0
$\begingroup$

In perturbative quantum field theory, one starts with interaction Lagrangian density

$$\mathcal{L} = \mathcal{L}_{\text{free}}+g\mathcal{L}_{\text{I}}\tag{1}$$

Where $g$ is coupling strength. Generating functional of the theory with background field $J$ is then given as

$$Z[J] = \int \mathcal{D}\phi \; e^{i \int d^kx \, (\mathcal{L}_{\text{free}}+g\mathcal{L}_{\text{I}})+\int d^kx J\phi}.\tag{2}$$

Taylor expanding the interaction amplitude in coupling constant $g$

$$e^{i g\int d^4x \mathcal{L}_{\text{I}}} = \sum^{\infty}_{n= 0}\frac{(ig)^n}{n!}\biggr(\int \mathcal{L}_{\text{I}}d^ky\biggr)^n = \sum^{\infty}_{n = 0}\frac{(ig)^n}{n!}\biggr(\int \mathcal{L}_{\text{I}}(y_1)\cdots \mathcal{L}_{\text{I}}(y_n)d^4y_1\cdots d^4y_n\biggr).\tag{3}$$

Then one obtains

$$\begin{align}Z[J] &= \int \mathcal{D}\phi \; e^{i \int d^kx \, \mathcal{L}_{\text{free}}+\int d^ky J\phi}\biggr(1+ig\int \mathcal{L}_{\text{I}}(y)d^ky_1+\mathcal{O}(g^2)\biggr)\tag{4}\\ &= \underbrace{\int \mathcal{D}\phi \; e^{i \int d^kx \, \mathcal{L}_{\text{free}}+\int d^ky J\phi}}_{ = Z_0[J]} + ig \int d^k y\int \mathcal{D}\phi \biggr(\mathcal{L}_{\text{I}}(y)\biggr)\; e^{i \int d^kx \, \mathcal{L}_{\text{free}}+\int d^kx J\phi} + O(g^2)\tag{5}\end{align}$$

Functional derivative of which leads us to

$$\begin{align}\frac{1}{Z[0]}\frac{\delta^n Z[J]}{\delta J(x_1)\cdots\delta J(x_n)}\biggr\vert_{J = 0} &= \langle \mathcal{T}\{\phi(x_1)\cdots \phi(x_n)\}\rangle_{\text{free}}-g\int d^ky \langle \mathcal{T}\{\phi(x_1)\cdots \phi(x_n)\mathcal{L}_{\text{I}}(y)\}\rangle_{\text{free}}+\mathcal{O}(g^2)\end{align}\tag{6}$$

Is this how we perturbatively compute the correlation functions of the interacting theory?

$\endgroup$
0

2 Answers 2

1
$\begingroup$

I'm not an expert (I'm a student myself) but I think I can give a quick answer.

Your expansion in powers of $g$ look good so yes this is a perturbative expansion for the interacting correlation function. Nevertheless, we usually don't stop here. If $\cal{L}$ is made of terms like $\phi^k (\partial \phi)^m$ you can see that the expectation values you need to compute are actually free correlation functions or derivatives of free correlation functions calculated with $(k + m)$ points identified with the integration variable, i.e., with only one point. This identification of points in the free correlation functions turns the trivial Feynman diagrams of the free theory in interesting diagrams where the integration variables are internal points and the $\{x_i\}$ variables are the external points. This is how the perturbative expansion is usually understood.

$\endgroup$
1
  • $\begingroup$ Addition: when you identify multiple points in the free correlation function with a single internal point many diagrams of the free theory become the same diagram of the interacting theory, this is what gives rise to the "combinatorial factor" (which is only part of the total prefactor of the diagram, the rest comes from the fact that internal points are mute variables and perhaps also from numerical factors present in $\cal{L}$ itself) $\endgroup$ Commented Oct 16 at 10:06
1
$\begingroup$

QFT is a vast topic. OP's perturbative expansion (6) is not wrong per se but ineffective because of a bunch of disconnected terms. Instead, perturbative calculations are typically organized as follows:

  1. Firstly, correlation functions are broken into connected correlation functions, which in turn are broken into 1PI parts.

    It is therefore enough to calculate 1PI correlation functions.

  2. Secondly, there is the issue of regularization & renormalization.

$\endgroup$

Your Answer

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

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.