Afshar Wikipedia entry incl. views of Unruh and Lubos Motl

[Hans de Vries]

The Wikipedia entry is here

http://en.wikipedia.org/wiki/Afshar_experiment

Afshar's blog is here:

http://irims.org/blog/index.php/2004/09/25/questions_welcome

So the idea seems to be that opening the second hole causes an
interference pattern that increases the probability for photons
(going through the first hole) to go through the wire grid placed
on the lows of the interference pattern.

Essential is then that the probability seems to be affected even
though the path is known. In a "single photon at a time" experiment
this would then mean that the "wave" goes through both holes but
the photon choses only one hole.

Regards, Hans.

 

[Gokul43201]

Shouldn't this be in the QM sub-forum ?

 

[Hans de Vries]

Shouldn't this be in the QM sub-forum ?

For a discussion of Afshar's experiment sure. I placed it on Lubos Motl's
home ground because his view was included in the wikipedia entry.

Regards, Hans

 

[Hans de Vries]

caribou withdraw his post while I was editing, so I'll withdraw the quote from his post as well, Hans

It's true that the detection is well after going through the interference
pattern at the wires. So the detected photons could have which way
information without destroying the interference pattern they went
through because the detection was made later. (The desctruction
doesn't go back in time)

This seems a satisfactory explanation.

The above is different however from what Unruh and Motl say:

"the photons that interact with the wire grid allow us to measure
the interference pattern, but they don't carry the "which way
information", as explained by Unruh,

while the photons that avoid the wires are absorbed by the detectors.
These photons carry the "which way information" but they don't
create any interference pattern"

The latter would not explain why the intensity increases when an
interference forms because the photons that reach the detector
would not take part in the interference process.

Regards, Hans

 

[Caroline Thompson]

There has been a lot of discussion about the Afshar experiment, in sci.physics.research and elsewhere, but does anyone know if it has actually been published in PRL yet? I feel we need more details before trying to analyse it. I did a Google search on Afshar and found the following:

http://www.irims.org/quant-ph/030503/

but, since it is 33 pages long, I'm reluctant to download it unless I know in advance that it is relevant. Hmmm ... I think I'll have to. I've just checked back and the abstract is:

Bohr’s principle of complementarity predicts that in a welcher Weg (“which-way”) experiment, obtaining fully visible interference pattern should lead to the destruction of the path knowledge. Here we report a failure for this prediction in an optical interferometry experiment. Coherent laser light is passed through a dual pinhole and allowed to go through a converging lens, which forms well-resolved images of the respective pinholes, providing complete path knowledge. A series of thin wires are then placed at previously measured positions corresponding to the dark fringes of the interference pattern upstream of the lens. No reduction in the resolution and total radiant flux of either image is found in direct disagreement with the predictions of the principle of complementarity. The theory of measurement and non-perturbative techniques are aslo briefly discussed.​

Caroline

 

[caribou]

I deleted my earlier post because I decided it was a bit aggressive. The post basically just said the experiment is not saying anything new and that the lack of interaction at the wires due to destructive interference at those points is not a "detection" so complementarity is not violated.

Also, I'd add that in the Wikipedia entry, Figure 1 and Figure 3 actually reflect a superposition state of detectors. Light is not behaving like a particle at any time, it is still behaving like a wave but we then get decoherence and a "particle" detection at one or the other detectors.

So the effect of the decoherence is being ignored by the suggestion that there is particle behaviour, when particle behaviour doesn't exist until detection.

 

[Hans de Vries]

The Afshar Wikipedia page was vandalized by somebody of
the university of Graz in Austria. (It seems that it has now
been restored)

I personally find this one of the more interesting experiments lately.
Regardless of the various claims by different people. Experiments
stand for them self. The information they provide is independent
of the different opinions and explanations.

Regards, Hans

 

[vanesch]

The Afshar Wikipedia page was vandalized by somebody of
the university of Graz in Austria. (It seems that it has now
been restored)

I personally find this one of the more interesting experiments lately.
Regardless of the various claims by different people. Experiments
stand for them self. The information they provide is independent
of the different opinions and explanations.

Well, I'm not particularly impressed, honestly.

Let us consider the symbolic quantum mechanics of this thing.

You have an initial state which is:

|state0> = (|hole1> + |hole2>)|coldwire>|det1x>|det2x>

The "grid" states correspond to |hole1> + |hole2> (and this transits the state), and |hole1> - |hole2> (and this "blocks" the state), we will say that the "wire heats up".

After interaction with the wire (which comes down to the unitary evolution:
sqrt(2) |nophoton>|hotwire>(<hole1| - <hole2|) +
(|hole1>+|hole2>)|coldwire>(<hole1|+<hole2|),

we obtain the state:
(|hole1> + |hole2>)|coldwire>|det1x>|det2x>

(the state is simply transmitted), and then we will have the interaction with the "lens + detector" system:

(|det1+>|det2-><hole1| + |det1->|det2+><hole2| +
|det1->|det2-><nophoton| )<det1x|<det2x|

if we apply that to our state, we find:
(|det1+>|det2-> + |det1->|det2+>)|coldwire>

So we get 50% chance that detector 1 clicks, and 50% chance that detector 2 clicks.

If we now start with the state |hole1>|coldwire>|det1x>|det2x>

which we can rewrite as:

1/2 (|hole1>+|hole2> + |hole1> - |hole2>)|coldwire>|det1x>|det2x>
then after the grid we obtain:


1/sqrt(2) |nophoton>|hotwire> +
1/2 (|hole1>+|hole2>)|coldwire>

After lens + detection, we have:

1/sqrt(2) |det1->|det2->|hotwire>
+ 1/2 (|det1+>|det2-> + |det1->|det2+>)|coldwire>

So we have that half of the photons do not get through,
and 1/4 of the photons are seen by detector 1 and 1/4 of the photons are seen by detector 2.

Of course, this is the simple case where the grid has a 50% coverage. The normalizations (which I treated sloppily here) will change when your grid hasn't got a "dutycycle" of 50%.

I really don't see where the mystery is. The grid makes you loose the original "hole" origin (if it is 50% 50%), because it enforces a pattern (it is a hologram of two holes if you want to).

cheers,
patrick.

 

[vanesch]

I really don't see where the mystery is. The grid makes you loose the original "hole" origin (if it is 50% 50%), because it enforces a pattern (it is a hologram of two holes if you want to).

In fact, the data that is missing is the ratio between the two pinhole images when the grid is in place and when one pinhole is covered. Because of the hologram function of the grid, some photons will exactly be scattered onto the "image" of the covered pinhole.
Also, you have to take into account the real "cross section" of the wires, and not their physical dimensions. For instance, even with a very thin wire, you can capture radiowaves.
I do not especially expect a "degradation of the resolution". I do expect a slight diminishing of the light intensity and a corresponding rescattering of the light onto the other peak. It is not clear to me in the data how this is contradicted.

Caveat: I didn't read the paper, just browsed through it...

cheers,
Patrick.

 

[Hans de Vries]

Because of the hologram function of the grid, some photons will exactly be scattered onto the "image" of the covered pinhole.

This seems to be the only all-optical explanation possible for Afshar's
experiment.

From an optical viewpoint with light being linear there should be no
influence from one pathway on the other. The interference takes
place only in the middle of the experiment. The light after the pattern
should not be influenced because of the linear interaction.

Other optical explanations of why light of one pinhole influences the
intensity of the spot of other seem very unlikely:

1) The interference pattern in the middle can only influence through
non-linear interaction, which can be neglected here.

2) Normal scattering on the grid would not get concentrated on the
other pinhole's spot.

Since the angle corresponds with the grid you may indeed expect that
the light is holographically deflected on the other spot.

The fact that the light is deflected more in the low bands may be the
result of the angle with respect of the grid. The light of one pathway
can reach the shadow area (at the wires) of the other one.

The conclusive evidence would be if one could see such a deflection
concentrated on the spot of the other also when there is only one
pathway active.


Regards, Hans

 

[vanesch]

The conclusive evidence would be if one could see such a deflection
concentrated on the spot of the other also when there is only one
pathway active.

This is indeed what I tried to point out: that's the measurement that's missing. But it is not a big deal in fact: you illuminate a grating and you can calculate the bragg diffraction, which corresponds to it. It fits perfectly with the other peak !
The perfect holographic image would only be obtained with a "grating" which has a transmission function equal to the diffraction pattern (a kind of sine function). If you change it to a square function, you will "generate harmonics" in the spatial distribution, and hence multiple images, of which the strongest will still be the "fundamental", and corresponds to the image of the covered slit. The other images will be further away, and that's exactly Bragg's law.

There's a whole branch in classical optics which does exactly this kind of things (holography is a part of it), it is called Fourier Optics. Usually you do more sophisticated things than that in Fourier optics. For instance, you can filter, in a very similar setup, low frequency or high frequency (spatial image frequency) components of your image, to enhance edges, or to smoothen out glitches... (but nowadays, people use Photoshop to do that :-)
I really, really don't think that there is anything 'very deep' in it.

cheers,
Patrick.

 

[Hans de Vries]

I really, really don't think that there is anything 'very deep' in it.

So this seems indeed the explanation of Afshar's experiment.

The effect that the grid sends part of the other path's light
exactly to the opposite spot can easily be checked with the
principles of the Phased Array Radar. (see further below)

It works equally well in the optical picture as with Feynman's
popular QED representation. Thus: In the photon representation
we can say that according to QED there is a non-zero chance that
a photon detected at one of the spots came through the other
one's hole.

A Phased Array Radar can send radiation into any direction by
adjusting the phases of the individual transmitters in the array.
In case of Afshar's grid we know that both lightpaths have
equal phases because the wires are placed on the lows of the
interference pattern.

This means that by removing part of the phase information
with the grid we get the result that some of the light (cq. photons)
is directed to the other one's spot (detector)

Feynman would have loved to use Afshars experiment as another
example of how his beloved QED can predict the behavior of quantum
mechanical particles!

Regards, Hans

P.S. This experiment has fooled both proponents and opponents of
Afshars interpretation, which only shows how careful one must be
in the interpretation of an experiment ! There are other famous
experiments which play similar tricks with the interpreters.

 

[vanesch]

I think we agree 100%.

cheers,
Patrick.

So this seems indeed the explanation of Afshar's experiment.

The effect that the grid sends part of the other path's light
exactly to the opposite spot can easily be checked with the
principles of the Phased Array Radar. (see further below)

It works equally well in the optical picture as with Feynman's
popular QED representation. Thus: In the photon representation
we can say that according to QED there is a non-zero chance that
a photon detected at one of the spots came through the other
one's hole.

A Phased Array Radar can send radiation into any direction by
adjusting the phases of the individual transmitters in the array.
In case of Afshar's grid we know that both lightpaths have
equal phases because the wires are placed on the lows of the
interference pattern.

This means that by removing part of the phase information
with the grid we get the result that some of the light (cq. photons)
is directed to the other one's spot (detector)

Feynman would have loved to use Afshars experiment as another
example of how his beloved QED can predict the behavior of quantum
mechanical particles!

Regards, Hans

P.S. This experiment has fooled both proponents and opponents of
Afshars interpretation, which only shows how careful one must be
in the interpretation of an experiment ! There are other famous
experiments which play similar tricks with the interpreters.

 

[Hans de Vries]

Simplified Afshar Experiment.

The image here:

http://www.geocities.com/zekise/unoba1.jpg

Shows a simplified version of Afshar's experiment. It does away with the
lens which I agree plays no essential role in the experiment.

Again it shows that the interference pattern only exist in the area were
the two light cones overlap. Undergraduate physics predicts that the
two light cones should come out unmodified after the interference region
since light doesn't interact.

This is, strangely enough, generally not recognized. People seem to apply a
sort of Bohmian picture were by the photons only pass through the hights
of the fringes and not through the lows, and thereby forgetting the wave
picture.

Photons in the Bohmian picture have a longer path from source to detector
than the light wave since they don't go straight. There are the implications
with the speed of light... See the image here:

http://plato.stanford.edu/entries/qm-bohm/figure1.gif

from

http://plato.stanford.edu/entries/qm-bohm/#ge

(The Bohmian Interpretation doesn't have this problem with particles
with mass)


Afshar claims that the wire grid doesn't diffract. (Which it should do,
diffracting light from one cone exactly on the other, which explains
his result using classical theory only, see the posts above )

He is testing this however with both pinholes open. (see figure 7. in
his preprint below) The way to test this should be with only one pinhole
open while looking at the detector of the of the closed one.
http://www.irims.org/quant-ph/030503/Afshar Complementarity All.PDF

:
Figure 7 (a) The configuration testing the effect of the wires in the wire grating (WG).
(b) Data representing the images of pinholes 1 and 2. No reduction in the resolution of
the images is found at the image plane 2 s . This implies that no diffraction is produced
by the WG and thus WWI is still complete

Regards, Hans

 

[dankomed]

Hi,

Afshar's interpretation is more seriously flawed than posted in this forum. Basicly the error is clearly shown in my recent paper, where I discuss almost exclusively Unruh's experiment. Unruh's experiment is very nice because you can write only 10 math expressions for the quantum amplitudes for each of the branches of the interferometer. Then in this relatively easy math environment I understandably point out why "pure state" experiments cannot be interpreted as "which way" one. This is mathematical theorem, and I will be glad if someone discusses it.
http://philsci-archive.pitt.edu/archive/00003048/

Indeed I show that Afshar's error starts from the very assumption that there is "which way" information when both pinholes are open plus no grid present. Indeed because of such a fundamental error a lot of antirealist positions occurred including that of Heisenberg and others. see
"The moon is not there where no one looks at it"
http://www.eequalsmcsquared.auckland.ac.nz/sites/emc2/tl/philosophy/moon.cfm

If one takes the realist position, and thinks that quantum formalism/mathematics models adequately the reality out there, then one cannot make "which way" claims without consulting the mathematical formalism first.

I can try to explain better what I mean with an analogy - if one believes that Einstein's relativity describes adequately the reality, then he cannot sum velocities in "Gallilean way: v1 + v2". Instead he must look at relativistic mathematics first.

The Afshar's and Unruh's setups are analogous - there is no "which way" claim at first place, and I don't understand why one does not consult the mathematics first? Maybe the issue is masked additionally by the fact that when one have expressions of the type x1 - x2 + x1 + x2, he is confused what anihilates with what.

 

[CarlB]

Oh... To read this thread thinking that it was new and then to be reminded that Caroline Thompson died this past year, and we shall have no more of her posts forever.

 

[Hans de Vries]

Oh... To read this thread thinking that it was new and then to be reminded that Caroline Thompson died this past year, and we shall have no more of her posts forever.

That's really sad news.

http://en.wikipedia.org/wiki/Wikipedia:Deceased_Wikipedians#Caroline_ThompsonRegards, Hans

(edit) some more over here: http://en.wikipedia.org/wiki/User_talk:Caroline_Thompson#Sad_news