# Convergence of Feynman path integral in QFT

In Greiner field quantization book, when discussing the Feynman path integral approach, the book tries to calculate the path integral

$$\tag{12.35} \int \mathcal{D}\phi \exp\bigg[\frac{i}{\hbar}\int d^4x \bigg(\frac{\hbar^2}{2}\partial_\mu\phi\partial^\mu\phi-\frac{1}{2}m\phi^2+J\phi\bigg)\bigg].$$

The book then proceeds with the standard way of how physics textbooks would do this: they treat $$\mathcal{D}\phi$$ as if it were just the usual Riemann integral $$d\phi$$, with $$\phi$$ treated as a single dummy variable. The result is the usual Gaussian integral.

However, at the beginning of chapter 12, we defined the precise meaning of $$\mathcal{D}\phi$$: first slice spacetime into a set of $$M$$ "elementary cells" of volume $$\Delta V$$, which for simplicity will be taken of equal size, centered at the coordinates $$x_l, l=1,...,M$$. In this way the continuous field function $$\phi(x,t)$$ is made into a finite dimensional veector $$\phi_l(t)=\phi(x_l,t)$$ with discrete index $$l$$ etc." (page 366 has more details on what Greiner means.)

My question is: how do we calculate 12.35 using this precise definition? I.e. how to show convergence of the prescription above to the nonrigorous result obtained by treating the Feynman path integral as a Riemann integral.

$$\textbf{EDIT:}$$ I now realized that Rothe's lattice gauge theory book chapter 3 has an explicit calculation using the prescription above, for the scalar field case. This appears to be exactly the kind of calculation that I was looking for.

• it does not converge :) Oct 26, 2022 at 21:50
• Can you give a reference? I would expect in the Wick rotated case it should converge though. Oct 26, 2022 at 21:55

Greiner is in the beginning of chapter 12 describing a heuristic discretization (rather than a precise mathematical meaning) of the Feynman path integral. A rigorous mathematical definition of path integrals is a huge topic, cf. e.g. this and this Phys.SE posts. For starters Greiner should include an $$i\epsilon$$ prescription for convergence. Discretization is mainly used in numerical work rather than in analytical calculations.