More thoughts on quantum eraser experiments

[nrqed]

Patrick posted the following interesting link in another thread:

...

http://grad.physics.sunysb.edu/~amarch/



cheers,
patrick.

I'd like to discuss this some more. Whenever I have read about these experiments I have always hoped to see the results of one in particular but I have never found what I wanted. I can guess what the results would be, but maybe some would disagree.

An obvious extension of the third experiment, the delayed erasure, would be the following:

let photon s hit the screen first. *Then* (after the impactof s) randomly decide whether a polarizer will be inserted in the path of photon p. This way, the choice of inserting the polarizer in the path of p is made *after* photon s has hit the screen.

Even better: make the path of beam p so long that even a signal at the speed of light could not get from s to p to exchange any info between those two.

Now, I can guess what the result would be (I can't see any other possibility unless completely new physics would come into play but maybe some will disagree):

I think that the pattern on the screen will not be a clear interference or non-interference pattern. It will be more messy. But then, if someone separately plot the impacts of s corresponding to the case where the polarizer was in the path of p and the impacts of s corresponding to when the polarizer was absent, one would see a non-interference and an interference pattern, respectively. I can't think of anything else that could happen (unless, again, something drastically new emerges).

I would find this an interesting experiment to do. I know that some would say that if we accept the experiments testing Bell's inequality, we already know about the nonlocality aspect of QM and that my experiment does not bring anything new. Still, I think it would be an interesting different check. And one that would clearly show not only a nonlocality in space but also one in time (the s photon would not only know what happens very far, but also what happens in the future!). I know that some people will say (I'm used to having my ideas shot down, as you see ) that it's obvious that a nonlocality in space leads to a nonlocality in time (we just have to view a n experiment testing Bell's inequality in a different Lorentz frame) but still, those are all theoretical considerations. We should do experiment and let Nature tells us what is correct. And the experiment I am thinking about here offers a new twist, I would think.

Of course, I am sure it's not a new idea, so I would also like to have reference to this type of experiment, if anyone knows.

Cheers

Pat

 

[vanesch]

I think that the pattern on the screen will not be a clear interference or non-interference pattern. It will be more messy. But then, if someone separately plot the impacts of s corresponding to the case where the polarizer was in the path of p and the impacts of s corresponding to when the polarizer was absent, one would see a non-interference and an interference pattern, respectively. I can't think of anything else that could happen (unless, again, something drastically new emerges).

In fact, the experiment which I linked up is, if you think about it, tricking you. It is (classical) subset selection. Look at the double slit experiment with different QW plates: clearly left-handed photons are not going to interfere with right-handed photons, so it is normal that you only get a blob when looking at ALL the photons. I also guess you get that same blob after having introduced the polarizer in p *if you do not look for coincidences* but simply look at all the s-photons. It is only when you select the subset of photons that get through the 45 degree polarizer that you get a subset of s-photons which interfere.
Due to the correlation between the photons, the 45-degree polarizer in fact (true, this is QM at its best !) makes us have also 45-degree s-photons, which are NOT altered through the QW plates and which DO interfere. It is just by writing out:

|psi> = 1/sqrt(2) (|x>|y> + |y>|x>) = 1/2 1/sqrt(2) ( [|45->+|45+>]x[- |45-> + |45+>] + [|45-> + |45+>]x[|45->+|45+>])
= 1/sqrt(2) (|45+>|45+> - |45->|45->)

which shows that we can think of our original entangled xy and yx state also as an entangled state with polarization +45 for both photons or -45 for both photons.
So subselecting the p photon under +45 degree selects only +45 degree photons for the s-photon, and hence they pass as +45 degree photons through the QW plates and the slits ; this means interference.

As the QW plate introduces different indices of refraction for +45 and -45 degree photons, this SHIFTS the interference pattern (contrary to what is said in the text!) so the interference pattern of the +45 degree photons together with the SHIFTED pattern of the -45 degree photons make up the blurry blob when we count all the photons.

Much ado for nothing special in fact.

A true Bell type experiment is much more mysterious. This one just *looks* mysterious, but consists of subselecting two shifted interference patterns.

cheers,
Patrick.

 

[vanesch]

|psi> = 1/sqrt(2) (|x>|y> + |y>|x>) = 1/2 1/sqrt(2) ( [|45->+|45+>]x[- |45-> + |45+>] + [|45-> + |45+>]x[|45->+|45+>])
= 1/sqrt(2) (|45+>|45+> - |45->|45->)

Oops, a minus sign forgotten (doesn't affect the result, I was taking it over from a piece of paper):

|psi> = 1/sqrt(2) (|x>|y> + |y>|x>) = 1/2 1/sqrt(2) ( [|45->+|45+>]x[- |45-> + |45+>] + [-|45-> + |45+>]x[|45->+|45+>])
= 1/sqrt(2) (|45+>|45+> - |45->|45->)

 

[LGuard332]

Simple Question?

I apologize if my terminology is wrong.

What is the result if the signal photon hits the screen before a choice is made about whether or not to measure the idler photons to determine the path? Two bands or an interference pattern.

If the result depends on whether or not it is ultimately decided to measure which path the photon goes down. Could we put the choice whether or not to measure the photons' path on Alpha Centuri and receive the answer through a interference pattern or no interference pattern here on Earth before the choice is ever made!

The only problem I see is it bumps up against Einstien's pesky little speed limit since we get an answer quicker than immediately.