On Peskin and Schroeder's QFT book page 282, the book considered functional quantization of scalar field.

First, begin with $$\left\langle\phi_b(\mathbf{x})\left|e^{-i H T}\right| \phi_a(\mathbf{x})\right\rangle=\int \mathcal{D} \phi \mathcal{D} \pi \exp \left[i \int_0^T d^4 x\left(\pi \dot{\phi}-\frac{1}{2} \pi^2-\frac{1}{2}(\nabla \phi)^2-V(\phi)\right)\right] $$ Since the exponent is quadratic in $\pi$, the book evaluates the $\mathcal{D}\pi$ integral and obtains $$\left\langle\phi_b(\mathbf{x})\left|e^{-i H T}\right| \phi_a(\mathbf{x})\right\rangle=\int \mathcal{D} \phi \exp \left[i \int_0^T d^4 x \mathcal{L}\right]. \tag{9.14} $$ There needs to be some complicated coefficients, but the book omit here.

I am puzzled how this integral finished? ie. $$\int \mathcal{D} \pi \exp \left[i \int _ { 0 } ^ { T } d ^ { 4 } x \left(\pi \dot{\phi}-\frac{1}{2} \pi^2\right)\right] $$ Since now the integral argument is $\pi$, which is a function, how to understand it's upper and lower limit? Also, their have a term $i \int_0^T d^4 x$ inside the exponent. So how to understand this integral?

  • 2
    $\begingroup$ Standard Gaussian integral - complete the square. $\endgroup$ Commented Sep 21, 2022 at 14:32
  • $\begingroup$ @SeanE.Lake I am familiar with the Standard Gaussian integral where the limit of $x$ goes from $-\infty$ to $\infty$. What I am puzzled is the $\mathcal{D}\pi$, it’s limit and other things as I point in the last paragraph of my post. Thanks! $\endgroup$
    – Daren
    Commented Sep 21, 2022 at 15:00
  • 2
    $\begingroup$ $\mathcal{D}\pi = \prod_x \mathrm{d}\pi(x)$ and the limits are all negative infinity to infinity. In other words, you're integrating over the space of all functions with support on the specified part of space-time (specified in the integral inside the exponential function). $\endgroup$ Commented Sep 21, 2022 at 15:04
  • $\begingroup$ I just want to clarify the integration should have $\sqrt{2\pi}$, is it dropped due to it is just a phase change? $\endgroup$
    – Li Chiyan
    Commented Dec 21, 2022 at 4:24
  • $\begingroup$ @LiChiyan Which integration do you mean? $\endgroup$
    – Daren
    Commented Dec 27, 2022 at 11:13

1 Answer 1


Briefly, there are 2 issues:

  1. It is safest to Wick rotate $t_E=it_M$ to make the Gaussian integrals exponentially damped rather than oscillatory. (NB: Don't also Wick rotate the momentum field $\pi_M=i\pi_E$, cf. my related Phys.SE answer here.)

  2. Truncate spacetime to a finite large box and discretize it. The result of the Gaussians integrations will be a number ${\cal N}$ that doesn't depend on any physically important parameters, but diverge in the continuum limit. Define the path integral measure ${\cal D}\pi$ to contain the reciprocal constant ${\cal N}^{-1}$.


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.