Skip to content

Green Function Table

This table collects common Green-function conventions used in quantum dynamics. A Green function is an inverse kernel plus a boundary prescription. The same local operator can have retarded, advanced, outgoing, incoming, time-ordered, Euclidean, or finite-temperature Green functions.

The conventions below match the Dynamics pages:

R(z)=(z−H)−1,R(z) = (z-H)^{-1},

and, for time-independent HH,

GR(t)=−iℏθ(t)e−iHt/ℏ,GA(t)=iℏθ(−t)e−iHt/ℏ.G^R(t) = -\frac{i}{\hbar}\theta(t)e^{-iHt/\hbar}, \qquad G^A(t) = \frac{i}{\hbar}\theta(-t)e^{-iHt/\hbar}.

The energy transform convention is

G(E)=∫−∞∞dt eiEt/ℏG(t).G(E) = \int_{-\infty}^{\infty}dt\, e^{iEt/\hbar}G(t).

With different Fourier conventions, factors of 2π2\pi, ℏ\hbar, and signs of i0i0 may move. Always translate conventions before comparing formulas from different sources.

ObjectFormulaPrescriptionMain use
ResolventR(z)=(z−H)−1R(z)=(z-H)^{-1}z∉σ(H)z\notin\sigma(H)spectral inverse
Retarded resolventGR(E)=(E−H+i0)−1G^R(E)=(E-H+i0)^{-1}upper half-planecausal response
Advanced resolventGA(E)=(E−H−i0)−1G^A(E)=(E-H-i0)^{-1}lower half-planecomplementary boundary value
Spectral functionA(E)=i(GR−GA)A(E)=i(G^R-G^A)real-axis discontinuityspectral weight
Density of statesρ(E)=−(1/π)Im⁡Tr⁡GR(E)\rho(E)=-(1/\pi)\operatorname{Im}\operatorname{Tr}G^R(E)trace and retarded signstate counting
Coordinate Green function$G(x,x’;z)=\langle x(z-H)^{-1}x’\rangle$

The resolvent is

R(z)=(z−H)−1,R(z) = (z-H)^{-1},

where zz lies away from the spectrum of HH. It satisfies

(z−H)R(z)=I.(z-H)R(z)=I.

For a discrete nondegenerate spectrum,

R(z)=∑n∣n⟩⟨n∣z−En.R(z) = \sum_n \frac{\lvert n\rangle\langle n\rvert}{z-E_n}.

The poles locate eigenvalues, and the residues give projectors onto eigenspaces. For continuous spectra, the real-axis structure is usually read through boundary values.

The retarded and advanced energy-domain Green functions are

GR(E)=lim⁡η→0+1E−H+iη,G^R(E) = \lim_{\eta\to0^+} \frac{1}{E-H+i\eta},

and

GA(E)=lim⁡η→0+1E−H−iη.G^A(E) = \lim_{\eta\to0^+} \frac{1}{E-H-i\eta}.

The sign of i0i0 records the time-support condition:

ObjectTime supportEnergy boundary value
retardedzero for t<0t\lt0E+i0E+i0
advancedzero for t>0t\gt0E−i0E-i0

For self-adjoint HH,

GA(E)=(GR(E))†G^A(E) = \bigl(G^R(E)\bigr)^\dagger

as a boundary-value statement when the relevant limits exist.

The distribution identity

1x+i0=PV⁡1x−iπδ(x)\frac{1}{x+i0} = \operatorname{PV}\frac{1}{x} - i\pi\delta(x)

implies

GR(E)−GA(E)=−2πi δ(E−H).G^R(E)-G^A(E) = -2\pi i\,\delta(E-H).

Thus the spectral function convention

A(E)=i(GR(E)−GA(E))A(E) = i\bigl(G^R(E)-G^A(E)\bigr)

gives

A(E)=2π δ(E−H).A(E) = 2\pi\,\delta(E-H).

Some many-body texts define spectral functions with additional signs, factors, matrix elements, or 1/(2π)1/(2\pi) normalizations. The safest habit is to state both GR−GAG^R-G^A and the chosen definition of AA.

The total density of states is

ρ(E)=Tr⁡δ(E−H).\rho(E) = \operatorname{Tr}\delta(E-H).

Using the retarded resolvent,

ρ(E)=−1πIm⁡Tr⁡GR(E).\rho(E) = -\frac{1}{\pi} \operatorname{Im} \operatorname{Tr}G^R(E).

The local density of states in coordinate representation is

ρ(x,E)=−1πIm⁡GR(x,x;E),\rho(x,E) = -\frac{1}{\pi} \operatorname{Im} G^R(x,x;E),

with the trace or integral over xx recovering the total density when the normalization is appropriate.

The coordinate-space energy Green function, derived in Energy Green Function, is

G(x,x′;z)=⟨x∣(z−H)−1∣x′⟩.G(x,x';z) = \langle x|(z-H)^{-1}|x'\rangle.

It solves

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

with boundary conditions determined by zz and the Hamiltonian domain. For example, the same differential expression on the line, a half-line, a box, or a ring gives different Green functions.

This is the energy-domain inverse-kernel analogue of the propagator table’s warning: boundary conditions are part of the object.

For a driven equation

(iℏ∂∂t−H)ψ(t)=η(t),\left( i\hbar\frac{\partial}{\partial t} -H \right)\psi(t) = \eta(t),

a retarded solution is

ψR(t)=∫dt′ GR(t−t′)η(t′).\psi_R(t) = \int dt'\, G^R(t-t')\eta(t').

Since GR(t−t′)=0G^R(t-t')=0 for t<t′t\lt t', the response at tt depends only on earlier source values. This is the causal-response meaning of the retarded prescription.

The advanced Green function solves the same local equation but obeys the opposite support condition. It is a useful analytic boundary value, not the usual initial-value response.

For a one-dimensional free particle,

H=p22m,E=ℏ2k22m,k>0,H = \frac{p^2}{2m}, \qquad E = \frac{\hbar^2k^2}{2m}, \qquad k\gt0,

the outgoing boundary value is

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

The incoming boundary value is obtained by k→−kk\to -k in the phase:

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

In three dimensions, with R=∣r−r′∣R=|\mathbf r-\mathbf r'|,

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

These are Green functions of E−H±i0E-H\pm i0, not time-evolution kernels. They solve Helmholtz-type inverse problems with outgoing or incoming boundary conditions.

The time-evolution kernel is

K(xf,tf;xi,ti)=⟨xf∣e−iH(tf−ti)/ℏ∣xi⟩.K(x_f,t_f;x_i,t_i) = \langle x_f|e^{-iH(t_f-t_i)/\hbar}|x_i\rangle.

The resolvent is related to time evolution by a one-sided transform. For Im⁡z>0\operatorname{Im}z\gt0,

R(z)=1iℏ∫0∞dt eizt/ℏe−iHt/ℏ.R(z) = \frac{1}{i\hbar} \int_0^\infty dt\, e^{izt/\hbar} e^{-iHt/\hbar}.

Thus propagator kernels and Green functions are related, but they are not interchangeable. The propagator evolves initial data in time; the Green function inverts an operator with a boundary prescription.

  • Saying “the Green function” without specifying the operator and boundary prescription.
  • Confusing K(t)K(t) with (E−H+i0)−1(E-H+i0)^{-1}.
  • Dropping the i0i0 sign when taking imaginary parts.
  • Mixing retarded, advanced, outgoing, incoming, and time-ordered prescriptions.
  • Forgetting Fourier-transform conventions when comparing signs and factors of ii.
  • Treating local density of states as a total density without tracing or integrating.
  • Using free-space Green functions in bounded domains without imposing boundary conditions.
  • E. N. Economou, Green’s Functions in Quantum Physics, 3rd ed., Springer, 2006.
  • G. D. Mahan, Many-Particle Physics, 3rd ed., Kluwer Academic/Plenum, 2000.
  • A. L. Fetter and J. D. Walecka, Quantum Theory of Many-Particle Systems, Dover, 2003.
  • J. R. Taylor, Scattering Theory: The Quantum Theory of Nonrelativistic Collisions, Dover, 2006.
  • R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
  • M. Reed and B. Simon, Methods of Modern Mathematical Physics I: Functional Analysis, revised and enlarged ed., Academic Press, 1980.
  1. Derive the spectral function identity from the retarded and advanced resolvents.
Solution

Using

1x+i0−1x−i0=−2πi δ(x),\frac{1}{x+i0} - \frac{1}{x-i0} = -2\pi i\,\delta(x),

and applying the identity through the spectral theorem with x=E−Hx=E-H, one gets

GR(E)−GA(E)=−2πi δ(E−H).G^R(E)-G^A(E) = -2\pi i\,\delta(E-H).

Therefore, with A(E)=i(GR−GA)A(E)=i(G^R-G^A),

A(E)=2π δ(E−H).A(E) = 2\pi\,\delta(E-H).
  1. Show that the density-of-states formula follows from the spectral function.
Solution

The retarded imaginary-part identity is

−1πIm⁡GR(E)=δ(E−H).-\frac{1}{\pi}\operatorname{Im}G^R(E) = \delta(E-H).

Taking the trace gives

−1πIm⁡Tr⁡GR(E)=Tr⁡δ(E−H)=ρ(E).-\frac{1}{\pi} \operatorname{Im} \operatorname{Tr}G^R(E) = \operatorname{Tr}\delta(E-H) = \rho(E).
  1. Verify the one-dimensional outgoing free Green function away from x=x′x=x'.
Solution

For x≠x′x\ne x',

G+(x,x′;E)=−imℏ2keik∣x−x′∣G^+(x,x';E) = -\frac{im}{\hbar^2k} e^{ik|x-x'|}

is proportional to a solution of

(d2dx2+k2)G=0.\left( \frac{d^2}{dx^2}+k^2 \right)G=0.

Since

E−H=ℏ22m(d2dx2+k2),E-H = \frac{\hbar^2}{2m} \left( \frac{d^2}{dx^2}+k^2 \right),

the equation (E−H)G=0(E-H)G=0 holds away from the source. The derivative jump at x=x′x=x' supplies the delta function, and the phase eik∣x−x′∣e^{ik|x-x'|} gives outgoing waves on both sides.

  1. Why can two Green functions solve the same differential equation but represent different physics?
Solution

The local differential equation fixes only the inverse relation away from boundary data. The Green function also includes a boundary or support prescription. Retarded and advanced Green functions solve the same local equation but have opposite time support. Outgoing and incoming Green functions solve the same Helmholtz-type equation but impose different behavior at infinity. The prescription states the physical problem.