I Have You Heard of the SEW Experiment? Thoughts?

[ChadGPT]

TL;DR Summary

Have you heard of the SEW? Do you find it significant?

The SEW Experiment, named after its authors, Marian 0. Scully, Berthold-Georg Englert &Herbert Walther, was published in 1991 under the title, "Quantum optical tests of complementarity," and can be found here: https://www.nature.com/articles/351111a0.pdf

They built atom interferometers with detectors and used caesium atoms instead of photons or electrons. A couple of well defined beams of caesium atoms are sent towards a double slit, eventually to end up at a back screen. Before each slit are micromaser cavities that the beams must pass through. A laser beam can be turned on or off to excite the caesium atoms into a higher energy level, such that when an atom enters a micromaser cavity in an excited state it will emit a photon which is then stored in the micromaser cavity. Thus, which-way information can be obtained by reading out whether the photon is in cavity 1 or cavity 2.

If the laser is on, they got a particle pattern. If the laser is off, they got an interference pattern.

In a second configuration the two cavities have a common internal wall covered by a thin film semiconductor which absorbs photons and acts as a photo detector, but a pair of electro-optical shutters are also placed in front of the detector wall to keep the two cavities separated. With the shutters open, any photons that end up in the cavities will be absorbed by the detector wall, and thus there is no which path information. With the shutters closed, the photons will end up in either cavity 1 or cavity 2, and which-path information can be obtained by reading out which.

When they run the experiment with the shutters open and the laser on, they got an interference pattern. When they run the experiment with the shutters closed and the laser on, they got a particle pattern. If they waited until after the experiment was run and the particle pattern was already accumulated on the back screen, and then subsequently opened the shutters to erase the which-path information, they still got the particle pattern.

However, they were able to recover the interference pattern after the fact by correlating detections by the detecting wall with back screen hits, and not detections with back screen hits. Apparently the detector wall only absorbs photons 50% of the time when the shutters are open, and the other 50% of the time the photons just bounce around in both cavities, where it is still impossible to determine which-path information. When they correlated detections by the detector wall they produced a fringe interference pattern, and when they correlated no detections by the detector wall they produced an anti-fringe interference pattern (pi phase shift). When added together they produce the particle pattern jointly, as seen on the back screen.

I find this experiment interesting because at first it seems like something about the laser being on might have something to do with the loss of the interference pattern. Then it turns out erasing the possibility of which-path information despite the laser beam being on reproduces the interference pattern.

It's also interesting because it shows there is no retrocausality in DCQE in a much clearer way than the more popular Kim et al. 2001 experiment. Here we see clearly an interference pattern if the choice is made before the atoms reach the back screen, and never an interference pattern if the choice is delayed until after the atoms have reached the back screen.

Though, there is still the strangeness about recovering the interference patterns after the fact using correlations, and one of the interference patterns being anti-fringe pi phase shifted.

Thoughts?

 

[DrChinese]

This article is behind a paywall. However, there is an excellent discussion of the paper in the following. See Figure 8 and related discussion on page 8.

Delayed-choice gedanken experiments and their realizations

Note that according to my reference, the SEW paper was a proposed (gedanken) experiment. I do not know if it was ever executed. However, there are literally dozens of Delayed Choice Quantum Eraser (DCQE) and Delayed Choice Entanglement Swapping (DCES) experiments that have been executed in the past 25 years.

They all have results that would be considered as "consistent" with retrocausality. That does not necessarily mean there is actual retrocausality, because that is something that is interpretation dependent. What is generally agreed is that the order of execution of observation steps in any* quantum experiment is not relevant to the quantum expectation value. As the order is varied, the results remains (apparently) the same. This renders classical notions of cause-effect relationships untenable.


*As @PeterDonis correctly points out, there are a few classes of setups where measurement order is baked in. So excluding those...

 

[PeterDonis]

What is generally agreed is that the order of execution of observation steps in any quantum experiment is not relevant to the quantum expectation value.

I don't think this is true of any quantum experiment. (For example, consider a pair of spin measurements of the same particle along different spin axes.) But it is true of the class of experiments under discussion, which never involve more than one spin measurement on a given particle.

 

[Cthugha]

TL;DR Summary: Have you heard of the SEW? Do you find it significant?

The SEW Experiment, named after its authors, Marian 0. Scully, Berthold-Georg Englert &Herbert Walther, was published in 1991 under the title, "Quantum optical tests of complementarity," and can be found here: https://www.nature.com/articles/351111a0.pdf

They built atom interferometers with detectors and used caesium atoms instead of photons or electrons. A couple of well defined beams of caesium atoms are sent towards a double slit, eventually to end up at a back screen. Before each slit are micromaser cavities that the beams must pass through. A laser beam can be turned on or off to excite the caesium atoms into a higher energy level, such that when an atom enters a micromaser cavity in an excited state it will emit a photon which is then stored in the micromaser cavity. Thus, which-way information can be obtained by reading out whether the photon is in cavity 1 or cavity 2.

[...]
I find this experiment interesting because at first it seems like something about the laser being on might have something to do with the loss of the interference pattern. Then it turns out erasing the possibility of which-path information despite the laser beam being on reproduces the interference pattern.

It's also interesting because it shows there is no retrocausality in DCQE in a much clearer way than the more popular Kim et al. 2001 experiment. Here we see clearly an interference pattern if the choice is made before the atoms reach the back screen, and never an interference pattern if the choice is delayed until after the atoms have reached the back screen.

You seem to be missing one very important point: This is a review article (not a regular one) and they never did these experiments. As pointed out by DrChinese these are just "gedankenexperiments", which were never actually performed at that time. They just present the results expected from calculations. The only experimental data is figure 2 - the interference pattern produced by slow neutrons in a double slit. If I remember correctly, later on some somewhat related experiments with Rydberg atoms were done, but not exactly in the way suggested by Scully. I might be wrong, though. I did not follow the literature too closely in recent years. However, to the best of my knowledge people used other implementations because Rydberg atoms are incredibly sensitive to external fields and it was close to impossible to keep the surfaces of the cavities free from any remaining charges. The interaction between these harges and the Rydberg atoms is an effective which-way measurement.

The whole rationale of the Kim et al. experiment was to map this proposed experiment onto something which could actually be realized in a lab. Therefore, I do not really agree that this is clearer than the Kim et al. experiment because - well - the latter experiment has actually been performed.