So the transition amplitude for a free Klein-Gordon field for a space-like interval is finite and non-vanishing (decays exponentially). What does one make of this physically, i.e. what is the meaning of a quantum field leaking out of the light cone?

  • 2
    $\begingroup$ See this and links therein, e.g. this. Your question for sure has been asked before here. I again advice you to use the search function. $\endgroup$ Commented May 15 at 18:37
  • $\begingroup$ @TobiasFünke I have got the search bar working now. However, The links you provide talk about causality. I am not concerned with whether or not QFT is causal, but rather with the fact that quantum fields have non-zero probabilities for exiting outside the light cone. $\endgroup$ Commented May 15 at 19:16
  • 1
    $\begingroup$ See this paper for the causality of the Dirac field: Phys. Rev. A 107, 032209 (2023) $\endgroup$
    – Quillo
    Commented May 15 at 23:06
  • $\begingroup$ @Quillo Yes, I was indeed thinking that any correlations between space-like events would have to be "subluminal" in nature..as the commutator vanishes. Thanks for that reference I will check it out. $\endgroup$ Commented May 15 at 23:10

1 Answer 1


The posts linked in the comments give a good explanation of why the microcausality condition $[\phi(x),\phi^\dagger(y)]=0$ for spacelike $x-y$ is sufficient to make sure that only causal signals can propagate. But there are a few ways to think about the physical meaning of the exponential decay in the correlation function $\langle\phi(x)\phi^\dagger(y)\rangle$.

The first way is that even if $x,y$ are spacelike separated, they have a common history. That is, there is another point $z$ in the past of both $x,y$ such that $z-x$ and $z,y$ are timelike. So what happens at $z$ affects what happens at both $x,y$, and so the relevant correlation function is non-zero.

The second way is to realize that in the vacuum state $|0\rangle$, spatial subregions are highly entangled. If you try to split space into two regions $A,B$ and compute the entropy $S(\rho_A) = -\text{Tr}_A[\rho_A \ln \rho_A]$, where $\rho_A = \text{Tr}_B[|0\rangle\langle 0 |]$, you actually find that this is infinite, so the vacuum is extremely entangled. The technical way to phrase this is that observables in continuum QFT form a Type III von Neumann algebra. If spacelike regions were unentangled, then we actually would have that $$\langle0 |\phi(x)\phi^\dagger(y)|0\rangle = \langle 0 |\phi(x) |0\rangle\langle 0|\phi^\dagger(y) | 0 \rangle = 0$$

A third way to think about it is that the Hamiltonian density has a term in it that goes like $\mathcal{H} \sim |\nabla \phi|^2$, so it must be that $\phi(x)$ is correlated at spacelike separation in the vacuum. If this wasn't the case, then there could be sharp jumps in the profile of $\phi(x)$ in the vacuum, and this would produce a large energy. This is very related to the previous point about vacuum entanglement.

Finally, another perspective is that spacelike correlations of a field don't come from one particle, but from multiple identical particles. You might measure a particle at $x$ and your friend might measure another at $y$ to give a non zero answer for $\langle \phi(x)\phi(y)\rangle$, but that doesn't mean they were the same particle. It only says that they are indistinguishable, and come from excitations of the same field.


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