From: "Caroline Thompson" <ch.thompson1@virgin.net>

Newsgroups: sci.physics.research

Subject: Re:  A possible loophole in the Delayed Choice Quantum Eraser

Date: Fri, 26 Nov 2004 07:22:26 +0000 (UTC)

 

"Schneibster" <wbishop@mbay.net> wrote in message news:Schneibster.1g91h2@physicsforums.com...

> Hi, I've been honing this idea for about 3 or 4 months, and

> now I'd like to bring it here and discuss it. I'll start out by stating

> that I am not a professional physicist; I have been studying

> physics on an amateur basis for most of my life, so most of my

> knowledge is from various general-reader-level books. However,

> I do have a degree in electronics engineering, so hopefully I will

> not make a complete fool of myself here.

 

Hi, I too am not a professional physicist, though I became interested in physics 10 years ago, starting with trying to find out what really happened in the Bell test experiments.  I too have looked Kim et al's experiment.  It was discussed in sci.physics in 1999, and I wrote to the author to ask for more information.  I'm afraid I never did feel I had quite enough facts to be able to say definitely what happened, though I am convinced that this experiment, together with all others currently classified as "quantum optics" ones, can be explained by ordinary wave theory once you take account of the clever ways in which phase, frequency, polarisation and exact direction are all tied together in the outputs from a pumped nonlinear crystal.

 

> As a review, the Delayed Choice quantum eraser was proposed by

> Scully and Drühl in 'this'

> (http://prola.aps.org/abstract/PRA/v25/i4/p2208_1)

> article, and then performed by Kim, Kulik, Shih, and Scully in 1999 as

> reported in 'this'

> (http://xxx.lanl.gov/PS_cache/quant-ph/pdf/9903/9903047.pdf) article.

> Basically, photons are led through a dual-slit apparatus, and then

> split with a &#946;-BBO crystal spontaneous parametric down-converter

> (SPDC) into orthogonally-polarized pairs that are separated with a

> Glan-Thompson prism and sent to two separate measurement areas.

 

Ah, thanks for reminding me!  The prism is not shown on their diagram.  Perhaps I'd better go away and re-read the paper before commenting too much.

 

> In the first area, the photons are allowed to fall on a movable

> detector that checks various positions for photon count. This is

> essentially looking for interference fringes ...

 

I'm not so sure this is the right way to think of it.  The actual interference in the experiment takes place not here but when the beams from the two parts are combined at the beamsplitter.  It is not measured directly as an interference pattern but detected by means of variations in coincidence counts.  [This technique for identifying interference, incidentally, occurs in many quantum optics applications, including the one I'm currently interested in: Grangier's team's new proposal for a "loophole-free" Bell test, http://arxiv.org/abs/quant-ph/0403191 ]

 

> ... which would show up as high incidences at some locations and

> low or zero incidences at others. These photons are labeled "signal"

> photons in the 1999 paper.

>

> In the second measurement area, the photons are either captured

> by one of two detectors that determine their "which-path" data,

> which is valid for determination of the which-path data for their

> associated signal photon, or are sent to a "scrambler" that makes

> it impossible to reclaim the which-path data from them.

 

Yes, but I think there is a completely different explanation for the whole experiment and it does not assume that we know which path the light went by since it always goes both ways.  We are dealing here with coherent light.  The signal and idler beams pretty certainly have a definite phase relationship.  I haven't checked the details but the potential for interference is there, so long as the light is passed through, say, a polarising cube or equivalent that is set at 45 deg to the polarisation directions.  Another feature of the light that may be relevant (it is very important in certain experiments, though I haven't yet checked here) is that it is likely to be in a mixture of two phase relationships to the pump laser.  [See http://arxiv.org/abs/quant-ph/9912082]

 

> These are labeled as "idler" photons in the 1999 paper. The

> scrambler is labeled as a "quantum eraser" in the 1982 paper.

>

> The "Delayed Choice" is introduced by making the path to the idler

> measurement area longer than the path to the signal measurement

> area.

 

If my kind of explanation is right, this path length difference is not relevant.  There is some real interference at the beamsplitter, and path lengths to there look equal.  The actual measurement is a coincidence one, so presumably signals are counted that arrive at the same time and must have been emitted at the same time.

 

> The experimenters report that, after correlation, the collection of

> signal photons that are correlated with idler photons that went to the

> quantum eraser show interference, whereas the collection of signal

> photons that are correlated with idler photons that went to the

> which-path determination detectors do not show interference.

 

The part that does not show interference is the part in which there could not have been any since the beams did not mix at the beamsplitter.

 

> By conventional causality, because

> a) No matter what we do to the signal photons, the idler photons are

> unaffected as far as we can tell,

> but

> b) depending on whether we measure the which-path information we

> do or do not see interference in the signal photons,

> we interpret the idler photon measurement or non-measurement of

> which-path information as the "cause" and non-interference or

> interference as the "effect." Therefore, by that conventional

> interpretation, this experiment appears to violate the normal time

> direction of causes preceding effects.

 

As you say, this is the conventional interpretation.

 

> I wondered whether it might be possible to make this time reversal

> explicit. My direction of attack was to attempt to differentiate the

> interference case from the non-interference case -without resort to

> the correlation, or in fact to any measurement of the idler photons.-

>

> After developing this into something like a paper or proposal, I

> realized that I had forgotten to consider that the superposition of

> the interference patterns detected in the initial experiment might

> add up to the non-interference case. After spending some time

> being frustrated at such an oversight, I evaluated the published

> experimental results, and found that

> a) they are not sufficiently detailed to absolutely confirm or deny

> that there is a difference between the interference and

> non-interference patterns,

> but

> b) they tend to indicate that there will be a difference.

 

[snip details]

 

> The trough-to-peak difference in local minima and maxima of impacts

> appears to be 17-25% of the maximum for the non-interference case,

> but 23-30% for the interference case. In addition, there are many more

> apparent "deviations" from the best curve in the interference case than

> in the non-interference case ...

 

Agreed, the fit to their predicted curves is not that good, but nevertheless I think the graphs show an interference pattern.  Other similar experiments -- the "ghost interference" ones -- have produced similar results.

 

> My proposal is therefore as follows:

> 1. Replace the stepper-motor-driven detector with a CCD array that can

> read single lines. Such arrays 'exist'

> (http://naysika.mi.iasf.cnr.it/epic/pnfocalplane.htm) and if properly

> oriented would allow the detection of evidence of interference fringes

> very quickly, if combined with appropriate electronics and fast image

> processing.

> 2. Use optical delay lines, preferably variable switched delay lines

> with external fiber-optic hookups to facilitate the use of measured

> cable lengths to introduce the desired delay.

> 3. Replace the beam splitters that randomly direct one half of the

> idler photons to the eraser, and the other half to the which-path

> detectors, with LCD optical switches. Such switches are currently in

> use in optical networking switches available from over fifty companies

> worldwide.

 

Yes!  That really would test the theory!  I would predict that this would destroy the interference pattern.

 

> I would begin by attempting to successfully differentiate the

> interference pattern from the non-interference pattern. If this were

> successful, then I would proceed with attempting to apply negative

> feedback to the experiment, in which the detection of interference

> would cause the measurement of the idler photons' which-path

> information, and detection of non-interference would cause the idler

> photons to be sent to the eraser.

>

> I seek opinions on whether this experiment is possible (i.e., whether

> the interference and non-interference cases can be differentiated), and

> on what the results might be if it -is- possible, and on what might

> happen if it is possible and negative feedback is applied.

 

I'm afraid I have not studied your idea sufficiently to really judge it, but have the impression that it would be a step in the right direction.  How about applying your mind to rather easier experiments first, though?  The Bell test experiments are not nearly as intricate as this one.  They need repeating with slight modifications, designed to compare the explanatory power of the rival models (local realism with reliance on the various "loopholes" [see my web site] versus quantum theory) under a range of conditions, not just concentrating on the Bell test.

 

Cheers

Caroline

 

 

From: "Caroline Thompson" <ch.thompson1@virgin.net>

Newsgroups: sci.physics.research

Subject: Re: A possible loophole in the Delayed Choice Quantum Eraser

Date: Fri, 26 Nov 2004 09:24:22 -0000

 

Re Kim, Kulik, Shih, and Scully in 1999 article,

http://xxx.lanl.gov/PS_cache/quant-ph/pdf/9903/9903047.pdf

 

 

Just a small correction to what I wrote yesterday:

 

There is pretty certainly interference at the screen with the movable detector as well as at the central beamsplitter.  In both places if you look at the total events (i.e. if you look at "singles" rather than coincidences) you don't easily spot this, the main reason being that there are two interference patterns superposed, 180 deg out of phase.  The patterns at the two locations are the same, so they show up when you look at coincidences.

 

That would be in the perfect case, if all frequencies were monotonic and there were no extraneous light.  In the real experiment it is not surprising to find the interference pattern rather poor.

 

Caroline

 

ch.thompson1@virgin.net

http://freespace.virgin.net/ch.thompson1/

 

 

Return to front page

or Read more letters