Quantum mechanics is not weird, unless presented as such

[stevendaryl]

Isn't this just a problem with the language? The two events are spacelike seperated, but you say that Alice measures before Bob, which is meaningless. I know you said in a certain reference frame, but the notion before is meaningless for spacelike separeated events. If you acknoledge that how do you get a problem with the second answer?

Well, it's interesting if it turns out the interpretation of QM necessarily involves an interpretation of SR, so QM is inherently relativistic. I think that's just a complication that doesn't actually change anything, though. Whether or not there is a meaningful notion of "before" and "after" for spacelike separated events, we can certainly make sense in SR of a spacelike slice of spacetime. I'll have to think about it.

But in any case, to address the claim that QM is not weird by bringing up ways that things are even more complicated to reason about isn't really an argument in favor of non-weirdness. At best, it is a plausibility argument that there might be a way to sort things out.

 

[stevendaryl]

Why do you write ''almost''? It never comes into play since it is not objective.

Well, there have been lots of experiments testing the weirdness, such as tests of Bell's inequality, delayed-choice, etc. So "never" is too strong. But those experiments are sort of peripheral, and outside of the main stream of physics.

 

[stevendaryl]

My question was about something else. If you have no preferred time coordinate it makes no sense to talk about the universe. It is a frame depended consept.

I would say that that claim is slightly circular. The issue is about nonlocality in QM. If there is nonlocality, then that implies that our ideas of SR are either wrong, or that we don't completely understand it. So you can't really use SR as an argument against nonlocality.

 

[A. Neumaier]

If the universe is infinite, it's never isolated.

Isolated just means that nothing flows into or out of the system. Thus a universe without boundary satisfying the usual conservation laws is completely isolated. Something can flow further away but not leave it.

 

[stevendaryl]

Isn't this just a problem with the language? The two events are spacelike seperated, but you say that Alice measures before Bob, which is meaningless. I know you said in a certain reference frame, but the notion before is meaningless for spacelike separeated events. If you acknoledge that how do you get a problem with the second answer?

Another thought about your points. In nonrelativistic physics, the way that we talk about dynamics is:

• Specify the state of the system at time $t_0$
• Use the evolution equations to evolve the state to a future time $t_1$

In relativistic physics, what I thought was that the only change is to generalize this in the following way:

• Specify the state of the system along a spacelike hypersurface.
• Use the evolution equations to evolve the state to a future spacelike hypersurface.

So instead of the issue of "what's true at time $t$", you have the analogous issue of "what's true along such-and-such spacelike hypersurface".

Now, the way that calculations for relativistic QFT are done, the state almost never comes into play. It's there in the background, since the fields of QFT are operators on Fock space. But usually, there is no other state used in calculations other than the vacuum.

 

[A. Neumaier]

the only change is to generalize this in the following way:

• Specify the state of the system along a spacelike hypersurface.
• Use the evolution equations to evolve the state to a future spacelike hypersurface.

There are problems with this, as in the relativistic scenario an observer can prepare only a localized part of the state. This part is then evolved into the common causal future of this localized part. This implies an additional loss of information compared to the nonrelativistic situation. Thus relativistic analyses of few particle experiments are difficult.

the way that calculations for relativistic QFT are done, the state almost never comes into play. It's there in the background, since the fields of QFT are operators on Fock space. But usually, there is no other state used in calculations other than the vacuum.

This is far from true. Even when things are expressed in terms of multiparticle vacuum expectation values, these can always be reinterpreted as matrix elements of few-particle states. Indeed, this is their operational interpretation.

In QFT scattering calculations one needs the (asymptotic) states.

When one has to do dynamical calculations one needs time-dependent states explicitly, and uses them in a Kadanoff-Baym approximation.

The book by Mandel and Wolf on quantum optics is full of computations with non-vacuum states. But most of the time they are not states of fixed photon number but coherent states or squeezed states. (Already preparing states with fixed photon number is an experimental challenge, though it can be done to a reasonable accuracy.)

 

[stevendaryl]

The book by Mandel and Wolf on quantum optics is full of computations with concrete states - but most of the time they are not states of fixed photon number but coherent states or squeezed states. (Already preparing states with fixed photon number is an experimental challenge, though it can be done to a rasonable accuracy.)

Okay, this points out the limitations of my education. When I studied QFT, I only learned to use it for scattering problems where the only states involved were states in the far past and far future. States at intermediate times didn't come into play.

 

[A. Neumaier]

States at intermediate times

See, e.g., J. Berges, Introduction to Nonequilibrium Quantum Field Theory, AIP Conf. Proc. 739 (2004), 3--62.
(preprint version: hep-ph/0409233)

 

[stevendaryl]

Okay, this points out the limitations of my education. When I studied QFT, I only learned to use it for scattering problems where the only states involved were states in the far past and far future. States at intermediate times didn't come into play.

The interesting thing about the two limits $t \rightarrow -\infty$ and $t \rightarrow +\infty$ is that it neatly avoids the issue of "what's the state of a distant particle right NOW". The first limit (far past) is in everybody's far past, and the second limit (far future) is in everybody's far future, so relativity of simultaneity doesn't matter.

 

[A. Neumaier]

it neatly avoids the issue of "what's the state of a distant particle right NOW".

It also precludes any discussion of states at any fixed time, and hence of entanglement experiments.

At finite times - i.e., times where interactions are appreciable so that cluster decomposition (independent bound states at large spatial separation), which is at the heart of the scattering approach, is no longer applicable -, the relativistic particle picture breaks down completely and one needs a field approach, typically in a hydrodynamic or kinetic approximation. These approximations have an interpretation as quantum fluids, with the associated classical intuition but nonclassical interactions featuring quantum corrections. The liquid drop model in nuclear physics (which predicts that iron, Z=26, has the most tightly bound nucleus, which sort of explains the iron core of the Earth) was the earliest approximation of this kind. It is already meaningful for nuclei containing very few protons and neutrons. This indicates again that, in the microscopic domain, the field picture is a better informal picture than the particle picture.

 

[zonde]

Any talk about knowledge. Knowledge requires a subject, hence turns the problem into something subject-dependent = subjective.

Let us modify the setting a little. A machine makes the decisions and records the responses in the experiment while Alice sleeps. Next morning Alice wakes up and reads the records of all decisions and responses. At this moment her knowledge changes and the states of all the photons collapse, long after they stopped to exist. It is obvious that in this setting the states only refer to Alice's knowledge not to any underlying physical reality. Here approximation $\rho_A$ to the true physical state $\rho$ improves as she draws the consequences of having obtained new information.

Do you understand that we can't cut out "any talk about knowledge" as that's a bit too much, right? You have to be more specific. But in particular context I suppose I can guess what you mean by that.
Ok suppose we can cut out the talk about knowledge of Alice or Bob about remote states. But we can't cut out any talk about detection records, do you agree?

 

[A. Neumaier]

Do you understand that we can't cut out "any talk about knowledge" as that's a bit too much

No, it is not too much, but precisely what is needed to make quantum mechanics objective. My whole book on quantum mechanics is quite close to applications but mentions knowledge only in two contexts:

• in a section on modeling, which is done by a human,
• in two sections on information theory, where I criticize basing the foundations of physics on a notion of knowledge.

The only kind of knowledge Nature knows about (and hence may be mirrored in a theory and its discussion) is the one directly encoded into the equations and the initial conditions. Any other talk of knowledge is anthropomorphic and leads to problems.

 

[A. Neumaier]

At finite times, the relativistic particle picture breaks down completely

This holds already classically. See the discussion in https://www.physicsforums.com/posts/5183296/

 

[A. Neumaier]

But we can't cut out any talk about detection records, do you agree?

Detection records define the results of experiments, and are indeed needed to interpret how Nature responded to the experimental setting.

 

[zonde]

No, it is not too much, but precisely what is needed to make quantum mechanics objective.

Objective is property of knowledge. There is not much sense of "objective" that is independent from knowledge.

 

[martinbn]

@A. Neumaier : Now I am even more confused. It seems that by the universe you mean the space-time!

@stevendaryl : I didn't mean to say your example was not good for showing weirdness, I am just worried if there might be something that isn't there just because of the phrasing. I don't think the intial valiu problem is relevant here, because the events are still in such causal relationship, that we cannot talk in an invariant way about which is first and which is second. This may be just me not being able to see the forest because of the trees.

 

[zonde]

Detection records define the results of experiments, and are indeed needed to interpret how Nature responded to the experimental setting.

I think this can be used as common basis to reach some agreement. I would agree that collapse is not really needed if it does not follow from detection records.
So would you agree that collapse is needed if it follows from detection records without involving subjective knowledge of Alice and Bob? (I am not proposing to start discussion on the subject here.)

 

[A. Neumaier]

Objective is a property of knowledge.

No. objective is what is the case in Nature.
Knowledge depends on who has the knowledge, making it subjective. It is precisely as objective as it agrees with the true state of Nature.

Even in science, what counts as knowledge changes in the course of new research, otherwise our scientific knowledge couldn't grow with time. This shows that is is to some extent subjective.

Since we do not know the true state of Nature we replace in our Gedankenexperiments and our analyses of real experiments Nature by a (drastically simplified) quantum mechanical model of it. Then for the purposes of the Gedankenexperiment or analysis, objective is what is the case in the quantum mechanical model - not what the (Gedanken)people involved in the experiment know about it.

 

[A. Neumaier]

It seems that by the universe you mean the space-time!

The space-time including all the quantum fields in it, and the Heisenberg state that defines the QFT expectations. (Or, actually, the reality that is described by this, but in a scientific discussion, this is indistinguishable from it since talking about items in the universe we have to address them by their physical identity.)

 

[A. Neumaier]

I think this can be used as common basis to reach some agreement.

OK, so start form here. As to your other question, I don't invest time in counterfactual reasoning.

 

[zonde]

No. objective is what is the case in Nature.
Knowledge depends on who has the knowledge, making it subjective. It is precisely as objective as it agrees with the true state of Nature.

Even in science, what counts as knowledge changes in the course of new research, otherwise our scientific knowledge couldn't grow with time. This shows that is is to some extent subjective.

Since we do not know the true state of Nature we replace in our Gedankenexperiments and our analyses of real experiments Nature by a (drastically simplified) quantum mechanical model of it. Then for the purposes of the Gedankenexperiment or analysis, objective is what is the case in the quantum mechanical model - not what the (Gedanken)people involved in the experiment know about it.

I agree with all this except some semantics related things. But I don't want to continue this semantics discussion if it seems that I can't make the point.

OK, so start form here. As to your other question, I don't invest time in counterfactual reasoning.

You don't do that at the level of detection records. But you use counterfactual statements at statistics level. When you spell out quantum mechanics prediction you are making counterfactual statement.

 

[zonde]

OK, so start form here. As to your other question, I don't invest time in counterfactual reasoning.

I suppose that it should be possible to formulate minimal interpretation without collapse (if we refuse to go into any details about detection events and how we get statistics) but then we have to stay agnostic about locality or non-locality as there would be no real connection to individual spacetime events.

 

[A. Neumaier]

When you spell out quantum mechanics prediction you are making counterfactual statement.

Don't talk about 'you' when you mean yourself; it sounded like you addressed me. But when I spell out quantum mechanics prediction I don't make counterfactual statements.

It is easy to avoid any subjunctive form (and hence any counterfactual statements) in the discussion. You can make case distinctions if you need.

it should be possible to formulate minimal interpretation

I am not interested in discussing subjunctive statements. Do it, or don't claim it!

 

[A. Neumaier]

to stay agnostic about locality or non-locality

If one has to stay agnostic about (non)locality in the absence of invoking subjective knowledge, it is proof of my earlier assertion that nonlocality is a purely subjective feature of the interpretation, without any objective content.

 

[zonde]

Don't talk about 'you' when you mean yourself; it sounded like you addressed me. But when I spell out quantum mechanics prediction I don't make counterfactual statements.

It is easy to avoid any subjunctive form (and hence any counterfactual statements) in the discussion. You can make case distinctions if you need.

I think I was a bit sloppy.
If you make two mutually exclusive predictions and then test one of them the other one becomes counterfactual statement.

 

[A. Neumaier]

If you make two mutually exclusive predictions and then test one of them the other one becomes counterfactual statement.

No. In this case you made two predictions, tested one of them, and remain silent about the validity of the other prediction. I see nothing counterfactual in this.

It becomes only counterfactual if you start arguing after the test ''If I would have done...'' - but such arguments must also be cut out since they are not testable. Testable are only statements about the future.

Counterfactual statement may be useful for planning an experiment but have no place in an objective analysis of the results.

 

[lightarrow]

Who is Alfred?

Sorry. Arnold.

--
BlueRay

 

[zonde]

Counterfactual statement may be useful for planning an experiment but have no place in an objective analysis of the results.

Right

No. In this case you made two predictions, tested one of them, and remain silent about the other prediction. I see nothing counterfactual in this.

I remain silent about factual truth or falsity of that other untested statement.

It becomes only counterfactual if you start arguing after the test ''If I would have done...'' - but such arguments must also be cut out since they are not testable. Testable are only statements about the future.

Counterfactual statements can't be tested, that's right.
But "what if" type of reasoning is irreplaceable tool in consistency checks (thought experiments). Say if I intend to subject physical configuration to external parameter "a" theory makes prediction "A" but if I intend to subject physical configuration to external parameter "b" theory makes prediction "B". Now I want to check that description of physical configuration from first case is consistent with description of physical configuration from second case. For this I need "what if" type of reasoning. And consistency is a valid requirement for a theory.
So we can analyze a theory from perspective of consistency. Say we can add something to the theory or remove something from it and look if it remains consistent.

 

[A. Neumaier]

Say we can add something to the theory or remove something from it and look if it remains consistent.

But this means that you make different predictions of the same situation with two different models. There is no need to use counterfactual language, which is inappropriate in science.

I remain silent about factual truth or falsity of that other untested statement.

Yes, and this removes any counterfactuality. it is just an irrelevant prediction, so it doesn't need to be discussed - and in science, it shouldn't.

 

[zonde]

A. Neumaier,
But science requires that theories are consistent. So it has to include tools to check this consistency.
Edit: I mean self-consistent.

 

[A. Neumaier]

So it has to include tools to check this consistency.

The scientific tools for this are mathematical deductions (where logic guarantees consistency) and the comparisons of experiments with predictions that assume only what was actually prepared.

Counterfactual comparisons are logically meaningless since they neither apply to reality (where the link has to be made in ordinary language) nor have the formal stringency of a mathematical argument (where one can decide consistency objectively). They are one of the traditional backdoors that allow weirdness to creep into an otherwise rational and beautiful theory.

 

[ddd123]

But this means that you make different predictions of the same situation with two different models. There is no need to use counterfactual language, which is inappropriate in science.

Yes, and this removes any counterfactuality. it is just an irrelevant prediction, so it doesn't need to be discussed - and in science, it shouldn't.

Sorry for my naughty thinking, but it seems you're saying that just because counterfactuals are a big part of the weirdness, not because of a rock-solid reason. In another universe without QM you could be saying counterfactuals are absolutely needed for consistency for whatever reason. Though making theories not-weird is not science's purpose, the purpose is to seek the truth about nature. If weirdness is the truth (not saying it is, but we don't know that it isn't either) it wouldn't be a good idea to sweep it under the rug.

 

[A. Neumaier]

In another universe without QM

You use counterfactual reasoning to justify counterfactual reasoning. Isn't this already weird? (As I argued earlier, there is only one universe to which we have access.)

 

[ddd123]

Because, if anything, this way of thinking has been a propeller for science, like Einstein's Gedanken experiments (of which, EPR was intended as a reductio, so that's already a purposefully impossible envisioning, though in reality and not in purpose it ended up being true). If we declare we've solved, say, the measurement problem and sit down, while our resolution isn't fully satisfactory but we've just deluded ourselves into thinking it is, we're stymieing possible groundbreaking ideation in the future.

 

[A. Neumaier]

Einstein's Gedanken experiments

Good gedanken experiments like Einsteins are not counterfactual. Making simplifying assumptions in a situation where better models do not exist or are intractable is not meant counterfactual, but is the usual state of affairs where we substitute the always unknown true model of reality by an approximate model. It becomes counterfactual only if you change the model to try to make a point about reality.

If you change the model to make a point about different predictions there is nothing counterfactual: It is legitimate to say that two models predict X and Y, experiment shows X, hence the model predicting Y is ruled out. This is correct science, no counterfactual language is involved,