Can the quantum eraser experiment produce two different interference patterns?

[nomadreid]

In the explanation of the quantum eraser, Wikipedia https://en.wikipedia.org/wiki/Quantum_eraser_experiment states
"...the 'which-path' information is 'erased,' whereupon the interference pattern is recovered. (Rather than removing or reversing any changes introduced into the photon or its path, these experiments typically produce another change that obscures the markings earlier produced.)
A key result is that it does not matter whether the eraser procedure is done before or after the photons arrive at the detection screen."
Suppose I run the experiment twice:calling the two patterns K = which-path information known, and U = which-path information is unknown, and distinguishing between U1 and U2 as follows,
(1) one time, I keep the screen there the whole time, both while I am getting a pattern K, and then erasing, and ending up with interference pattern U1
(2) I withdraw the screen while I am erasing, so that on the screen is already imprinted a K pattern, and then put the second screen in its place, and get the interference pattern U2.
It would seem that the U1 and U2 are not the same, since I should be able to superimpose K onto U2 to get U1. However, the last sentence in that Wiki explanation seems to say that this is wrong, since otherwise I could do the screen switch between when the erasure took place and when the photons reached the screen, thereby making an "unauthorized" difference.
What I understand from that last sentence is that the difference between the pattern takes place at the moment of erasure , both at the point of erasure and the screen because the screen is part of the same system as the apparatus where the erasure takes place (entangled), but that it simply takes a bit of time from this quantum information to become classical information as in the case of quantum teleportation. But I am still confused about whether or not U1 and U2 would be the same, since that is classical information. Any clarification would be appreciated, thanks.

 

[andrewkirk]

A key result is that it does not matter whether the eraser procedure is done before or after the photons arrive at the detection screen.

I don't know what the writer of the wiki article means by this. There is no screen. There are only two detectors, one of which is moved around by a stepper motor. Perhaps the writer meant "before or after the photons arrive at the detector", but which one - the signal detector $D_s$ or the idler detector $D_p$? The erasure is done by the second polarising filter after each slit so, by definition, that can't happen to a photon after it reaches $D_s$.

So perhaps they mean $D_p$. But why should it be surprising that the pattern disappears even if the signal photon passes the erasure filter before the idler photon hits $D_p$?

It looks like your questions assume the existence of a screen with an imprinted pattern. There is no such screen in the experiment, just a bunch of time-location pairs for detections at $D_s$ and a bunch of detection times at $D_p$. Perhaps it will be easier to understand with the recognition that there is no screen, and no imprinted pattern.

 

[Cthugha]

This article is pretty bad. It seems to have been subject to several subedits, which in total make little sense. So before we start discussing details of the experiments, let us start about discussing the basics. The article states "A key result is that it does not matter whether the eraser procedure is done before or after the photons arrive at the detection screen." This is misleading.

Are you aware that there are two kinds of quantum eraser? Either you directly see an interference pattern when shining light through the double slit. Then you can put e.g. orthogonal polarizers or wave plates at the two slits. This yields which-way information and no interferenec pattern will be seen on any screen. Afterwards, you can place another polarizer in front of the detector, which is at 45 degrees to the two polarizations from the slits and thus let's them through with 50% probability and erases the which-way information. In this case, you will see an interference pattern on the screen. In this setting, there is no way to even erase something after the detection of the photons. You can only do it beforehand.

The kind of eraser experiments that allows for erasing the which-way information after detection of the signal photons - usually termed delayed erasure - relies on entangled photons, where you use or erase the path information on the entangled partner. This necessarily corresponds to a filtering process. In this case, you will only get interference patterns in the coincidence counts between the photons and never a pattern that is directly visible on any screen.

The article should really clarify that these are different experiments. In both cases, it never happens that there are patterns that show up on a screen and suddenly disappear and come back into existence.

 

[nomadreid]

Thanks for the clarifications, andrewkirk and Cthugha. I will go back and look at the arrangement again (and other sources) with these points in mind.