Delayed choice interpretations

[sunrah]

I just came across the delayed choice quantum eraser experiments (kim et al 1999) through Wheeler's astronomical delayed choice thought experiment and I can't stop thinking about it :-(

I would like to get some ideas on how people interpret this, either the general physics community consensus or your own personal feelings, if you'd be so brave

I keep telling myself that entanglement allows spooky action at a distance in space, so why not over time as well. Still that doesn't get anywhere close to sufficiently rationalising its effects and implications.

So first off, how do we currently reconcile entanglement with relativity if at all?
If I remember my undergrad physics, entanglement is due to QM allowing for superposition of wavestates and nothing more. Do other theories (like string theories) say anything more? Do we think we need a quantised spacetime to understand this?
Does delayed choice break causality (please no)?
Going back to the first question, do we think information is actually being exchanged between measurement events?
What does the Kim experiment suggest about the nature of time?

Thank you

 

[Strilanc]

The delayed choice experiment is a refutation of the idea that a photon has to be either exactly a particle or exactly a wave at all times. That's it. All the 'backwards in time' ridiculousness is just people misunderstanding or oversimplifying it.

Consider that the Copenhagen interpretation explains the delayed choice result without any time travel or even any FTL effects. The measurement of the signal photon causes a collapse that biases the later measurement of the choice photon. It's a bit surprising that the bias ends up looking wavy in some cases, but really that's all there is to it. (The reason there's no FTL effects needed is because the delay has to be long enough to ensure the choice photon is measured after the signal photon. That's the whole point of it being a delayed choice.)

I'd focus on Bell inequality experiments instead of on the delayed choice experiment, if you're interested in how relativity and quantum mechanics interact.

 

[sunrah]

The measurement of the signal photon causes a collapse that biases the later measurement of the choice photon.

I still don't see how can you bias correctly for a measurement that has not yet been conducted - after all the splitters are 50-50 and the signal stream doesn't measure the path. Can a signal photon striking the screen really bias its entangled idle photon which still has to negotiate a pair of 50% beam splitters?

EDIT: So Bell's Theorem requires detection of one photon to change the distribution of its entangled twin instantaneously, regardless of distance, but the signal stream doesn't carry any path information so surely it is measurement of the idle photon that biases the signal photon :-(

 

[Strilanc]

I still don't see how can you bias correctly for a measurement that has not yet been conducted

Do you know the actual math behind quantum mechanics? You set the thread to "Advanced" (i.e. graduate level), so I'm going to assume you don't just know it but are familiar with using it and looking at quantum information from different perspectives.

Here's a quantum circuit analogue of delayed choice:

Now consider: what's the state of the bottom qubit after the top qubit is measured, supposing that the rotation around the X axis was 45 degrees and that the measurement result was OFF?

Right, because of the entanglement, it's as if the second qubit started in the OFF state and was itself rotated around the X axis by 45 degrees. Specifically it's proportional to $\cos(45^\circ) |0\rangle - i \sin(45^\circ) |1\rangle$.

In that state, when you measure the second qubit along the $Z$ axis, you'll get differing probabilities ($\approx 14.6$% chance of being ON). But, because the spin is lying on the YZ plane, there's no bias along the $X$ axis. There's a 50% chance of +X and a 50% chance of -X, if you measure along the X axis.

Now, apply Bayes rule to flip things so that instead of predicting the second qubit's measurement based on the first qubit's measurement, you're using the second qubit's measurement to predict the first qubit's measurement. Along the X axis the 50/50 results give you no predictive power. Along the Z axis, the deviation from 50/50 gives predictive power. Call the X axis measurement "erasing", and the Z axis measurement "not erasing".

If you then hit yourself in the head with a hammer, so that you forget that we're working with correlations instead of causation, and also forget the order things happened in, and also forget that we're conditioning on both qubits instead of just one in order to see anything that looks like interference, then it kinda looks like the second qubit's measurement retroactively determined whether the first qubit's measurement showed interference. How amazing !

 

[sunrah]

I marked the thread advanced in line with posting guidelines, but I am a graduate student in astronomy and have very little recourse to QM. Please would you post a link to the experiment your talking about, thanks.

 

[Strilanc]

I marked the thread advanced in line with posting guidelines, but I am a graduate student in astronomy and have very little recourse to QM. Please would you post a link to the experiment your talking about, thanks.

It's not an experiment, it's an argument.

 

[StevieTNZ]

I marked the thread advanced in line with posting guidelines

See this post: https://www.physicsforums.com/threa...-existence-of-randomness.874153/#post-5489322

 

[bhobba]

I would like to get some ideas on how people interpret this, either the general physics community consensus or your own personal feelings, if you'd be so brave

These days, and this is the general view of physicists, is its simply an example of how in simple cases decioherence can be undone (although if it can be undone its not really decoherence - but that is a whole new thread). In modern times decoherence and observation are largely synonymous, so, roughly its simply that observations can be undone - strange but true.

Here is the text where all this is explained:
https://www.amazon.com/dp/3540357734/?tag=pfamazon01-20

At a less advanced level see:
http://quantum.phys.cmu.edu/CQT/index.html

See Chapter 20 but you will likely benefit from reading the lot.

Thanks
Bill

 

[bhobba]

So first off, how do we currently reconcile entanglement with relativity if at all? If I remember my undergrad physics, entanglement is due to QM allowing for superposition of wavestates and nothing more.

Deep questions Grasshopper.

First you are correct - that's all entanglement is.

All this Bell stuff is, is correlations. Its exactly the same as if you put a red slip of paper in an envelope, and a green slip in another. Open one envelope and you immediately know the other. Nothing mysterious. So what's going on? The interesting thing about quantum correlations is they have statistical properties different to the paper example. Why is it so? The reason is, unlike the paper slips that remain red and green at all times, quantum theory is silent on the proprieties of objects unless observed. That's the crucial difference. So where does this non locality stuff come from? Well since QM is silent on if objects have properties independent of observation what if we insist? Then it turns out you need FTL communication and locality is violated.

How is this reconciled with relativity? In QM the combination of relativity and QM is called Quantum Field Theory (QFT). Locality is built into QFT via the so called cluster decomposition property:
https://www.physicsforums.com/threads/cluster-decomposition-in-qft.547574/

Guess what? Entangled systems are correlated so the cluster decomposition property does not apply. Neat hey?

Thanks
Bill

 

[Swamp Thing]

This paper has some more insights: http://arxiv.org/abs/1007.3977 [Demystifying the Delayed Choice Experiments, B. Gaassbeek]

Wheeler's thought expt. is discussed in Appendix A.

 

[Demystifier]

Niels Bohr famously said: If quantum mechanics hasn't profoundly shocked you, you haven't understood it yet.

I like to paraphrase him with: If delayed choice quantum eraser shocked you more than the rest of quantum mechanics, you haven't understood the rest of quantum mechanics yet.

 

[sunrah]

This paper has some more insights: http://arxiv.org/abs/1007.3977

That's interesting. So if I understand the concept, entangled particles are a superposition state. Measuring one will make the probability distribution of the other conditional on the collapsed state of the measured particle. This conditional probability doesn't depend on either space or time and therefore, seeing as all events take place in spacetime, this conditionality transfer is instantaneous from our point of view.

The article ends by saying this isn't very mysterious at all, but I find it very strange. I guess it is easier to handle, if you accept that an entangled pair is like a single particle?

 

[sunrah]

I like to paraphrase him with: If delayed choice quantum eraser shocked you more than the rest of quantum mechanics, you haven't understood the rest of quantum mechanics yet.

Well clap me in irons and cast me hence into the wet. I can live with that, the rest of QM is not exactly insubstantial. But sometimes things take longer to sink in and I'd be the first fool with his hand up when it came to admitting, I didn't understand it at all when I did those courses (although understanding is not always required at university).

 

[ddd123]

That's interesting. So if I understand the concept, entangled particles are a superposition state. Measuring one will make the probability distribution of the other conditional on the collapsed state of the measured particle. This conditional probability doesn't depend on either space or time and therefore, seeing as all events take place in spacetime, this conditionality transfer is instantaneous from our point of view.

This insights article has allowed me to stop thinking about it too much: https://www.physicsforums.com/insig...elayed-choice-no-counterfactual-definiteness/

Basically, it's similar to the least action principle: whatever the history in time, it's always satisfied globally. So are these constraints on the measurement outcomes.

 

[Swamp Thing]

Sunrah, One thing that is not stressed heavily in Gaasbeek's paper (but is implied throughout) is that the interference only shows up when we match up "twin" photon pairs according to their arrival times, and plot the scatter diagram only for those photons on the screen whose twins go to D1 or D2. So we need to set up a telegraph that sends data from Bob to Alice for comparison and selection.

So I guess one could say, "If a photon happens to land on a null in the interference pattern at detector D0, then its twin will never go to D1 or D2 but will only go to D3 or D4. And if we reject / ignore all D3 and D4 events then we are left with those events where all the D0 photons have carefully avoided the nulls in the interference pattern, therefore this kind of plot will show interference". I have added some graphics to the figure in the paper:

On the other hand, photons that happen to land on interference nulls at D0 --> their twins will be going to D3 or D4 pretty often, so that this pattern will not reveal any nulls and peaks... like so:

 

[sunrah]

Thanks, I think the DCQE has been slightly demystified now, but I'm still not really comfortable with it.
If Alice (D0) and Bob (D1,D2,D3,D4) do the experiment, and Alice's data represents a complete dataset, then Bob's data represents the key telling Alice how to read her data properly. (People will disagree with this I'm sure, but I think qualitatively it's not a bad way to explain the behaviour at D0). The information containing fringe patterns must really be there, yet without knowledge of Bob's data, Alice won't be able to extract it. According to this paper: http://arxiv.org/pdf/1009.2404v2, a single-maximum curve will be observed, even if Bob doesn't bother to measure which-path information, i.e. too lazy to set up detectors D_i where 0<i<5. This suggests to me that wave function collapse of both photons is not necessary to already deviate from a classic double slit experiment.

Is all the strangeness surrounding the DCQE experiment simply related to the stochastic nature of QM? That is we can be sure that measurements made on an entangled photon pair will be correlated, therefore whenever/wherever collapse occurs for the second (idler) particle, the total measurements at D0 are correlated with the idler photons and therefore correspondingly distributed because the results of idler photon measurements are also similarly distributed? In this way delaying the idler photon measurement by an arbitrary amount of time makes no difference, so even if they are never measured, the idler photons still affect the entangled signal photons due to correlated probability distributions.

 

[sunrah]

In QM the combination of relativity and QM is called Quantum Field Theory (QFT).

Thanks, QFT is something I wish I had studied. Can you recommend some introductory textbooks?

 

[bhobba]

Thanks, QFT is something I wish I had studied. Can you recommend some introductory textbooks?

Most definitely. Some really good ones suitable for studying after a first course in QM have just started to appear. I have this one and its good:
https://www.amazon.com/dp/019969933X/?tag=pfamazon01-20

Thanks
Bill

 

[Ilja]

The reason is, unlike the paper slips that remain red and green at all times, quantum theory is silent on the proprieties of objects unless observed. That's the crucial difference.

No, that's not a difference. This is simply a side effect of sloppy language, which names some interaction a "measurement", suggesting that what is "measured" depends only on the state of the other system, and not on our own influence.

 

[Demystifier]

No, that's not a difference. This is simply a side effect of sloppy language, which names some interaction a "measurement", suggesting that what is "measured" depends only on the state of the other system, and not on our own influence.

But bhobba did not use the word "measurement". He used the word "observed". Even though you are right that in QM we don't "measure" things, bhobba is also right that we do observe them.

 

[sunrah]

Even though you are right that in QM we don't "measure" things, bhobba is also right that we do observe them.

Would you just expand on this a little? Does a measurement imply the system is already in a given state just before we measure?

 

[Demystifier]

Would you just expand on this a little? Does a measurement imply the system is already in a given state just before we measure?

Yes. The so called "quantum measurements" really change the properties of the system, so they are not really measurements in that sense.

 

[bhobba]

Would you just expand on this a little? Does a measurement imply the system is already in a given state just before we measure?

Yes - sort of.

With decoherence superposition is converted to an improper mixed state. If it was a proper one then it would be in a given state before we measure, but we can't say that with improper states - they may be - but we don't know. There is no way to tell the difference, so we can 'kid' ourselves it is. That's the essence of my ignorance ensemble interpretation.

Measurements are just a special type of observation.

Thanks
Bill

 

[MrRobotoToo]

That's interesting. So if I understand the concept, entangled particles are a superposition state. Measuring one will make the probability distribution of the other conditional on the collapsed state of the measured particle. This conditional probability doesn't depend on either space or time and therefore, seeing as all events take place in spacetime, this conditionality transfer is instantaneous from our point of view.

The article ends by saying this isn't very mysterious at all, but I find it very strange. I guess it is easier to handle, if you accept that an entangled pair is like a single particle?

A lot of the so-called 'mystery' is interpretation-dependent. Make sure to read Appendix B of Gaasbeek's paper: he shows that from the Everettian point-of-view there is nothing at all mysterious about the correlations observed in the delayed choice experiments.

 

[sunrah]

A lot of the so-called 'mystery' is interpretation-dependent. Make sure to read Appendix B of Gaasbeek's paper: he shows that from the Everettian point-of-view there is nothing at all mysterious about the correlations observed in the delayed choice experiments.

If I understand this interpretation, the outcome of any observation here has already been decided before an initial measurement is made. Pre-decided in the sense that the observers have to live in one of the superpositioned-worlds. It is not mentioned what might cause the initial world-splitting event, I guess that might be generating an entangled photon pair?

 

[bhobba]

A lot of the so-called 'mystery' is interpretation-dependent.

.

Very true.

Thanks
Bill

 

[MrRobotoToo]

If I understand this interpretation, the outcome of any observation here has already been decided before an initial measurement is made. Pre-decided in the sense that the observers have to live in one of the superpositioned-worlds. It is not mentioned what might cause the initial world-splitting event, I guess that might be generating an entangled photon pair?

From my admittedly crude understanding, world-splitting occurs when a subsystem which is initially in a superposed state becomes irrevocably entangled with all other subsystems. I'm not going to embarrass myself by trying to give a detailed Everettian analysis of the DCE, but I assume that the entangled photon pair corresponds to the initially superposed subsystem and that the detectors are 'all other subsystems'.

 

[sunrah]

Pre-decided in the sense that the observers have to live in one of the superpositioned-worlds.

Actually they would have to live in both world states, but are not conscious of the existence of the other world. This might explain correlated measurements and wave function collapse, but I don't think it makes things less strange. Going to take a break from this now guys, thanks for all the input.