How entanglement harvesting works? Recently, I learned about entanglement harvesting from this video from the Institute of Quantum Information (Waterloo). So as I understand it you have two particles, let's consider two spins 1/2 particles that are very far from each other (space-like) and are in the same reference frame. Thus we can describe the state in a separable product
$|m_1\rangle\otimes|m_2\rangle$ where $m_1,m_2=\pm1/2$ (projection over some given axis).
If I understand entanglement harvesting correctly, you can entangle the states of the two spins by using the vacuum in between (no interactions), faster than if you used interactions (faster than the time it takes a photon to travel in between them). I guess it is not necessarily a maximally entangled state and not even a pure state.
What I do not understand is what is meant by the vacuum not being in its ground state even if there are no photons. I tried to read the associated paper E. Martín-Martínez,B.C. Sanders 2016 NewJPhys. but it provides a more complicated scheme. The original paper by Reznik (2003) does not help either as it considers accelerated detectors.
I do not understand if this can be triggered experimentally and what would be the implications of this faster-than-light action. I am guessing that the no-communication theorem still holds, so is it any better than usual entangling procedures?
Could somebody provide a simplified explanation on what are the steps and ingredients to produce entanglement harvesting (in particular related to the interaction between the vacuum and the spins)?
 A: The video doesn't say that the vacuum isn't in the ground state; it just says that the ground state isn't a tensor product of localized quantum harmonic oscillator states, which would be an unentangled state.
There is no faster-than-light action. The idea is that at the time you decide to "harvest" entanglement, it's already present, because the universe has been doing its own thing for a long time, not waiting for your decision. It is not intrinsically stranger than the fact that, for instance, aliens elsewhere in the Milky Way that measure the CMBR at a spacelike separated position from us will see a similar pattern. (If your two labs are so far apart that they actually have nonoverlapping past light cones all the way back to the beginning of time, then I suppose you couldn't get entangled results this way, but it isn't clear that such points can even exist.)
The two particles themselves are scarcely relevant, since you can always transfer information to them from whatever other form you may obtain it in, using what amounts to a CNOT gate. All that matters is whether there is some experiment you can do at spacelike separated points that yields correlated results. I think that detailed analysis of Unruh radiation is alleged in these papers to work. The detectors have to accelerate in a certain way, but the particles that you ultimately wish to entangle don't have to.
