Recently, I noticed that the Dyson equation $$G=G_0+G_0\Sigma G$$ is used not only in quantum field theory but in some other branches of physics. For instance:

1. Wave equation

From the wave equation in a random media $$\mathcal{L}G(x,x_0)\equiv \left[\Delta_x+k^2 (1+\epsilon(x))\right]G(x,x_0)=\delta(x-x_0)$$ and $$\mathcal{L}_0G_0(x,x_0)\equiv\left[\Delta_x+k^2 \right]G_0(x,x_0)=\delta(x-x_0)$$ , we get $$\mathcal{L}_0G(x,x_0)=-k^2\epsilon(x)G(x,x_0)+\delta(x-x_0)$$ $$G(x,x_0)=\int dx_1\,\,G_0(x,x_1)\mathcal{L}_0G(x_1,x_0)=G_0(x,x_0)-\int dx_1\,G_0(x,x_1)k^2\epsilon(x_1)G(x_1,x_0)$$ , which obviously has the form of the Dyson equation.

2. Single-particle propagator - 1

Let $U(t,t_0)$ be the time evolution operator. The single particle propagator $$K(xt;x't')=\langle x|U(t,t_0)|x'\rangle$$ satisfies follwoing equations: $$\left[i\hbar\partial_t -\mathcal{H}\right]K(xt;x't')=\delta(x-x')\delta(t-t')$$ $$\left[i\hbar\partial_t -\mathcal{H}_0\right]K_0(xt;x't')=\delta(x-x')\delta(t-t')$$ where $\mathcal{H}=\mathcal{H}_0+V$. Those equations are mathematically identical to the wave equations above, i.e. $G\leftrightarrow K,\,-k^2\epsilon\leftrightarrow V$.

3. Single-particle propagator - 2

Using the Dyson series for $U$ in the interaction representation, $$U=1+\sum_{n=1}^{\infty}\frac{1}{n!}\left(-\frac{i}{\hbar}\right)^n\int\int\cdots \mathcal{T}\left[V(\tau_1)\cdots V(\tau_n)\right]$$ , we can also get the Dyson equation; strictly speaking, we can get the Dyson expansion of $K$ if we expand the Dyson equation in an iterative manner: $K=K_0+K_0\sum(K_0+K_0\sum(\cdots))$.

4. Green function in condensed matter physics

At zero temperature, the casual Green function is $$iG_{\alpha \alpha'}=\frac{\langle \psi_0|\mathcal{T}U(\infty,-\infty) \Psi_{\alpha}(x,t)\Psi_{\alpha'}^{\dagger}(x',t')|\psi_0 \rangle}{\langle \psi_0 |U(\infty,-\infty)|\psi_0 \rangle }$$ and it is not a Green function in a mathematical manner, i.e. $\left[i\hbar\partial_t -\mathcal{H}_0\right] iG_{\alpha \alpha'} \neq \delta$. Indeed, the unperturbed Green function $iG_{0,\alpha \alpha'}$ is a mathematical Green function. Using the Dyson expansion for $U$ in the interaction picture and the Wick's theorem, we can express $iG_{\alpha \alpha'}$ in terms of $V$ and $iG_{0,\alpha \alpha'}$. Such expression is known to satisfy the Dyson equation: $G=G_0+G_0\sum(G_0+G_0\sum(\cdots))$.


Obviously, the case 1 and 2 are closely related to each other. The case 3 and 4 are connected, either. However, I am not sure if there is a mathematical explanation about the reason that the Dyson expansion (case 3 & 4) leads to the result identical to the simple algebra (case 1 & 2).

I think the fact that the Dyson expansion produces not the Dyson equation itself $G=G_0+G_0\Sigma G$ but its iterative version $G=G_0+G_0\sum(G_0+G_0\sum(\cdots))$ would be a clue; but I have not made any progress so far.


1 Answer 1


Green's functions and example 1

First off, you are using the symbol $G$ and $K$ in 1 & 2 for Green's functions.

So let's define a Green's function first as $G(t,u)$ such as: $$ \mathcal{L}\,G(t,u) = \delta(t-u), $$ where $\mathcal{L}$ is a linear differential operator governing the evolution of the system.

The Green's function is used in solving the dynamics $x(t)$ caused by a source $f(u)$ by integrating over it: $$\mathcal{L}\,x(t) = \int \mathrm{d}u\,\mathcal{L}\,G(t,u)\, f(u) = \int \mathrm{d}u \ \delta(t-u)f(u) = f(t).$$

These are widely used in classical physics, such as in example 1.

Example 2 & Quantum mechanics

Quantum mechanics (first quantisation) is described by the Schrödinger equation, i.e. just a linear differential operator $\mathcal{L}$ like above. Similarly, then, one has:

$$ \phi(x,t_x) = \int \mathrm{d}y\, G^+(x,t_x,y,t_y) \phi(y,t_y), $$

where the Green's function $G^+$ propagates the particle, described by a wavefunction $\phi$, from position $y$ and time $t_y$ to a position $x$ at time $t_x$. The $+$ sign is to prevent particles from travelling back in time, so technically $G^+ = \theta(t_x-t_y)G$ is the retarted Green's function.

Hence the connection between a Green's function and a propagator in quantum mechanics. Generally, the propagator $G^+$ can be written: $$ G^+(x,t_x,y,t_y) = \theta(t_x-t_y)\, \langle x(t_x) | y(t_y)\rangle, $$ i.e. the probability amplitude that a particle in state $|y\rangle$ at time $t_y$ ends ip in state $|x\rangle$ at time $t_x$.

The fact that the propagator is non-zero for causally disconnected events, i.e. events that are space-like separated, is the failure of first quantisation and the need of second quantisation (quantum field theory)

Dyson's equation

Green's function in the energy domain can be written as:

$$G^+(x,y,E) = \sum_n \frac{i\phi_n(x) \, \phi_n^\ast (y)}{E-E_n} \propto \frac{1}{E-H_0}. $$

easily justified by taking the Laplace equation for the electric potential $\phi$ in vacuo $\nabla^2\phi = 0$. The Green's function solution is $\epsilon_0 \nabla^2V(\mathbf{r}) = - \delta^{(3)}(\mathbf{r}-\mathbf{r}_0)$, to which you know the solution to be a point charge with $V(\mathbf{r}) = 1/|\mathbf{r}-\mathbf{r}_0|$

Green's functions allow us to interpret a perturbation problem in terms of a propagating particle. A perturbation interrupts the propagation via a scattering process, following which free-particle propagation resumes.

A generic Hamiltonian $H = H_0 + V$ is decomposed in to its solvable, free-particle propagator-giving solution $H_0$ and in a non-analytically solvable perturbation $V$.

The Green's function is given by $$G \propto \frac{1}{E-H} = \frac{1}{E-H_0 -V},$$ which would be called the full propagator to distinguish it from the free propagator $G_0 \propto 1/(E-H_0)$.

In a perturbative approach, where $V \ll H_0$, the expression above can be written as:

$$ G = \frac{1}{E-H_0-V} = \frac{1}{E-H_0} + \frac{1}{E-H_0}V\frac{1}{E-V_0}+\frac{1}{E-H_0}V\frac{1}{E-V_0}V\frac{1}{E-V_0} + \dots $$ or $$G = G_0 + G_0 V G_0 + G_0 V G_0 V G_0 + \dots $$

which is a geometric series, rewritten as:

$$ G = G_0(1 + VG_0 + VG_0VG_0 + \dots) = \frac{G_0}{1-V G_0} = \frac{1}{G_0^{-1}-V},$$

which is known as Dyson's equation.

Example 3

The Schrödinger equation for a generic Hamiltonian $H$ is:

$$ i\hbar \frac{\mathrm{d}}{\mathrm{d}}|\psi_t\rangle = H |\psi_{t_0}\rangle.$$ $$ \implies |\psi_t\rangle = U(t,t_0) |\psi_{t_0}\rangle, $$

where $U$ is the time-evolution operator.

In the interaction picture, the Hamiltonian is split between its free part $H_0$ and interacting part $H'$ such that the time-evolution of the interaction picture Haimltonian $H_I$ is given by $H_I(t) = e^{iH_0 t/\hbar}H'e^{-iH_0t/\hbar}$.

The time evolution operator in this picture then becomes $U_I = e^{iH_0t/\hbar}Ue^{-iH_0t/\hbar}, and satisfies the following equation of motion:

$$ i\hbar \frac{\mathrm{d}}{\mathrm{d}t}U_I = H_I U_I. $$

The solution to the above is : $$ U_I(t,t_0) = 1 - \frac{i}{\hbar} \int_{t_0}^t \mathrm{d}t' H_I(t', t_0)\,U_I(t', t_0) ,$$

and it can be iterated by keeping plugging in the expression for the $U_I$ in the integral:

$$ U_I(t,t_0) = 1 - \frac{i}{\hbar} \int_{t_0}^t \mathrm{d}t' H_I(t', t_0)\ + \left (-\frac{i}{\hbar} \right )^2 \int_{t_0}^t \mathrm{d}t'' \int_{t_0}^t \mathrm{d}t''' H_I(t'', t_0)\,H_I(t''', t_0) +\dots $$

Going back to $U = e^{-iH_0t}U_Ie^{iH_0t}$, idenfitying $G_0 = e^{iH_0t}$ and $H_I = V$, we get the Dyson's equation as per the point above.

Hence $U$ is also called the Dyson's operator and the perturbative expansion known as Dyson series/expansion.

The time ordering operator $T$ should also be added.

Quantum field theory & example 4

While the first quantisation propagator is the same as a Green's function, in second quantisation the propagator is only termed as such by analogy.

Indeed: $$G^+(x,y) = \theta(x^0 - y^0) \langle \Omega| \hat{\phi}(x)\hat{\phi}^\dagger (y) |\Omega \rangle, $$ where $|\Omega\rangle$ is the interacting vacuum, $x$ and $y$ are now four-vectors in spacetime, and $\hat{\phi^\dagger}(y)$ is a field operator creating a particle at $(y^0, \mathbf{y})$.

Your expression in just obtained by requiring the overlap $A$ between an incoming and outcoming state to be expressed in terms of the free, non-interacting states: $$ A = _{\mathrm{interacting}}\langle q|p\rangle_{\mathrm{interacting}} = _{\mathrm{free}}\langle q|S|p\rangle_{\mathrm{free}}, $$ with $S$ being the scattering matrix.

The scattering matrix $S$ is related to the evolution operator $U$ above by $S = \lim_{t\rightarrow \infty, t_0 \rightarrow -\infty} U(t,t_0)$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.