Alain Aspect's experiment setting

[Joao]

Hi everyone! Sorry for the bad english.

I'm a psychologist from Brazil, so I don't know much of physics nor English.

I'm having a hard time understanding his setting.
In a very simple way, he made two entangled photons, each went to a polarization analyzer, that was set to random positions. Then it showed if each photon passed or no through the analyzer.
Is it correct?

If so, the expected by classical physics would be that 75% of the time at least one photon would pass, quantum mechanics predicted that 85% of the time at least one would pass.

So, the experiment gave the number 85%. Therefore quantum mechanics is right?

Thanks

 

[DrChinese]

I'm having a hard time understanding his setting.
In a very simple way, he made two entangled photons, each went to a polarization analyzer, that was set to random positions. Then it showed if each photon passed or no through the analyzer.
Is it correct?

If so, the expected by classical physics would be that 75% of the time at least one photon would pass, quantum mechanics predicted that 85% of the time at least one would pass.

So, the experiment gave the number 85%. Therefore quantum mechanics is right?

Thanks

I will try to keep my explanation short and simple, but with this subject that is rarely possible.

What you refer to as "classical physics" is more commonly referred to in the community as "Local Realism". Local Realism is composed of 2 ideas: that no cause can go faster than the speed of light (Local); and that any physical observable property of a system must have a specific value at all times (Realism).

Local Realistic theories have some requirements that conflict with some predictions of Quantum theory (QM). The Aspect experiment shows that in this conflict, the predictions of QM are correct and those of Local Realism are not. Since then, thousands of experiments have confirmed this result.

The Aspect experiment checked the polarizations of pairs of photons, whether they matched or did not match. (Individual photons in those experiments always passed a polarizer 50% of the time.)

 

[Joao]

I will try to keep my explanation short and simple, but with this subject that is rarely possible.

What you refer to as "classical physics" is more commonly referred to in the community as "Local Realism". Local Realism is composed of 2 ideas: that no cause can go faster than the speed of light (Local); and that any physical observable property of a system must have a specific value at all times (Realism).

Local Realistic theories have some requirements that conflict with some predictions of Quantum theory (QM). The Aspect experiment shows that in this conflict, the predictions of QM are correct and those of Local Realism are not. Since then, thousands of experiments have confirmed this result.

The Aspect experiment checked the polarizations of pairs of photons, whether they matched or did not match. (Individual photons in those experiments always passed a polarizer 50% of the time.)

Thanks! It really helped a lot!

Please, are you Dr.Chinese from the website that goes with this name? That's a great website! Really awesome! =)

 

[DrChinese]

Thanks! It really helped a lot!

Please, are you Dr.Chinese from the website that goes with this name? That's a great website! Really awesome! =)

Yes, and glad you like it!

-DrC

 

[Mathematech]

Bell's original proof of his inequality is based on what can be called a classical local hidden variable model which is only one very specific type of local realist model. If one defines "local realism" to mean the type of model Bell considered then yes, Bell's inequality and Aspects results show that QM is not "locally real". However this requires a very narrow definition of realism. There are realist models in a more fundamental sense of the term which are not Bell's classical local hidden variable models.

The inequality actually applies to a wider range of realist models than what Bell considered. Eberhard's version derives it without assuming Bell's classical hidden variables, he merely assumes counterfactually definite local random variables. Fine's and Landau's versions generalize the scope further to include possibly non-local and possibly non-deterministic models, with Landau's being the most general - he derives it for any model with the relevant variables defined on a common joint probability space, and his argument holds regardless of whether the model is local or non-local and whether it is deterministic or non-deterministic. The scope of Bell's inequality cannot be extended further as satisfying the inequality is a sufficient condition for the variables to be definable on a common joint probability space. Bell's inequality does not and can not rule out models in which the variables are not defined on a common joint probability space even if those models are real and local.

The way formal classical mathematical probability works is as follows: For a given experimental context, if the experiment can be modeled by classical probability, one associates a "Kolmogorov probability space" with the experiment and things being measured in the experiment are modeled by what are known as random variables on the space. In simple terms what this means is that for an experiment you have a set of possible fundamental outcomes, a specification of subsets of this set of outcomes that one can measure the size of in some reasonable manner and this is then used to define probabilities of the variables attaining certain values when the experiment is run. Generally the probability space will be different for different experimental contexts and for some things it might not be possible to model the measurement results as random variables defined in this way. (For example if the outcomes correspond to points in an infinite dimensional space, there is no reasonable way to measure the sizes of sets of outcomes.) Classical physics (including relativistic physics) however has a very privileged position when it comes to using probability - probability can be used as described and moreover one can use a common probability space for varying experimental contexts (something which doesn't happen say when you use probability in social science or biology applications).

Besides Bell's inequality there is another result, the Kochen-Specher Theorem, which basically demonstrates that quantum mechanics is very different to classical mechanics in that it cannot in general use a common classical probability space for different experimental contexts to describe all possible quantum mechanical measurement results. Bell's inequality takes it up a notch and shows that even for one particular experimental scenario, that of two entangled particles, the measurement results cannot be modeled as random variables on single classical probability space.

Aspect's and others results show that QM's predictions are correct and violate Bell's inequaity. Thus QM is not a misleading approximation of some theory that works in the same way as classical physics, it is more accurate and more fundamental than classical physics. But contrary to what sensationalist science journalists as opposed to scientists might say, it does not indicate that there are faster than light interactions or that one must choose between locality and realism in a general sense.

 

[Leo1233783]

Hello,
who can cite , not a thousand, just one photons experiment where 85% of the detection was observed and give the references of the publication with the detection details. Or just 2%. Believing things heard in public lectures is an act of faith.

Some talk seriously of 20 to 30%. But in the details, it is 30% of a part of the photons . Perhaps, their method is still interesting. But it is not 75.1 %.
The other limit used is 86.6% of detection by arm, the root square of 75% for the crossed detection.

With electrons, physicists claim explicetely 75%++, giving credible details ( but not yet the raw data ) . Honestly, it must be said that it is not over. If the realists find a perfect cos² at 75% , they also know how to find similar curves at 85%. Thus, the experimental raw results will be reviewed. However, we roll away the act of faith.

 

[vanhees71]

I don't know, whether this experiment meets your requirements:

http://dx.doi.org/10.1103/PhysRevLett.115.250402
https://arxiv.org/abs/1511.03189

 

[zonde]

Bell's inequality does not and can not rule out models in which the variables are not defined on a common joint probability space even if those models are real and local.

As I understand (spacetime events based) model in which variables are not defined on a common joint probability space has to exploit fair sampling loophole.
Or is it not so?

 

[DrChinese]

Hello,
who can cite , not a thousand, just one photons experiment where 85% of the detection was observed and give the references of the publication with the detection details. Or just 2%. Believing things heard in public lectures is an act of faith.

Some talk seriously of 20 to 30%. But in the details, it is 30% of a part of the photons . Perhaps, their method is still interesting. But it is not 75.1 %.
The other limit used is 86.6% of detection by arm, the root square of 75% for the crossed detection.

With electrons, physicists claim explicetely 75%++, giving credible details ( but not yet the raw data ) . Honestly, it must be said that it is not over. If the realists find a perfect cos² at 75% , they also know how to find similar curves at 85%. Thus, the experimental raw results will be reviewed. However, we roll away the act of faith.

The fair sampling assumption, and photon detection efficiency, is a subject for another thread. The OP is asking about the groundbreaking Aspect experiment generally, and at a B level. The state of open or closed "loopholes" completely different, and is probably more A level.

 

[Joao]

Thanks a lot! Now I have a more profound understanding of the experiments mechanics and significance!

So, sorry to be repetitive, but it seems that each YouTube video gives a different parameter about the % that would indicate the "hidden info" and the % that indicates "Quantum Mechanics", as the winner of the argument...

So, imagining a very very simple experiment, just pair of photons, each of the photons being measured only one time, about his polarization, and their polarization could be measured in three possible angles, 1, 2 or 3.

Sooooo, if we put both photons with the same angle of measurement:
Both will pass or both will fail, always. And the chance of passing would be 50%.

And if we start the randomization of the angles at the detectors, here what we would have:
Some photons would pass angle 1, but not 2 and 3, some would pass 1 and 2, but not three...

And here's that fancy table I can't feel I understa t it:

1 2 3 1and2 1and3 2and3
+ + + S S S
+ + - S D D
+ - - D D S
+ - + D S D
- - + S D D
- + + D D S
- + - D S D
- - - S S S

S goes for same result (both passing or not) D for different results (one pass, the other not). + is pass, - is not pass

And so on.

Soooo, the probability of a pair of photons having the same results (pass or fail both), is 12 in 24. Or 1 in 2.

If we don't consider the situations that the photon always pass or always fails, we have 6 in 18, or 1/3.

But then, in this last scenario, Quantum Mechanics happens, and instead of 1/3, we have 1/2 ! And that's impossible to explain using local physics!Did I understood correctly?

Thanks a lot for your time! Really Thanks! =)

 

[DrChinese]

Thanks a lot! Now I have a more profound understanding of the experiments mechanics and significance!

So, sorry to be repetitive, but it seems that each YouTube video gives a different parameter about the % that would indicate the "hidden info" and the % that indicates "Quantum Mechanics", as the winner of the argument...

So, imagining a very very simple experiment, just pair of photons, each of the photons being measured only one time, about his polarization, and their polarization could be measured in three possible angles, 1, 2 or 3.

Sooooo, if we put both photons with the same angle of measurement:
Both will pass or both will fail, always. And the chance of passing would be 50%.

And if we start the randomization of the angles at the detectors, here what we would have:
Some photons would pass angle 1, but not 2 and 3, some would pass 1 and 2, but not three...

And here's that fancy table I can't feel I understa t it:

1 2 3 1and2 1and3 2and3
+ + + S S S
+ + - S D D
+ - - D D S
+ - + D S D
- - + S D D
- + + D D S
- + - D S D
- - - S S S

S goes for same result (both passing or not) D for different results (one pass, the other not). + is pass, - is not pass

And so on.

Soooo, the probability of a pair of photons having the same results (pass or fail both), is 12 in 24. Or 1 in 2.

If we don't consider the situations that the photon always pass or always fails, we have 6 in 18, or 1/3.

But then, in this last scenario, Quantum Mechanics happens, and instead of 1/3, we have 1/2 ! And that's impossible to explain using local physics!Did I understood correctly?

Thanks a lot for your time! Really Thanks! =)

The 50% pass rate for an individual photon is because its orientation is random. Actually the classical (local realistic) theories also predict that.

The actual same/match rate is 25% rather than 50%. That is because the quantum prediction is cos^2(theta) where theta is the angle difference between the settings. Usually that is made to be 120 degrees or 60 degrees or similar in an example. There are actually (infinitely) many different combos of angles that highlight the issue, but as you might expect, the math can get complicated pretty quick. And it is harder to explain.

Your chart is perfect by the way. If you play with it enough, and try hand-picking results, you will eventually see that the only way to get a result close to the QM value is to "cheat". That is, you know in advance what the results will be and pick those settings that will match the QM rate (25%) on average. Other than that, you cannot - over any decent sample size - below the Local Realistic average of 1/3 (33.3%).

 

[Joao]

Thanks a lot! Really! =)

Please,

The actual same/match rate is 25% rather than 50%.

Are you referring to the prediction when all the possible outcomes is considered, this one:

Soooo, the probability of a pair of photons having the same results (pass or fail both), is 12 in 24. Or 1 in 2.

Thanks again!

 

[DrChinese]

Are you referring to the prediction when all the possible outcomes is considered, this one...

The match rate at various angles is sensitive to the difference. That wouldn't be a problem except that the QM prediction is not linear, it varies as the square of the cosine. While that statement may not seem relevant, it explains that the choice of measurement (what theta you select) "appears" to alter the outcome statistics.

Keep in mind: when theta=0, you get perfect correlation. Perfect correlation implies predetermination (local realism/classical ideas). But predetermination is completely in conflict with the idea that choice of measurement alters the outcome statistics. You can't have both!

QM predicts everything correctly, but is silent on the concept of predetermination. So it doesn't have a problem with this.

 

[rede96]

Bell's inequality takes it up a notch and shows that even for one particular experimental scenario, that of two entangled particles, the measurement results cannot be modeled as random variables on single classical probability space.

This is something I've been trying to understand for a while, but being a layman with no background in physics or great at math, I've found it difficult. But with reference to the statement above, in layman's terms (forgive my poor terminology) is it possible to explain the experimental results of entangles particles using multiple classical probabilities for a given set of measurement angles selected at random?

The reason I ask is that I can model the QM prediction from experiments using classical probabilities if I make certain assumptions about how the particles may react when their spin states (e.g.the Z direction of spin) are at different angles relative the measurement device. I.e. I use a different probability function depending that range of angles. Doing this I can artificially create the same QM results as seen in experiment using a single function for each particle independent of the other. But each time a different set of angles is chosen for measurement I have to make a new set of probability functions.

Now I know that means nothing in the real world, it's just me playing with numbers, but what I concluded from that is that is although I can make a formula for any given set of three angles used in experiment that will match the quantum mechanical outcome, and where the formula for each particle is independent (e.g. does not rely on knowing which of the three angles has been selected by the measurement devices) it is not possible to have a single formula that will give the right quantum mechanical prediction for all angles. I have to change the formula each time I change the three selected measurement angles.

Which is probably a very long winded and nonsensical way for me to arrive at the conclusion that what a violation of bell's inequality means is that although it is possible to make a function to predict the outcome of experiment for any given set of three angles using a set of classical probabilities, there is no classical formula based on a single set of probabilities that can be produced to explain the results of experiments done on entangled particles for all angles. (without knowing both angles selected in advance.)

Am I sort of the right lines here?

 

[DrChinese]

... there is no classical formula based on a single set of probabilities that can be produced to explain the results of experiments done on entangled particles for all angles. (without knowing both angles selected in advance.)

Am I sort of the right lines here?

That is correct, the only way you can make it work out is to know the angles in advance. Which of course defeats the point of modeling "as random variables on single classical probability space." If you fix it here, you are bound to break it there.

 

[Joao]

Wonderful! Really wonderful!

There's just one last thing that I can't understand, because each video that I watch in the internet gives a different information:

Assuming perfect equippaments, tuned to the most perfect angles (to show the difference between the local physics prediction and quantum predictions), and if we run this test a very large number of times, and the par of particles have the best kind of random plans (not cheating). The difference of the predictions from local physics and quantum mechanics would be?

According to this video, from Khutoryansky:
At 4:16

Local physics 75%
Quantum Mechanics 85%
(It's about spin, not polarization)

This video, from DrPhysicA

At 24:25
Local physics: 33%
Quantum Mechanics 25%
(About polarization, and I also didn't understood why he ruled out if the photons would pass all tests or fail all tests)

This video, from Veritassium
At 6:36

Local physics: 5/9 or more
Quantum Mechanics: 50%
Different results measuring spin

Here, in minute physics:

At 04:10
Local physics: 50%
Quantum mechanics: 85%
They do not talk directly about the experiment, and it's about polarization.

This video, from minutephysics:

At 10:33
Local realism (doesn't say!)
Quantum mechanics: aa- 100%, ab- 85%, bc 85% and ac 50%
About polarization

So, please, after all, how much of the time actually local physics and quantum mechanics would predict that both photons would be blocked, both would pass, or just one would pass and the other would be blocked?

And what Aspect found?

I have a very good friend that loves statistics and I would like to say this to him:

"Hey Leandro! See how nature is weird! Did you know that photons have a polarization (show him a pair of polarization glasses), and did you know that scientis can make photons with exactly the same polarization? Here's when it gets weird! You know, if we measure both photons with angles 1, 2 and 3, we would expect that only (I CAN'T FIGURE OUT THIS NUMBER! ) % of times both photons would pass both measurements, but check it out! Quantum theory says that both photons would pass (I CANT FIGURE OUT THIS NUMBER!) and there's this clever guy, Aspect, that actually made this experiment and quantum mechanics was right! Pretty weird!

Thanks again! =)

 

[vanhees71]

I can't watch all these movies. The point is that you have to tell the precise experimental setup with which the violation of Bell's inequality is demonstrated to get clear quantitative predictions. It's better to study one setup in detail first. For a very detailed explanation for the most simple case of spin-correlation measurements see

J. J. Sakurai, S. F. Tuan, Modern Quantum Mechanics, Addison Wesley (1994)

I'm pretty sure it's also in the newer edition of this book co-authored by Napolitano, but I can't check this right now.

 

[Drakkith]

Thread locked for moderation.