What counts as a measurement in QM?

[evk]

What counts as a "measurement" in QM?

"Measurements" in quantum mechanics seem to have an important role. They collapse the Schrodinger wave equation, for example (please correct me if that's no longer the modern view). The state of a system BETWEEN MEASUREMENTS is what is inherently probabilistic and therefore subject to interference phenomena (as in the two-slit experiment).

My question is, what counts as a "measurement"? Is it ANY interaction of the system being analyzed with the outside world? Surely it can't matter whether or not a conscious entity is involved in the observation, right? In Schrodinger's thought experiment with the cat in the box, can the system only remain in a "fuzzy" state so long as absolutely no interaction takes place between the box and the outside world? Suppose a physicist with measurement devices was inside the box instead of a cat, are you really telling me the analysis would then be different? How must one draw the dividing line between the system being analyzed and "the rest of the world"?

Thanks

 

[jcsd]

The postulates of quantum mechnics tell us the possible results of a measuremnt and they tell us the state the system will be in immediately after a measurment, but they never bother to tell us what a measurement is!

The simple answer is that nobody has given an entirely satisfactory answer to your question, that's why it's know as "the measurement problem".

 

[vanesch]

The simple answer is that nobody has given an entirely satisfactory answer to your question, that's why it's know as "the measurement problem".

I second that

If you want to see some discussion about it here, have for instance a look at the thread
https://www.physicsforums.com/showthread.php?t=96231

Happily that we have this measurement problem. It is what keeps discussion forums on quantum physics alive

 

[lalbatros]

interaction with a macroscopic system

isn't that a measurement ?

Macroscopic means a practically continuous set of possible states.
This brings the possibility for irreversible transitions.

This goes back to a trivial fact: discrete Fourier series cannot represent an exponential decay, but Fourier transforms can because frequencies in the Fourier transform expansion can be as close as necessary to produce to long-term behaviour of the exponential. Similarly, a discrete system in a mixed state can decay when interacting with a continuum. There is still micro-reversibility, but the 'small' system has been 'measured' irreversibly.

Developping the maths for that doesn't bring more than this simple conclusion.
It is nevertheless interresting to read the discussions on coherence and decoherence.
There, the point of view is complementary: how is it possible to isolate a small system as long as possible from external perturbations is such a way as to keep it in a coherent state!
Obviously, by studying this, one aknowledge that interactions can easily 'decohere' a system, leaving it in a 'incoherent' state. An incoherent state cannot be represented by a wave function, of course, but it can be represented by probabilities.


Is there really a 'measurement problem' ?

 

[vanesch]

Is there really a 'measurement problem' ?

It depends on whether you think that there is contained, in QM, an ontological state description (the only thing I see that qualifies is the state vector in Hilbert space) or whether you simply consider the "state description" in QM as epistemological (a summary of your knowledge of the system under study).
In the last case, there is no measurement problem ; however, there is a "real world out there" problem . You've lost the link between your mathematical objects in your theory and "the world out there" ; as such, the association between laboratory experiments and calculations seems impossible to me: at a certain moment you HAVE to associate physical objects (in the lab) with mathematical objects in your theory (this association is what I'd call ontological), and then we're back to case 1.
To try to make you see what I mean, consider classical physics. You may define a probability distribution over phase space of your system. That's a purely epistemological description. It is sufficient to LEARN something about the system (for instance, by LOOKING at it), and the probability distribution changes (because your knowledge of it changes). At no point (except a few weirdos around here ) you would consider that the system by itself IS the probability distribution. But there's no harm. You know that the state of the system (the ontological state, which exists, independent of what you know of it) is a point in phase space. You know that your epistemological description is a complete or incomplete knowledge of that point in phase space.
If you go to QM, the equivalence of your point in phase space is your statevector in hilbert space, and the equivalence of your probability distribution over phase space is your density operator.
However, QM has a property that classical physics hasn't, and that is entanglement: you cannot decide to assign a quantum state in hilbert space to subsystems. If a system interacts with another, the quantum state is only given in the product space, and the individual quantum state of the subsystem does not exist. You can now "coarse grain" and trace out the part of hilbert space that doesn't interest you, and you'll end up with a reduced density matrix. But I don't think you can call that the ontological state of the system, because it does WRONGLY describe correlations with the other system you decided to ignore (EPR style correlations). It DOES give you a good epistemological description of what you decided to consider (to know) of the system.

 

[lalbatros]

Vanesh,

If I understood well, you don't like the idea that the state of the 'measured' system depends on the action (point of view?) of the observer. But I think this is not the case, after the measurement-interaction, there is only one big universe wavefunction, just as before the measurement. The only difference is that the interaction has made it irreversibly different: very close to ideally collapsed states but not exactly. In principle, the correlations have not vanished and could be recovered by time-inversion. Practically, there is no way to perform this time-inversion (because all evolution in the universe would ultimately need to be time-inverted). The strange object we need to get familiar with is a large scale wavefunction.

This is just -for me- the same problem than for classical mechanics and the second principle. It is a difficult subject, but there is agreement that micro-reversibility is not contradictory to the second principle. And the second principle is compatible with the classical micro-description of the world.

Of course, there are differences too. Like the EPR question. Like the fact that in QM the irreversibility deals with a micro-system interacting with a macro system, while in CM it is macro-macro.

My feeling is that these two question are the same and should have a common answer. This makes QM measurement less a new problem, and more like an old problem.

 

[vanesch]

Vanesh,

Why is everybody here forgetting the "c" in my name ?
(my full name is Patrick Van Esch, and Esch is a little town in Luxemburg which seems to be the genealogical origin of it).

If I understood well, you don't like the idea that the state of the 'measured' system depends on the action (point of view?) of the observer. But I think this is not the case, after the measurement-interaction, there is only one big universe wavefunction, just as before the measurement.

I agree with that ! There has been a measurement interaction (the unitary time evolution operator attached to the physics of the measurement apparatus, the observer body and all that). I just have difficulties considering that an "observation" has *killed off* a big part of the entire wavefunction of the universe which is what standard state reduction prescribes.

The only difference is that the interaction has made it irreversibly different: very close to ideally collapsed states but not exactly. In principle, the correlations have not vanished and could be recovered by time-inversion. Practically, there is no way to perform this time-inversion (because all evolution in the universe would ultimately need to be time-inverted). The strange object we need to get familiar with is a large scale wavefunction.

Yes, that's also my POV. Except that I consider that *before* measurement, that large scale wavefunction was in a product state (or at least the relevant part of it was in a product state) of an "ignorant observer" and the "pure system", while the measurement interaction entangled both.

This is just -for me- the same problem than for classical mechanics and the second principle.

Well, it is slightly more subtle, in that, contrary to classical systems, there IS no individual state description of an entangled system, while you can always consider the projected sub-phase space in classical mechanics as describing de state exactly of the subsystem.

It is a difficult subject, but there is agreement that micro-reversibility is not contradictory to the second principle. And the second principle is compatible with the classical micro-description of the world.

Sounds like music in my ears

Of course, there are differences too. Like the EPR question. Like the fact that in QM the irreversibility deals with a micro-system interacting with a macro system, while in CM it is macro-macro.
My feeling is that these two question are the same and should have a common answer. This makes QM measurement less a new problem, and more like an old problem.

To me, the essential part of the measurement problem remains: how do you go from a SUM of states (this time, entangled and all that), to ONLY ONE with a certain PROBABILITY ? This cannot be done with a unitary operator.
With a unitary operator you can mix states, have a Schmidt decomposition with classical-looking states and everything, but at the end of the day, YOUR BODY ends up entangled if all is unitary. Each of these body states is in a product state with a classically looking universe if we are optimistic (using environmental decoherence) but it is STILL in SEVERAL states. YOU, as a conscious being, only PERCEIVE ONE of these states. *that* is puzzling. You ought to be seeing all of them, because your body sees all of them. It is the essence of the measurement problem, to me.
If, in classical physics, say, green light is coming from the left and red light is coming from the right, you are aware, you perceive, you are conscious of BOTH the green light and the red light, because that will be your body state (the ontological one, in the phase space of your body).
In quantum theory, in that big wavefunction of the universe, your one and the same single body (brain) HAS seen a click in one term, and HASN'T seen a click in the other. Both state vectors are there in your body state. But you only PERCEIVE one of them, with a certain PROBABILITY. How come ?

 

[lalbatros]

Patrick, sorry for the 'c' !

I don't understand your conclusion:

To me, the essential part of the measurement problem remains: how do you go from a SUM of states (this time, entangled and all that), to ONLY ONE with a certain PROBABILITY ?

, since you admit that the all-whole-wavefunction evolves as usual, in an unitary way.

Indeed, assume that after the measurement (analyser) you perform a second analysis (like with a crossed-analyser). You could model the outcome of the second experiment in two ways:

• either start from a density operator representing the collapsed state of the measured system

• or use the all-whole-wavefunction that includes the small system and the macro-measuring device entangled wavefunctions

The two models will give exactly the same result.

They could give different results only if your second analyser is able to 'unfold' the smeared correlations from the first measurement. And this is probably impossible for real-world measurements. However, this is may not be impossible for nano-scale interactions (is it called meso-scale maybe?), a scale where correlations are long-lived, not too much perturbed.

Sorry also for the musics, which sounds like poor knowledge. I don't know much more than classical mechanics, the Boltzmann H-theorem, and its holy proof. This is an uncomfortable region of physics for me. I hadly see a link between the H-theorem and the second principle. I hardly can understand the proof. And I cannot see any generality in this theorem. Nevertheless, this H-theorem is taken as the cornerstone for understanding irreversibility from reversibility.

In a sense, I see an easier track in QM, since I would analyse the density operator and think in terms of the correlations. And irreversibility would only be an excellent approximation of a state vector by a state density matrix.

Why should we be perturbed by two models that agree so easily?

Concerning you final remark:

In quantum theory, in that big wavefunction of the universe, your one and the same single body (brain) HAS seen a click in one term, and HASN'T seen a click in the other. Both state vectors are there in your body state. But you only PERCEIVE one of them, with a certain PROBABILITY. How come ?

Why do you imagine such complicated scenarios?
You will never be able to look at the universe from the outside!
Maybe this could be useful to develop new physics, but the predictions will only be compared to what we can see from the inside! If the model looks strange on the cover should not be important!?

 

[straycat]

... YOU, as a conscious being, only PERCEIVE ONE of these states. *that* is puzzling. You ought to be seeing all of them, because your body sees all of them. It is the essence of the measurement problem, to me.

You might be interested in Saunders' treatment of this subject in his paper [1] "Time, Quantum Mechanics, and Probability", section 4.2 . He argues that you should *not* expect to "see all of them," but rather you should expect (prior to measurement, at time t_1) to see *one* of them (after measurement, e.g. at time t_2), you just don't know which one.

David

[1] http://xxx.lanl.gov/abs/quant-ph/0111047

 

[vanesch]

Patrick, sorry for the 'c' !

No problem but about EVERYBODY here writes vanesh (probably imagining I'm of Indian origin ?).

since you admit that the all-whole-wavefunction evolves as usual, in an unitary way.
Indeed, assume that after the measurement (analyser) you perform a second analysis (like with a crossed-analyser). You could model the outcome of the second experiment in two ways:

• either start from a density operator representing the collapsed state of the measured system

• or use the all-whole-wavefunction that includes the small system and the macro-measuring device entangled wavefunctions

The two models will give exactly the same result.

I know ! But you're cheating here: you're USING the Born rule to calculate probabilities. You do that already when doing the partial trace for the reduced density matrix: WHY do you think that you can use the partial trace ? Because you sum then over the Born probabilities of the outcomes (in any basis, trace is basis-independent) of the outcomes for the environment or the system you're not interested in. But that means you ALREADY use the Born rule, so you cannot use that to DERIVE the Born rule (circular reasoning).

They could give different results only if your second analyser is able to 'unfold' the smeared correlations from the first measurement. And this is probably impossible for real-world measurements. However, this is may not be impossible for nano-scale interactions (is it called meso-scale maybe?), a scale where correlations are long-lived, not too much perturbed.

I only mentionned that because I had the impression that you considered the (reduced) density matrix as the *ontological* state of your subsystem (in which case these correlations would be unexplainable!). If you accept that the ontological state of the system is always given by a vector in Hilbert space, then we are in agreement on this. (the epistemological description "state" can be a density matrix: it writes down what you KNOW of the system, not what IS the system).

Concerning you final remark:
Why do you imagine such complicated scenarios?
You will never be able to look at the universe from the outside!
Maybe this could be useful to develop new physics, but the predictions will only be compared to what we can see from the inside! If the model looks strange on the cover should not be important!?

I don't have to look from the outside. In classical physics, you don't have to "look from the outside" to consider that there is a big phase space of all stuff in the universe, that you, as part of that universe, are also in a "state" (the chunk of phase space that corresponds to the degrees of freedom of your body). And THAT state of that body of yours will then determine what you will consciously experience (how that happens is then left as an exercise for philosophers and doesn't concern any physics as such). The FULL STATE of your body will determine what you will experience. But that's so evident in classical physics that nobody writes that down. You're dead when the ENTIRE STATE of your body satisfies certain conditions (like being splattered all over because you've been hit by a truck or so). You're alive when your body is in another state...

However, the ENTIRE STATE of your body, in the quantum description of the universe, is made up of different terms in the wavefunction of the universe. In certain terms there is a contribution of your bodystate which corresponds to looking at some birds in the field, in other terms you're splattered all over because while looking at the birds you didn't see the truck, in still other terms you decided not to go for a walk and you're chatting on PF...
This is the entire, ontological, state of your body: ALL these different terms are present, some with an important Hilbert norm, others with less. Apparently you only consciously observe ONE of these terms. With a probability given by the Born rule. And it is because of this probability given by the Born rule for your mind to see ONE of the states of your body who are present, that you ALSO find the Born rule for experiments in the lab and everything.

So the question is: why do you not experience the full quantum state of your body as it is present in the wavefunction of the universe ?

I answer this simply by POSTULATE, but you should be aware that this is a strange postulate, saying something about the conscious connection between a body state and a mind. And if you DON'T do that, I really, really don't see why I'm only aware of part of the state of my body and how this can be *derived* from the unitary evolved state (the wavefunction of the universe).

 

[jcsd]

At a more proletarian level: Bog standard quantum mechanical measurement theory does not consider the measuremnt appartus and the particle (or whatever) as a single quantum mechanical system. However when you do you still have the same basic problem, for example:

Consider a quantum mechanical system consisting of an electon in a superpostion of states (spin up, spin down) and some mesuremnt appartus; when the measuremnt apparatus measures the electron decoherence occurs. As now a measuremnt of the measuremnt apparatus-electron system will always correspond to a measuremnt having occurred and the electron not being in a superpostion of states, you may well be tempted to say "problem solved!" There are now two quite distinct 'branches' to the wavefunction one corresponding to spin up and one corresponding to a spin down measurement, however the system is still in a superpostion of states, so we still have the same basic problem: "Why when we measure the electron do we only get a single state?"

 

[ZapperZ]

What if you measure a non-commuting observable and deduce the superposition that way? I.e. [A,B] != 0, so to see the superposition of observable A, you measure B and deduce from the "weird" property of B that you have just detected the superposition of A. Example: energy gap measured in Stony Brook/Delft experiment.

Zz.

 

[jcsd]

But surely, though in the formal sense we have performed a measurement of some sort, we have not (again in the strictly formal sense) performed a measuremnt of A as the result of the measuremnt is not an eignevalue of the operator A.

 

[ZapperZ]

But surely, though in the formal sense we have performed a measurement of some sort, we have not (again in the strictly formal sense) performed a measuremnt of A as the result of the measuremnt is not an eignevalue of the operator A.

Correct. But if we want to preserve the superpostion of A, this is one way to "observe" it. I didn't say we were going to measure A. I simply pointed out how to detect the superpostion of A.

Zz.

 

[-Job-]

Can't the collapse of a particle's wave function collapse the wave function of another particle? By observing B are you collapsing both B and A?

 

[vanesch]

Can't the collapse of a particle's wave function collapse the wave function of another particle? By observing B are you collapsing both B and A?

This is the problem (?) with quantum mechanics as we know it: ALL interaction is described by a UNITARY operator. And with such an operator, you cannot do the collapse thing which is a projector. No matter how big and complicated the system is. That's the fundamental difficulty: the "collapse" operation is mathematically not compatible with a unitary operator. The best we can do is environmental decoherence, but that is STILL a superposition. It explains why we don't see INTERFERENCE but it doesn't explain why we only see ONE term.

A naive analogy: imagine, that after a lot of calculation in a new theory about light boxes, you find that out of a certain box with two holes must come red light out of the left hole and green light out of the right hole. When you do the experiment, however, sometimes there comes green light out of the right hole, and in other instances of the experiment, there comes red light out of the left hole.
So it seems that your calculation only predicted some STATISTICS of what was to happen. But no, the thing that corresponds to the "state of the box" said "red AND green light". It wasn't meant to be a description of an ensemble of boxes.
Then people say: hoho, but if you have red and green light, that must give you funny interference patterns! And then others do detailled calculations and show that for all practical purposes, you will NOT see interference patterns between the red and the green light, because of the influence of the environment. THIS is the equivalence of "environmental decoherence".

So, problem solved because you won't observe interference between red and green light ? I wouldn't think so: your theory said that you should see BOTH, and you only see one at a time, with probability 50%. So you introduce an extra postulate: whenever I OBSERVE a box which sends out red and green light, I only observe ONE of these states, probabilitically. But the box is in the state "red + green" allright. You can even check that, because when you do a careful experiment AVOIDING the environment, you DO see interference between red and green light. So it really WASN'T a statistical mixture which you erroneously took to be the state of a system.

This is an analogy of what we do when we consider the wavefunction of the universe, and then say that we "see only one term".

 

[ZapperZ]

Can't the collapse of a particle's wave function collapse the wave function of another particle? By observing B are you collapsing both B and A?

If you look at the mathematics carefully, you will see that if A and B do not commute, a measure of the observable A says nothing about the outcome of B. Only when A and B commute are you able to determine both quantities in a single measurement (for non-degenerate states). For example, Lz does not commute with Lx and Ly. A measure of Lz will STILL leave Lx and Ly values undetermine, preserving their superposition.

This is a very fundamental aspect of QM. In fact, the commutator relation [A,B] has often been called as the First Quantization. It is the origin of the uncertainty relations in QM, and the source of the so-called "measurement problem".

Zz.

 

[lalbatros]

Vanesch

I suggested that a pure-state POV or a mixed state POV may lead to exactly the same results when correlations are 'smeared'.

I don't understand your reaction:

I know ! But you're cheating here: you're USING the Born rule to calculate probabilities

The interpretation of the Born rule, for me, is just that a pure-state and a density matrix can represent very closely the same system and predict its evolution with very much the same precision. In such a situation, the Born rule is a useful language.

But is the Born rule really necessary ? Since the approximate equivalence has the same meaning and is easily integrated in the formalism without the Born rule. Maybe the Born rule is a simple matter of convenience ...

Could we not try without the Born rule?

 

[lalbatros]

Vanesch,

I almost agree with your remark:

... QM has a property that classical physics hasn't, and that is entanglement ...

But then, I think that classical space is shaped by particles (GR, Mach principle). And how could I assign a coordinate to a particle alone ? In a sense, this is already a kind of entanglement: it is an impossibility to talk about a part without considering the whole. But this is less a problem in CM, maybe because of the type of systems/interactions being studied.

So, then, what is non-classical with QM entanglement?

One aspect is for sure linked to the EPR paradox. And I don't see another.
The non-locality is the difficulty here. The Born postulate is not the difficulty.
That is to say that the 'measurement postulate' is not shaken by the question of entanglement.
But maybe the classical concept of space-time will be shaken sometimes by this question.

 

[Careful]

You cannot do without the Born rule:
In copenhagen and in consciouness it is explicitely there (this is the AND formalism).
In environmental decoherence (the OR formalism) there is nothing in the density matrix which tells you how to interpret it. There is for example no a priori reason why the different pure states (which span the density matrix as a convex combination) should be orthogonal; this is an extra postulate (and even if you would adapt this convention you still don't have a unique decomposition when equal weights are present). Moreover, as far as I know, all these so called decoherence proofs show is that *IF* the environment decoheres, *THEN* the environment + system decoheres given a reasonable interaction Hamiltonian (I may be wrong here, but I do not think so).


Concerning your second comments: coordinates in GR have nothing to do with physics, they are a mathematical artifact. Now, you might wonder about assigning physical coordinates to matter; this is an act of measurement and depends upon a priori hypotheses about (a) your measurement apparatus and (b) space time structure.
However, it has nothing to do with entanglement ! On the contrary, it is a famous and painfully correct statement of A. Einstein that ``if everything were entangled in the universe, then it would be impossible to find out any laws for it´´ and indeed our ability to construct laws of nature depends upon the assumption that isolated systems exist in nature (something which is not possible in QM). This is another ``translation´´ of the measurement problem: show why the world disentangles at reasonable scales of 10^{-8} meters.

To shake the classical concept of space time (which has a priori nothing to do with entanglement) is one of the most dangerous and uncontrollable games in town. People should have more and deeper respect for the principle of LOCALITY since the alternative is a 80 years old untamed monster which hasn't been shown by any means to be NECESSARY, albeit its apllications to microphysics are succesful.

Cheers,

Careful

 

[vanesch]

Could we not try without the Born rule?

I would like to point to a small paper that I've written on the subject:
quant-ph/0505059

It was refused by Proceedings of the Royal society, but not on grounds of being incorrect, but on grounds that the basic idea was already published before by Barnum and co (while it was recognized that my little paper was technically clearer but that this by itself was no sufficient reason to warrant a new publication).

As long as you take the pure state in the Hilbert space as being the state of the system (basic postulate in QM, no ?) then the only way of interpreting the diagonal elements of the density matrix as probabilities is by using the Born rule. The reason is that the (total) density matrix of a pure state, under unitary evolution, remains a pure state, and that the *reduced* density matrix is obtained by taking the partial trace, which finds its origin in the Born rule.
Now, writing down the density *matrix* (IN A SPECIFIC BASIS OF COURSE), and calling the diagonal elements "probabilities in a mixture" IS THE BORN RULE.

It suddenly makes the equivalence between a pure state and a statistical mixture. A pure state will NOT have a density matrix with non-diagonal elements zero (except if it is ONE of the basisvectors in which we write down the matrix). A mixture will. So the measurement PUTS THE NON-DIAGONAL ELEMENTS TO 0, which clearly is a non-unitary evolution, and if you do it too early, you simply make WRONG predictions (eliminating quantum interference).
The reduced density matrix only *mimics* this, because in order to obtain it, we already *used* the Born rule in the partial trace. It is only a tool.

I know that the density matrix formalism obscures the measurement problem because the language between the statistical description and the state description is mixed up. But each time you go and dig, you find the Born rule hidden somewhere. At a certain point, you associate hilbert norm and probability.

 

[Sherlock]

... it is a famous and painfully correct statement of A. Einstein that ``if everything were entangled in the universe, then it would be impossible to find out any laws for it´´ and indeed our ability to construct laws of nature depends upon the assumption that isolated systems exist in nature (something which is not possible in QM). This is another ``translation´´ of the measurement problem: show why the world disentangles at reasonable scales of 10^{-8} meters.

Correlations which qualify as quantum entanglement can be more or less reproduced by more or less reproducing the experimental environments. But whatever is happening at the level of the quantum interactions themselves still seems to be a mystery. So, the measurment problem is that the physical reality (at least the details) of whatever is being measured is still largely unknown. There's no unambiguous way to apprehend its precise qualitative features. No way to control the outcomes of individual measurements. A measurement problem is inevitable. There's a theoretical 'measurement problem' because there's a real physical *measurement problem*. And yet some discussions (not necessarily here) continue as if the whole thing is going to be resolved by some shift in the theoretical approach.

--------------

It doesn't seem to me that quantum theory forbids isolated systems -- just isolated subsystems of a system. Is entanglement a matter of context? That is, wrt the motion of the solar system as a whole, then is everything in the solar system entangled wrt to that motion?

Do the effects produced by two interfering water waves, or by two different locations of a single water wave front, qualify as being entangled? Are em signals transmitted by a local news station at a time t1, and received at a time, t2, on many tv sets tuned to that channel, entangled? Are the simultaneous receptions of the show entangled?
In some sense maybe? But these things aren't quite the same as quantum entanglement, although there do seem to be similarities. What then distinguishes quantum entanglement? Is it that the *essence* of quantum entanglement can be understood in terms of analogies to macroscopic phenomena, but that the details are missing?

What did Einstein mean by "entangled" in the statement you cited?

 

[Careful]

It doesn't seem to me that quantum theory forbids isolated systems -- just isolated subsystems of a system. Is entanglement a matter of context? That is, wrt the motion of the solar system as a whole, then is everything in the solar system entangled wrt to that motion?
Do the effects produced by two interfering water waves, or by two different locations of a single water wave front, qualify as being entangled? Are em signals transmitted by a local news station at a time t1, and received at a time, t2, on many tv sets tuned to that channel, entangled? Are the simultaneous receptions of the show entangled?
In some sense maybe? But these things aren't quite the same as quantum entanglement, although there do seem to be similarities. What then distinguishes quantum entanglement? Is it that the *essence* of quantum entanglement can be understood in terms of analogies to macroscopic phenomena, but that the details are missing?
What did Einstein mean by "entangled" in the statement you cited?

What a series of questions for a european sunday morning
(a) Obviously, the problem concerns isolated substems of a closed system. The only closed system is the entire universe. Jim Hartle has written good stuff on that.
(b) Let me clarify some entanglement issues to you:
(I) CLASSICAL : all matter fields and gauge fiels are REAL. Entanglement between two particles would mean that results of separate, but identical measurements (***) on both particles are correlated due to interactions between the particles in their mutual *causal* past through gauge fields.
(II) Quantum mechanically: nothing is real except the results of measurements. Here entanglement means the same except that it results from processes which are NOT restricted to the mutual causal past (such as the instantaneous collapse of the wavefunction).
(***) A measurement does NOT need to be a binar number due to interaction with a measurement apparatus, but could be the value of a function on phase space (of one particle) at a particular instant in time determined by the trajectory of the particle.
Since (II) is more liberal than (I), QM predicts correlations which cannot be reached by locally causal realist theories in THEORY (the practice is an entirely different matter).
Now, in scenario (I), suppose we want to check a property of a particle by making some measurement which takes about a nano second. Then, we know that if we control the environment in one meter around the experimental setup, we are safe since no events outside this region can influence the experimental outcome. Now, in scenario (II) this would be more troublesome since if our setup were entangled with an object outside this room and a measurement on this object were done between the preparation and registration of the relevant property of our particle, then the outcome of our experiment would have been influenced superluminally. If such superluminal influences were to occur randomly then it would be impossible to find any regularity in the outcome. This was an objection made by Einstein Podolsky and Rosen.
You can find good discussions about the difference between classical and quantum entanglement in the books of Bell - speakable and unspeakable in QM - and Selleri, the EPR paradox.

Of course, quantum physicists could argue that the influence of the environment on the subsystem would miraculously average out and therefore be of no importance. But that is exactly one aspect of the cat problem which they should SHOW and not assume. As you noticed, this issue gets different faces in different approaches (therefore I am not going to expand on this anymore). So, if you are interested in more measurement stuff, then specify your interpretation of QM.

Cheers,

careful

 

[DrChinese]

(b) Let me clarify some entanglement issues to you:

(I) CLASSICAL : all matter fields and gauge fiels are REAL. Entanglement between two particles would mean that results of separate, but identical measurements (***) on both particles are correlated due to interactions between the particles in their mutual *causal* past through gauge fields.

(***) A measurement does NOT need to be a binar number due to interaction with a measurement apparatus, but could be the value of a function on phase space (of one particle) at a particular instant in time determined by the trajectory of the particle.

You would think that the classical description would be reasonable, but of course there are all kinds of serious problems with it. Note the usage of the word "real" - which is at odds with the QM description. One of the interesting things about classical descriptions of the phenomena of entanglement is that this really isn't one - it is just nice words with no predictive force behind it.

For instance...if the quantum mechanical prediction $$P(A,B)=cos^2(\theta)$$ is wrong (which it must be, when you accept Bell's Theorem and still push local reality), then what is the "classical" prediction? This is where the local realist position always falls apart. There never is a specific prediction, because there are none that make any sense at all.

Careful, would you care to provide a specific formula for us to associate with the classical (local realist) position?

 

[Careful]

For instance...if the quantum mechanical prediction $$P(A,B)=cos^2(\theta)$$ is wrong (which it must be, when you accept Bell's Theorem and still push local reality), then what is the "classical" prediction? This is where the local realist position always falls apart. There never is a specific prediction, because there are none that make any sense at all.
Careful, would you care to provide a specific formula for us to associate with the classical (local realist) position?

Sigh, dear dr Chinese, I would kindly invite you to start reading the old paper of Pearle which I referred you to in the beginning of our amusing conversations and where it is PROVEN that you can reproduce EXACTLY the QM predictions, given that you allow for an angle dependent detector efficiency which has an upper limit of 87%. This is extremely high and can only be reached in experiments where you cannot exclude the locality loophole. I have told you before that there is at the moment NO UNIFIED local realist theory, this is what I repeated in another message which has been deleted from this forum. In the same message, I also said that QM is (partly for that reason) clearly stronger still on the microscale. Your intention is simply to close off any good news coming from the local realist side, and argue against anything which might have the potential to eventually clarify the mysticism in QM you like to embrace so much. If you want formula: you can enrich your mind in the many books written by A. O. Barut (the inventor of the self field approach), Franco Selleri and others. It would be nice to detect an effort from your side to TRY to understand my position, while I clearly express all the time the appropriate consideration for standard quantum mechanics. I do not react upon any message from your side anymore as long as such effort is not made clear. I am not here to explain BOOKS, since this is a never ending POINTLESS discussion and a waiste of my precious time. If you want to understand my position better, you should chat to Vanesch whom I have kindly explained this in some more detail.

Cheers,

Careful

 

[DrChinese]

Sigh, dear dr Chinese... I have told you before that there is at the moment NO UNIFIED local realist theory... I also said that QM is (partly for that reason) clearly stronger still on the microscale. Your intention is simply to close off any good news coming from the local realist side, and argue against anything which might have the potential to eventually clarify the mysticism in QM you like to embrace so much. If you want formula: you can enrich your mind in the many books written by A. O. Barut (the inventor of the self field approach), Franco Selleri and others.

You are at least honest enough to admit that there really is NO LOCAL REALIST THEORY. And THAT is the real reason there is no classical explanation for a purely quantum phenomena.

As to the good news from the LR front... you must be joking! It's all bad and getting worse. Every new experiment makes LR harder to hold onto for the dedicated LRist. The reason is that no LR theory can explain so many quantum phenomena simultaneously. A theory that explains entanglement can't reproduce something else; and vice versa. You must have noticed this already. And despite heroic attempts, the visibility argument has lost most of its steam since we now know that the results don't change one iota: As visibility increases from early Aspect experiments at about 5% all the way to 100% (Rowe - which you will of course tend to disregard), there is no deviation from the QM predictions. When is an experiment going to show results more along the lines of Bell's Inequality rather than QM? (Now, that would be good news for the LR front.)

BTW, I don't see where Barut ever gave a variant prediction on Bell test experiments in his self field theory.

 

[vanesch]

first statement:

You are at least honest enough to admit that there really is NO LOCAL REALIST THEORY. And THAT is the real reason there is no classical explanation for a purely quantum phenomena.

second statement:

The reason is that no LR theory can explain so many quantum phenomena simultaneously.

I think Careful's position is that although he grants the first statement (that is, in 2005, there is no local realist theory that can reproduce all QM results) he objects to the second statement, which has not yet any proof - only a very huge number of indirect indications.

So he thinks it is still worthwhile to pursue the QUEST for a *potential* LR theory. I tried to explain to him that this is a much better activity when you are retired and became rich in another way than to put your professional life at stake - because personally I wouldn't bet a dime on that possibility.

However, I think (I may be wrong) that *logically* the possibility is still open, so it is a matter of personal taste (and stubbornness :-) to *persue* that avenue.

The fundamental reason for him to do so is that he got personally convinced that QM as it stands is incompatible with the principle of general covariance. Others already came to such a conclusion - I am by far not competent enough to know whether this is only gut-feeling or whether it is almost for sure a no-go. Now, you can also argue from the relativist side, that GR has a lot of experimental support, so you could argue, in the same way as you can argue in favor of QM, that the principles of GR are untouchable ; in that case, you need to do something about QM. Clearly, QM has more illnesses (measurement problem, lots of maths difficulties in QFT...) than GR, which is, as a theory, rather crystal clear. And he then realized that there is still some hope to reconcile both, which would be the case if we could find a LR theory behind QM. That would of course mean that all EPR stuff goes out of the window (thanks to Bell) but he still sees this as experimentally not absolutely established.
Now, maybe there is something to the QM prediction of entanglement, but his hope is that if we first can establish a GR-compatible LR theory, that this may be another starting point from which to work towards eventual entanglement results by modifying both and that not all possibilities on that avenue have been explored. Talk about an ambitious research programme !

 

[Careful]

** You are at least honest enough to admit that there really is NO LOCAL REALIST THEORY. And THAT is the real reason there is no classical explanation for a purely quantum phenomena **

Sigh, I said there is no UNIFIED local realist theory (read well) for now yet, there are a few which capture particular features very well (now, you should be a good cook). Now, if there were such a unified theory, then LR would have quantum gravity ! (imagine yourself that ) So, you might imagine that, seen the unbalance in research efforts and so on, that such an accomplishment can still take a few decades...

** As to the good news from the LR front... you must be joking! It's all bad and getting worse. Every new experiment makes LR harder to hold onto for the dedicated LRist. The reason is that no LR theory can explain so many quantum phenomena simultaneously. A theory that explains entanglement can't reproduce something else; and vice versa. **

That is utterly false, we are still far away from any conclusive experiment and local realists actually hope there comes one since it might surpise YOU. Concerning your entanglement comments, you should see this on the appropriate scales. Moreover, you should try to answer my translation of the Einstein comment and not just focus on this issue, which is important for BOTH sides.

** You must have noticed this already. And despite heroic attempts, the visibility argument has lost most of its steam since we now know that the results don't change one iota: As visibility increases from early Aspect experiments at about 5% all the way to 100% (Rowe - which you will of course tend to disregard), there is no deviation from the QM predictions. **

You must understand that for local realists, the locality and detection loopholes are not separable for obvious physical reasons. This invalidates the Rowe experiment (conclusive experiments should be performed on scales of around one meter with very high efficiency). Until now, local realist MODELS actually fit experimental outcome certainly as good as QM does.

** BTW, I don't see where Barut ever gave a variant prediction on Bell test experiments in his self field theory **

Oh, but Barut did much more than just his self field theory (he was a true genius)! I shall give you one reference (which is on my to read staple, but I assume I do not have to justify recommending Barut) - you can check for more references therein : ``The fallacy of arguments against local realism in quantum phenomena´´ , in Waves and particles in light and matter, A. Van der Merwe et al, Plenum (1993)

Cheers,

Careful

 

[DrChinese]

I think Careful's position is that although he grants the first statement (that is, in 2005, there is no local realist theory that can reproduce all QM results) he objects to the second statement, which has not yet any proof - only a very huge number of indirect indications.

So he thinks it is still worthwhile to pursue the QUEST for a *potential* LR theory.

The fundamental reason for him to do so is that he got personally convinced that QM as it stands is incompatible with the principle of general covariance. Others already came to such a conclusion.

I certainly see the logic to placing GR above QM in the hierarchy (or vice versa). I think that new experiments to test the nuances of GR will being us some new data to help with this. Also, recent data shows that the universe is expanding at increasing rates (and much greater than c) which tells us that we still have a lot to learn about space-time. This may be accounted for in GR, but is it properly accounted for in QM?

After several decades of trying, quantum gravity has really gone nowhere - I don't think a single experimental test to date points in that direction. And really, the same is true of the local realist program. There has not been a single test pointing in that direction; nor even a single prediction yet emerge which is testable. It seems pretty clear to me that the LR idea of "visibility" as a saving grace for the program is doomed to failure. There has to be a point at which there is a connection between visibility and reported results, and so far there isn't. (I.e. as visibility increases, the QM prediction breaks down.)

But let's face it, no one knows where the next big breakthrough will come from. So maybe careful's direction will make sense in the end. Ultimately, the proof is in the pudding.

 

[DrChinese]

You must understand that for local realists, the locality and detection loopholes are not separable for obvious physical reasons. This invalidates the Rowe experiment (conclusive experiments should be performed on scales of around one meter with very high efficiency). Until now, local realist MODELS actually fit experimental outcome certainly as good as QM does.

I don't agree that Rowe is not meaningful. It certainly places severe restrictions on LR models. The idea for the LR program vis a vis Rowe must go something like this:

a. Rowe is efficient, but does not rule out local effects accounting for the correlations.

b. QM provides a good mechanism for the correlations regardless of local effects being ruled out or not.

c. The local realist must therefore add in a new, previously unknown mechanism to account for observed results.

So now we have traded Bell's Inequality for something just as bad. If there is such a new and previously unknown mechanism, what is it? Where is it? In other words, we are now requiring that the setting of one measurement device affects the results at the other measurement device. This must be explained by a candidate LR theory.

 

[Careful]

But let's face it, no one knows where the next big breakthrough will come from. So maybe careful's direction will make sense in the end. Ultimately, the proof is in the pudding.

As 't Hooft says it: let's everyone mud on, the solution will come from an unexpected direction The fact that the LR program has a clear guideline (quantum chemistry) is in my opinion a strength and not a weakness (it is very unlikely that in the next few decades we shall get any further indication for QG) - no other program has such clear goals.

For the record: I am not advocating necessarily pure LR, I think however it is the most sensible STARTING strategy for QG. Concerning this visibility, I have commented on that to you already in my email. I think it is not so unlikely as you seem to think it is, since the predictions of QM on scales below 1/10'th of a meter are **indisputable** also for me. It is entanglement above this scale which is very, very strange and cannot be explained by local realism (and would indeed need a nonlocal ``perturbation´´). Let's see what the future brings !

Cheers,

Careful

 

[Careful]

I don't agree that Rowe is not meaningful. It certainly places severe restrictions on LR models. The idea for the LR program vis a vis Rowe must go something like this:
a. Rowe is efficient, but does not rule out local effects accounting for the correlations.
b. QM provides a good mechanism for the correlations regardless of local effects being ruled out or not.
c. The local realist must therefore add in a new, previously unknown mechanism to account for observed results.
.

Agreed that Rowe adds restrictions (which could be seen as a good thing) - I did not say it was not meaningful (just that it is not an exclusive Bell test)- except for (c): local realism ALWAYS needs to take into account the detailed setup (that is its philosophy), it is not said that ``new physics´´ is needed.

**So now we have traded Bell's Inequality for something just as bad. If there is such a new and previously unknown mechanism, what is it? Where is it? In other words, we are now requiring that the setting of one measurement device affects the results at the other measurement device. This must be explained by a candidate LR theory **

Aha, that is were you have to dig deeper into the well known theories, and that is not obvious by any means. As I mentioned once, local realist explanations are going to be more difficult and that is why QM, when appropriatly apllied, is going to stay around a very long time as an effective theory and perhaps even as a fundamental one (in my worst nightmares :-) ). But the only thing we need to know for QG is that there exists a plausible, albeit complicated LR explanation (at least when you adapt the LR point of view).

Cheers,

careful

 

[DrChinese]

But the only thing we need to know for QG is that there exists a plausible, albeit complicated LR explanation (at least when you adapt the LR point of view).
Cheers,
careful

I want to make sure I am following this comment. To get Quantum Gravity to make sense and ultimately produce a good theory, one should adopt a local realist perspective as a starting point. Do I have the basics correct?

 

[Careful]

I want to make sure I am following this comment. To get Quantum Gravity to make sense and ultimately produce a good theory, one should adopt a local realist perspective as a starting point. Do I have the basics correct?

I did not say that one HAS to adopt a LR point of view to make QG make sensible. However, it is for sure an approach which avoids all philosophical as well as many (unsolvable) technical problems QM oriented approaches have. It is not a popular line of thought because it actually has to reproduce all QM successes on the scales I mentioned and is without the introduction of a non locality scale not capable of explaining quantum correlations at larger distances IF these exist in nature (but these correlations are also troublesome for QM itself - see my Einstein comment). However, you can compare the spirit of what I say *in some sense* with the study of interacting quantum field theories as a perturbation of the free field theory (with that difference that we do not know yet if nature is not ``free´´). Free field theories are the easiest ones to solve (and the only ones in which we have some reasonable particle notion), in the same way, realist locality is by far the most simplifying principle in nature and leads to the cleanest theories. So, it is ALWAYS useful to study these, even though they might not have the final word.

Cheers,

Careful

 

[Sherlock]

(a) Obviously, the problem concerns isolated substems of a closed system. The only closed system is the entire universe. Jim Hartle has written good stuff on that.
(b) Let me clarify some entanglement issues to you:
(I) CLASSICAL : all matter fields and gauge fiels are REAL. Entanglement between two particles would mean that results of separate, but identical measurements (***) on both particles are correlated due to interactions between the particles in their mutual *causal* past through gauge fields.
(II) Quantum mechanically: nothing is real except the results of measurements. Here entanglement means the same except that it results from processes which are NOT restricted to the mutual causal past (such as the instantaneous collapse of the wavefunction).
(***) A measurement does NOT need to be a binar number due to interaction with a measurement apparatus, but could be the value of a function on phase space (of one particle) at a particular instant in time determined by the trajectory of the particle.
Since (II) is more liberal than (I), QM predicts correlations which cannot be reached by locally causal realist theories in THEORY (the practice is an entirely different matter).
Now, in scenario (I), suppose we want to check a property of a particle by making some measurement which takes about a nano second. Then, we know that if we control the environment in one meter around the experimental setup, we are safe since no events outside this region can influence the experimental outcome. Now, in scenario (II) this would be more troublesome since if our setup were entangled with an object outside this room and a measurement on this object were done between the preparation and registration of the relevant property of our particle, then the outcome of our experiment would have been influenced superluminally. If such superluminal influences were to occur randomly then it would be impossible to find any regularity in the outcome. This was an objection made by Einstein Podolsky and Rosen.
You can find good discussions about the difference between classical and quantum entanglement in the books of Bell - speakable and unspeakable in QM - and Selleri, the EPR paradox.

Of course, quantum physicists could argue that the influence of the environment on the subsystem would miraculously average out and therefore be of no importance. But that is exactly one aspect of the cat problem which they should SHOW and not assume. As you noticed, this issue gets different faces in different approaches (therefore I am not going to expand on this anymore). So, if you are interested in more measurement stuff, then specify your interpretation of QM.

Thanks for the reply Careful. Sorry I intruded on your European Sunday morning. :-)

So far, my interpretation of qm would be the probabilistic one.

I gather from what you wrote above that quantum entanglement doesn't mean the same thing as classical entanglement in that classical entanglement requires some sort of prior physical connection between two particles, A and B, (not necessarily a direct connection between the two such as an interaction, but at least some common influence); and quantum entanglement doesn't require this. This doesn't seem correct to me. Wave function collapse by itself isn't what entangles two particles in qm, afaik. They have to be related in some way prior to the measurement which collapses the wave function in order for their wave functions to be linearly combined prior to the measurement, don't they?

I'm still not sure what you mean by saying that the assumption that isolated systems exist in nature isn't possible in qm.

In a Bell test setup for example, once a pair of disturbances have been produced and are on their way toward the polarizing filters, and before they interact with the polarizing filters, each disturbance is a closed, isolated system. (It's during this interval that their wave functions are combined and interfere via the application of the superposition principle, but this isn't a physical interaction.)

Every particle in the universe isn't quantum entangled with every other particle. They might be classically 'entangled' in some sense (eg. wrt the motion of the universe as a whole -- expansion, rotation, etc.; or wrt gravitational behavior of the macroscopic objects that they're part of), but these aren't quantum correlations.

I don't understand what you're saying about scenario (II) being more troublesome wrt the 1ns measurement example. It seems like you're assuming that quantum entanglement of spacelike separated events implies superluminal influences -- and, afaik, it doesn't. That is, for all anyone knows quantum entangled particles *are* affecting each other superluminally, but that conclusion isn't *required* from observations and qm (or Bell's theorem for that matter).

As for the S-cat problem, or why don't we see macroscopic interference? Well, we do see macroscopic interference, don't we? That's what ponderable objects are -- regions of interference/interaction wrt a hierarchy of various media, aren't they? After all, the idea of interfering *quantum* phenomena came from observing interfering macroscopic phenomena, didn't it?