3)}\\ &= \delta^{(3)}. \end{aligned}
  1. Show that the residue at a nondegenerate bound-state pole is the coordinate projector kernel.
Solution

Near z=Enz=E_n, separate the singular spectral term:

G(x,x′;z)=ψn(x)ψn∗(x′)z−En+Gregular(x,x′;z).G(x,x';z) = \frac{\psi_n(x)\psi_n^*(x')}{z-E_n} + G_{\mathrm{regular}}(x,x';z).

Multiplying by z−Enz-E_n and taking the limit gives

lim⁡z→En(z−En)G(x,x′;z)=ψn(x)ψn∗(x′)=⟨x∣∣n⟩⟨n∣∣x′⟩.\begin{aligned} \lim_{z\to E_n} (z-E_n)G(x,x';z) &= \psi_n(x)\psi_n^*(x')\\ &= \langle x\rvert \lvert n\rangle\langle n\rvert \lvert x'\rangle. \end{aligned}

For degeneracy, sum over an orthonormal basis of the eigenspace to obtain ⟨x∣Pn∣x′⟩\langle x\rvert P_n\lvert x'\rangle.

  1. Use the distribution identity to derive the discontinuity between outgoing and incoming resolvents.
Solution

For a scalar variable,

1x+i0−1x−i0=−2πiδ(x).\frac{1}{x+i0} - \frac{1}{x-i0} = -2\pi i\delta(x).

Apply this identity to each spectral value of HH through functional calculus:

R+(E)−R−(E)=1E−H+i0−1E−H−i0=−2πiδ(E−H).\begin{aligned} R^+(E)-R^-(E) &= \frac{1}{E-H+i0} - \frac{1}{E-H-i0}\\ &= -2\pi i\delta(E-H). \end{aligned}

Taking coordinate matrix elements gives the corresponding discontinuity of G+(x,x′;E)G^+(x,x';E) and G−(x,x′;E)G^-(x,x';E).