Causality in Quantum Mechanics: Exploring Meaning & Timing

[entropy1]

I am wondering about the meaning of 'preservation of causality' in quantum mechanics. Is there causality in QM? And does it act back in time? I have some ideas of myself, but want to learn about the general accepted view first.

 

[Bruno81]

In as much as there are 'particles', there's as much causality.

 

[entropy1]

I forgot to give an example: in this publication of Anton Zeilinger et al. he uses entanglement swapping via a Bell-state measurement. It appears that the swap works back in time. It seems to me this could be seen as causality back in time, and thus this might be a violation of causality depending on its definition.

What I want to know is what is ment by 'preservation of causality' in the context of QM. My own view is that there is no causality in any case, only correlations over space and time.

 

[bhobba]

What I want to know is what is ment by 'preservation of causality' in the context of QM. My own view is that there is no causality in any case, only correlations over space and time.

Schroedinger's equation is a partial differential equation hence causal.

Only in some interpretation's is there anything back in time going on in QM.

The issue of causality in QM is purely a matter of what is being talked about. The state is causal - but the results of observations is not necessarily - but we have some interpretations where it is.

Thanks
Bill

 

[entropy1]

Thanks for your reply. I am a layman studying QM on my own. I just got to the Schrödinger equation, so my understanding of it is limited right now. I don't understand what you mean by "Schroedinger's equation is causal" and "The state is causal". Would you like to explain that to me?

 

[bhobba]

Thanks for your reply. I am a layman studying QM on my own. I just got to the Schrödinger equation, so my understanding of it is limited right now. I don't understand what you mean by "Schroedinger's equation is causal" and "The state is causal". Would you like to explain that to me?

Ok. In QM we have this thing called the state - it's a generalisation of the wavefunction you may have heard about. The state and what you are observing determines the probabilities of the observations outcomes. It is a characteristic of QM that all you can predict is the probabilities of the outcomes of observations. If that is inherent in the theory or determined by something else is a matter of interpretation. If it is inherent then its obviously random and not causal. If it is determined by something else and that something else is causal then it may be causal. The theory doesn't say one way or the other.

The state however is determined by a causal equation.

Thanks
Bill

 

[entropy1]

So a 'state' is a (generalized) wavefunction? How can an equation be causal?

For me, the experiment Anton Zeilinger et al. have done shows that due to entanglement swapping (see Fig. 2a for clarity of what I want to discuss here) the cause of a correlation between two particles A(lice) and B(ob) lies in the future (by a Bell measurement by V(ictor)), thereby facilitating retrocausality, if you want to speak of causality in this context of course... You could also speak of correlations and leave the whole thing in the hands of probality. So, indeed, it could be a matter of interpretation; is that what you mean?

 

[bhobba]

So a 'state' is a (generalized) wavefunction? How can an equation be causal?

If initial conditions determine final conditions its causal.

For me, the experiment Anton Zeilinger et al. have done shows that due to entanglement swapping (see Fig. 2a for clarity of what i mean to discuss here) the cause of a correlation between two particles A(lice) and B(ob) lies in the future (by a Bell measurement by V(ictor)), thereby facilitating retrocausality, if you want to speak of causality in this context of course... You could also speak of correlations and leave the whole thing in the hands of probality. So, indeed, it could be a matter of interpretation; is that what you mean?

Correlations do not have to have a cause - they can just be correlations.

Thanks
Bil

 

[entropy1]

If initial conditions determine final conditions its causal.

But still probabilistic?

Correlations do not have to have a cause - they can just be correlations.

Thanks
Bil

Exactly But you could also view them in the light of retrocausality, am I right?

This article of Zeilinger et al. also describes what I mean (see Fig 1).

 

[bhobba]

But still probabilistic?

The state is deterministic, observations are probabilistic.

But you could also view them in the light of retrocausality, am I right?

Yes - that is one interpretation - but a quite uncommon one.

Thanks
Bill

 

[entropy1]

The state is deterministic

By that, do you mean the evolution of the state??

Yes - that is one interpretation - but a quite uncommon one.

Why is it uncommon??

 

[bhobba]

By that, do you mean the evolution of the state??

Yes.

Why is it uncommon??

Its just is. You can't explain taste.

I personally don't like it because its against the spirit of relativity. But that doesn't mean anything.

Thanks
Bill

 

[entropy1]

Thanks.

 

[Heinera]

For me, the experiment Anton Zeilinger et al. have done shows that due to entanglement swapping (see Fig. 2a for clarity of what I want to discuss here) the cause of a correlation between two particles A(lice) and B(ob) lies in the future (by a Bell measurement by V(ictor)), thereby facilitating retrocausality, if you want to speak of causality in this context of course... You could also speak of correlations and leave the whole thing in the hands of probality. So, indeed, it could be a matter of interpretation; is that what you mean?

As far as retrocausality goes, there is nothing in QM that allows you to communicate a message from a future observer to a past observer. For me, that fact settles the question.

 

[entropy1]

As far as retrocausality goes, there is nothing in QM that allows you to communicate a message from a future observer to a past observer. For me, that fact settles the question.

But it seems to be possible to force (cause) (V) a correlation (between A en B) from the present (V) to the past (A&B)...??

 

[bhobba]

But it seems to be possible to force (cause) (V) a correlation (between A en B) from the present (V) to the past (A&B)...??

its possible. None deny that. His issue is the same as mine. It can't be used to send information, so seems strange nature conspired to allow it but prevent that.

Thanks
Bill

 

[entropy1]

its possible. None deny that. His issue is the same as mine. It can't be used to send information, so seems strange nature conspired allow it but prevent that.

If nature would allow that, I think the consequences would probably be enormous, and we would probably live in an entirely different universe!

 

[entropy1]

It just occurs to me, (I don't know very much about it though); if you'd place A en B close together and V far away, and let V 'decide' whether to do a Bell measurement or not, you would have A and B entangled or not! In this scheme, the repeated measuring of entanglement between A en B, yes or no, could in principle send information! But as I understand a Bell measurement requires an 'event-ready' scheme (what is that??), and does not close the fair-sampling loophole!

I can't make out if the random correlation that occurs when there is no entanglement spoils the whole setup though...

I throw this up because I wonder if you guys can tell me why this wouldn't work, so that I gain more knowledge...

UPDATE: In fact, this wouldn't solve anything, because the scheme involves sending entangled photons to V, and V could just as well send a message with photons from V to A&B.

 

[zonde]

But it seems to be possible to force (cause) (V) a correlation (between A en B) from the present (V) to the past (A&B)...??

No, it doesn't seem so. It is like you find out what type of correlation (out of 4 possibilities) is between A and B. So no forcing or causing.

 

[entropy1]

No, it doesn't seem so. It is like you find out what type of correlation (out of 4 possibilities) is between A and B. So no forcing or causing.

I think I don't understand what you mean... A and B are initially not correlated at all, right?

 

[entropy1]

It seems to me you might say that you can't send information, but you can send behaviour.

 

[zonde]

I think I don't understand what you mean... A and B are initially not correlated at all, right?

When you sum these 4 types of correlations they cancel each other out and the sum looks like no correlation at all.

 

[zonde]

It seems to me you might say that you can't send information, but you can send behaviour.

No I say it's enough to view entanglement swapping as finding out information. Nothing has to go back in time.

 

[entropy1]

Thank you for your reply, Zonde. I have to say I currently lack the knowledge to be able to fully appreciate what you are saying. I will have to learn more about QM. But thanks everybody! I learned a lot more today!

Note: if they feel they like to contribute to the discussion, everyone is still invited to respond to this thread of course!

 

[zonde]

I have to say I currently lack the knowledge to be able to fully appreciate what you are saying. I will have to learn more about QM.

It's not so complicated. Let's say that experimenter who performs entanglement swapping get's measurement that says either 1, 2, 3 or 4.
Now if he get's 1 he says photons are correlated in H/V basis and anticorrelated in +45deg./-45deg. basis (rotated measurement)
for 2 - correlation in H/V basis and correlation in +45/-45 basis
for 3 - anticorrelation in H/V basis and anticorrelation in +45/-45 basis
for 4 - anticorrelation in H/V basis and correlation in +45/-45 basis

So depending on the number he gets he splits all photon pairs (at A&B) into four subsets that each show different correlation within that subset. But all subset together show no correlation as all differences cancel out.

 

[entropy1]

This probably shows how poorly I understand QM yet, but: what is "anti-correlation"?

 

[zonde]

This probably shows how little I understand of QM yet, but: what is "anti-correlation"?

Say you measure two photons with both polarizers oriented vertical and one photon passes polarizer but the other one doesn't. And it happens so for all photon pairs that only one photon out of every pair passes the polarizer. That would be perfect anticorrelation.
But if both photons either pass or both don't pass the polarizer it would be perfect correlation.

 

[atyy]

It depends on what one means by causal.

In relativistic QM, one cannot send information faster than light. In that sense QM is causal.

However, because of Bell's theorem, in relativistic QM, the cause of each event is not solely in the past light cone of that event. In this sense, QM is not causal.

 

[entropy1]

Thank you atyy, very clear answer! Thank you, Zonde, very helpful answer! I think I have to understand how a Bell-measurement is carried out!

 

[entropy1]

However, because of Bell's theorem, in relativistic QM, the cause of each event is not solely in the past light cone of that event. In this sense, QM is not causal.

Do these particular events lie outside past/future light cones or can they also lie in the future light cone?

 

[entropy1]

Do these particular events lie outside past/future light cones or can they also lie in the future light cone?

If someone feels ok with answering this, I would probably be helped a whole lot!

 

[DrChinese]

Discussing retrocausality implies an interpretation of QM. Personally, I believe there are a lot of reasons to consider retrocausal interpretations, I think they are given less consideration than they are due. However, an interpretation is still an interpretation and there are none that are currently "scientifically" preferred (as long as they are not local realistic).

So I think I should point out that none of the generally accepted interpretations are "causal" or "deterministic" in any normal fashion. The Bohmian is the closest. MWI and retrocausal (time symmetric) and the others cannot predict the evolution of all elements from any state to any other state. The Heisenberg Uncertainty Principle is always operating to blur the values of non-commuting observables. Know one, you still know nothing of the other. No interpretation fixes this, even the Bohmian doesn't really solve this (even though they try to claim they do). So saying a state evolves deterministically is exactly fair. The values of many observables cannot be explained at all.

So the term "retrocausal" may be a bit misleading, as will similarly be the notion that Bohmian Mechanics is "deterministic". BM inherently cannot provide the answer to a key variable which would need to be known to make it causal in the normal sense of the word.

 

[atyy]

Do these particular events lie outside past/future light cones or can they also lie in the future light cone?

See DrChinese's reply. Retrocausation is a loophole that allows local realism. (Hmmm, I realize DrChinese wrote that local realism excludes retrocausation ... Might be a terminology difference?)

 

[entropy1]

How is 'realism' compatible with 'probabilism'? I mean, in the sense that probabilism is not concerned with 'pre-existing values'? (Unless you take the state as a value)

 

[bhobba]

How is 'realism' compatible with 'probabilism'? I

Flip a coin. The heads and tails are real - the outcome is simply unknown.

Thanks
Bill