Insights Why the Quantum | A Response to Wheeler's 1986 Paper - Comments

[Fra]

Between Bohr's (mis)understanding of quantum theory and today are 83 years with tremendous progress

As I see it the probabilisitic foundation required for QM is anchored in the classical "certainty".

The fact that one can in principle describe classical systems as emergent from a complex many-body QM picture, does not mean we do not need the classical measurement device.

Such a fallacious conclusions sits in the same category as those that suggest solving the observer problem by removing the observer, and instead attaching things in a metaphysical or mathematical realm and claim its objective.

This is a deep necessary insight that Bohr appears to have had. You can not make certain statistical predictions, without a certain distributions, and certain symmetries. These are manifested only on the classical side of things in the infinite ensembles etc; or in the "observer" part of this, if we are to generalize beyond classical observers.

This is easy to see if you analyse this from the point of view of inference. It should also be intuitive for any experimental work as the accuracy and confidence in the statistical predictions, requires a solid control and knowhow of the classical measurement devices. But from the perspective of mathematical physics, the statistical predictions of QM is anchored in axioms, that sit in the mathematical realm and its very easy to be seduced and confused by this.
And that essense is what i read out of Bohrs original view as well is that he understood this, this is why a proper formulation of quantum theory itself REQUIRES the classical reference. I think this is a fundamental insight.

We certainly need to improve this to understand QG and unification, but can't see anyone so far has done better than Bohr. We obviously grossly improved and developed the SM for particle physics and QFT, but the foundations remain at Bohr level.

/Fredrik

 

[vanhees71]

To be more precise, we need something that behaves with good enough accuracy classically, and quantum many-body theory shows that many-body quantum systems are behaving to good accuracy classically. That's all you need to explain why quantum theory is successful in providing its probabilistic description of the outcome of measurements on quantum systems with macroscopic measurement apparati. There's nothing, however, hinting at a "quantum classical cut", i.e., there's nothing contradicting QT in favor of a classical description, but for many-body systems very often the classical description is a very accurate description for macroscopic "coarse-grained quantities", which are sufficiently accurate to describe the relevant behavior of many-body systems, including measurement apparati. Particularly there's no difference between measurement devices and any other kind of matter since indeed measurement devices are composed of the same elementary particles as anything around us.

 

[stevendaryl]

There's nothing, however, hinting at a "quantum classical cut", i.e., there's nothing contradicting QT in favor of a classical description

But the formalism doesn't actually make any predictions without such a cut. Without a distinction between measurements and other interactions, or between macroscopic and microscopic, there are no probabilities in QM, and the theories only predictions are probabilistic.

 

[vanhees71]

In QM of course everything is probabilistic from the very beginning, but there is no cut anywhere in the formalism. Where do you need that cut?

 

[stevendaryl]

In QM of course everything is probabilistic from the very beginning

The evolution of the wave function is deterministic. Probabilities come in when you make a division between a macroscopic system (the measuring device) and the system being measured. That division is necessary for there to be any probabilities at all.

 

[vanhees71]

The "wave function" is a probability amplitude by definition (within the standard minimal interpretation). Thus it's probabilistic from the very beginning, without any necessity to introduce classical concepts.

 

[stevendaryl]

The "wave function" is a probability amplitude by definition (within the standard minimal interpretation). Thus it's probabilistic from the very beginning, without any necessity to introduce classical concepts.

That's not true. The wave function gives probabilities for measurement results. Without distinguishing measurement results from other properties, there are no probabilities in QM.

To have probabilities you have to have events---the things that have associated probabilities. The events for QM are measurement results.

 

[lavinia]

That's not true. The wave function gives probabilities for measurement results. Without distinguishing measurement results from other properties, there are no probabilities in QM.

To have probabilities you have to have events---the things that have associated probabilities. The events for QM are measurement results.

Why aren't there situations where quantum states are naturally falling into eigenstates of some operator - without measurement - for instance on a star?
This could happen trillions of times and thus a probability distribution. Or is any time a quantum states projects onto an eigen state of an operator a measurement by definition?

 

[stevendaryl]

Why aren't there situations where quantum states are naturally falling into eigenstates of some operator - without measurement - for instance on a star?
This could happen trillions of times and thus a probability distribution. Or is any time a quantum states projects onto an eigen state of an operator a measurement by definition?

I wasn't giving my opinion about it---I was describing the orthodox interpretation of quantum mechanics, which is that the probabilities in quantum mechanics are probabilities of measurement results.

An alternative interpretation which I think is empirically equivalent is to forget about measurements, and instead think of QM as a stochastic theory for macroscopic configurations. What I think is nice about this approach is that it doesn't single out measurements, and it doesn't require the assumption that a measurement always gives an eigenvalue of the operator corresponding to the observable being measured. It doesn't require observers, so you can apply QM to situations like distant stars where there are no observations. On the other hand, it's got the same flaw as the orthodox interpretation, in that it requires a macroscopic/microscopic distinction.

Getting back to your specific comment, I'm not sure what you mean by "naturally falling into eigenstates". Could you elaborate?

 

[vanhees71]

That's not true. The wave function gives probabilities for measurement results. Without distinguishing measurement results from other properties, there are no probabilities in QM.

To have probabilities you have to have events---the things that have associated probabilities. The events for QM are measurement results.

What else than measurement results should any physical theory describe? Physics is about objectively observables facts of nature. It's not an empty mathematical game of thought, where you solve Schrödinger's equation just for fun without needing any "meaning" of the wave function, i.e., just because for some reason you like the puzzle to solve the equation.

 

[stevendaryl]

What else than measurement results should any physical theory describe? Physics is about objectively observables facts of nature. It's not an empty mathematical game of thought, where you solve Schrödinger's equation just for fun without needing any "meaning" of the wave function, i.e., just because for some reason you like the puzzle to solve the equation.

It's sort of funny that you simultaneously denigrate philosophy and take such strong philosophical positions.

But what you said doesn't change the fact that QM in the minimalist interpretation must make a distinction between measurements and other interactions. I'm just pointing out that you previously claimed that no such split is necessary.

 

[stevendaryl]

What else than measurement results should any physical theory describe? Physics is about objectively observables facts of nature. It's not an empty mathematical game of thought, where you solve Schrödinger's equation just for fun without needing any "meaning" of the wave function, i.e., just because for some reason you like the puzzle to solve the equation.

A theory of physics does not have to be based on measurements in order to have observational content. What you need for empirical content to a theory are correspondences: Such and such phenomenon described in the theory is assumed to correspond to such and such observation. You need for the theory to show how observations are affected by the objects and fields and so forth in the theory.

If human beings and measurement devices are physical objects described by the theory, then you should be able, in principle, to predict what happens to humans or measurement devices in this or that circumstance. That gives empirical content to the theory.

In every other theory besides quantum mechanics--special relativity, general relativity, electromagnetism, Newtonian mechanics, etc.--what is described is the behavior of particles and fields. That is enough to have empirical content if we (and our measuring devices) are ourselves made up out of those particles and fields.

 

[vanhees71]

It's sort of funny that you simultaneously denigrate philosophy and take such strong philosophical positions.

But what you said doesn't change the fact that QM in the minimalist interpretation must make a distinction between measurements and other interactions. I'm just pointing out that you previously claimed that no such split is necessary.

It's not philosophy, it's physics. I just take what my experimental colleagues do in the lab and try to make sense of quantum mechanics. The main difficulty in understanding quantum mechanics is that it is formulated by people who are too philosophical (Bohr, Heisenberg), and that it is very hard to get rid of their "doctrine" (as Einstein rightfully called it).

There is no distinction between measurements and other interactions. The interaction of a particle, say a pion, with a silicon chip within a detector at the LHC is just according to the interactions described by the Standard Model (usually it's of course the electromagnetic interaction for detecting particles or photons). There's not the slightest hint that there are different laws for the interaction of a pion with some semiconductor if it's used to detect the particle or with the same piece of matter if it's not used to detect the particle.

Again, you always claim that you need a split, but you never tell why you think so. Mostly this misconception comes about, because it's somehow diffused into the teaching of QT through taking Bohr et al as the authorities having the final word on the interpretation of QT, but that's not an argument at all. There is no evidence for such a "cut" by any modern experiment, as far as I know, or do you know any experimental evidence, published in a serious peer-reviewed journal, which claims to prove that there's distinction between interactions of particles with matter (i.e., many-body quantum systems) depending on whether this matter is used as a detector or whether it's not used as such? I'd be very surprised, to say the least ;-).

 

[stevendaryl]

It's not philosophy, it's physics.

No, it's philosophy.

There is no distinction between measurements and other interactions

That might be your belief, but it isn't consistent with the axioms of quantum mechanics in the minimalist interpretation.

 

[stevendaryl]

Again, you always claim that you need a split, but you never tell why you think so.

I'm saying that the minimal interpretation already has that split. Try formulating the probabilistic predictions of the minimalist interpretation without mentioning "measurement".

 

[Boing3000]

It's not philosophy, it's physics.

No, it is philosophy. It is stunning to hear a experimentalist pretend that his lab is made of quantum object and quantum observation. Every single one of your observation is classic, in the only un-philosophically possible sense.

I just take what my experimental colleagues do in the lab and try to make sense of quantum mechanics.

By counting classical "up" "down", not by observing some weird superposition. And you fail to recognize you have a cut of how many of those "identically prepared state" you'll have to classically observe before being content with the stochastic prediction.

 

[stevendaryl]

I'm saying that the minimal interpretation already has that split. Try formulating the probabilistic predictions of the minimalist interpretation without mentioning "measurement".

If you want to treat a measurement as just another interaction, then you should be able to formulate the probabilistic predictions of quantum mechanics without mentioning the word "measurement".

One attempt might be the following: We say that system $A$ (the measuring device) measures a property of a second system, $B$ if the interaction between the two systems causes an irreversible change in the state of system $A$ such that distinct values of the property of system $B$ reliably lead to macroscopically distinguishable states of system $A$. This definition of "measurement" seems to necessarily involve distinguish macroscopic properties from microscopic properties.

Of course, there are alternative interpretations, but the minimal interpretation seems to me to absolutely require such a distinction. You cannot make sense of the minimalist interpretation without this distinction (or something equivalent: macroscopic versus microscopic, irreversible versus reversible, measurement versus non-measurement).

I don't have a proof that it is impossible to make sense of Born probabilities without making such a distinction, I'm just claiming that the minimalist interpretation does not do so.

 

[vanhees71]

I'm saying that the minimal interpretation already has that split. Try formulating the probabilistic predictions of the minimalist interpretation without mentioning "measurement".

Sigh. It is really difficult to make this simple argument. Of course, I have to mention measurments. I have to state it, because physics is about measurements. What else should it be about? I never have to use the word "classical" in all these definitions. That's the point, not to avoid the word "measurement" or "observation". Again, where is, in your opinion, the necessity to invoke classical arguments here? You havent's defined, what you mean by "classically observe".

Let's take a photon. It's observed by letting it interact with a detector (in former days a photo plate, nowadays some electronic detector like a CCD). There's not the slightest hint that the interaction of the photon with the photo plate or CCD cam is any different from the electromagnetic interactions described by QED.

 

[vanhees71]

One attempt might be the following: We say that system $A$ (the measuring device) measures a property of a second system, $B$ if the interaction between the two systems causes an irreversible change in the state of system $A$ such that distinct values of the property of system $B$ reliably lead to macroscopically distinguishable states of system $A$. This definition of "measurement" seems to necessarily involve distinguish macroscopic properties from microscopic properties.

Sure, but the classical describability of macroscopic properties is not due to some cut, beyond which quantum theory isn't valid anymore, but it's explanable by coarse graining from quantum many-body systems.

 

[stevendaryl]

Sigh. It is really difficult to make this simple argument. Of course, I have to mention measurments. I have to state it, because physics is about measurements. What else should it be about?

That is not the truth. Newtonian physics is not formulated in terms of measurements. Neither is any other theory of physics besides the minimal interpretation. What you're saying is just not true. You're interpreting things through your personal philosophy.

What all theories of physics must have (if they are supposed to be fundamental) is a correspondence between observations and phenomena described in the theory. If you have a theory of light, then for it to have observational content, you need something along the lines of the assumption that seeing involves light entering our eyes and registering with sensors there. But the theory of light is not expressed in terms of observations. Maxwell's equations do not mention observations. Newton's laws don't mention observations. General Relativity doesn't mention observations. You don't need for a theory to be about measurements in order to have empirical content, you need to be able to describe how the phenomena described by the theory affects what is observable.

That's the point, not to avoid the word "measurement" or "observation". Again, where is, in your opinion, the necessity to invoke classical arguments here? You havent's defined, what you mean by "classically observe".

I didn't mention the word "classical" either. I said that the probabilistic predictions of QM (at least in the minimal interpretation---things are different in the Bohmian interpretation and the consistent histories interpretation and the many-worlds interpretation) depend on a distinction between "measurement" and other interactions.

 

[stevendaryl]

Sure, but the classical describability of macroscopic properties is not due to some cut, beyond which quantum theory isn't valid anymore, but it's explanable by coarse graining from quantum many-body systems.

No, coarse-graining doesn't explain anything. It's another way of formulating the split.

 

[vanhees71]

Of course, Newtonian physics is about measurements. To write down a position vector you already need to define it in terms of measurable quantities, e.g., the three Cartesian coordinates with respect to an appropriate reference frame (provided, e.g., by three rigid rods of unit length put together at a point or the edges in one corner of your lab, etc.). Physics is about measurable quantities.

Again you only stated that the minimal interpretation depends on a distinction between measurement and other interactions, but you did not tell WHAT difference this might be and why this distinction is even NECESSARY.

 

[Mentz114]

. That is not the truth. Newtonian physics is not formulated in terms of measurements. Neither is any other theory of physics besides the minimal interpretation..

All theories are written to express the outcomes of measurements ( or observations). It is not stated explicitly because it is obvious. J J Gleason identifies any formula that gives the value of a classical outcome as an operator, in analogy with QT. The insistence that 'measurement' is somehow different from other interactions is not justified.

 

[vanhees71]

No, coarse-graining doesn't explain anything. It's another way of formulating the split.

Ok, if you think so, I've to accept it, but then how can you explain the classical behavior of macroscopic objects from quantum theory at all, or are you really thinking, there's a cut on a fundamental level? If so, where's the empirical evidence for it?

 

[stevendaryl]

Of course, Newtonian physics is about measurements.

No, it is not. Certainly not in the sense that QM is about measurements. Newtonian physics is about the motion of particles under the influence of forces. The connection with measurement requires an assumption that the forces and/or particle motions have an affect on the measuring device. So what Newtonian physics says about measurement is derivable from Newtonian physics (possibly with other assumptions). It is not cooked into Newtonian physics.

If you assume that a spring deforms in a linear way when a force is applied to one end, then the spring can be used for measurement of forces. But it would be a mistake to define force in terms of the deformation of springs.

 

[stevendaryl]

Ok, if you think so, I've to accept it, but then how can you explain the classical behavior of macroscopic objects from quantum theory at all, or are you really thinking, there's a cut on a fundamental level? If so, where's the empirical evidence for it?

I'm saying that the minimalist interpretation of quantum mechanics makes a distinction between measurement interactions and other interactions. I'm not saying that it is impossible to come up with an interpretation of quantum mechanics that doesn't rely on such a split, only that your preferred interpretation requires it.

Let's suppose that we have a device that measures the spin of an electron along the z-axis as follows:

• If the electron is spin-up, a pointer on the device will point to the left.
• If the electron is spin-down, a pointer on the device will point to the right.

If you treat the pointer like a quantum-mechanical object, then you would have to conclude:

• If the electron is in a superposition of spin-up and spin-down, then the pointer will later be in a superposition of pointing left and pointing right. (Or more accurately, the entire universe will be in a superposition of a state in which the pointer points to the left and one in which the pointer points to the right).

But the Born rule says something different:

• If the electron is in a superposition of spin-up and spin-down, then the pointer will later either point left, with such-and-such probability, or point right, with such-and-such probability.

That rule is unlike anything you would say about microscopic systems.

 

[lavinia]

I wasn't giving my opinion about it---I was describing the orthodox interpretation of quantum mechanics, which is that the probabilities in quantum mechanics are probabilities of measurement results.

An alternative interpretation which I think is empirically equivalent is to forget about measurements, and instead think of QM as a stochastic theory for macroscopic configurations. What I think is nice about this approach is that it doesn't single out measurements, and it doesn't require the assumption that a measurement always gives an eigenvalue of the operator corresponding to the observable being measured. It doesn't require observers, so you can apply QM to situations like distant stars where there are no observations. On the other hand, it's got the same flaw as the orthodox interpretation, in that it requires a macroscopic/microscopic distinction.

Getting back to your specific comment, I'm not sure what you mean by "naturally falling into eigenstates". Could you elaborate?

by naturally I just meant without measurement.

 

[Mentz114]

No, coarse-graining doesn't explain anything. It's another way of formulating the split.

Can you expand that ? It might help to understand what the 'split' actually is.

 

[stevendaryl]

by naturally I just meant without measurement.

But under what circumstances would a star or whatever naturally make a transition into an eigenstate of some operator?

 

[stevendaryl]

Can you expand that ? It might help to understand what the 'split' actually is.

I sketched this in another post a while back.

But let's suppose that coarse-graining can be mathematically defined in terms of projection operators. Let $|\psi\rangle$ be the state of the complete system (environment plus measuring devices plus observers plus ...). Then we want a set of projection operators $\Pi_j$ such that:

• If the system is in a definite coarse-grained state $j$, then $\Pi_j |\psi\rangle = |\psi\rangle$.
• If the system is in a definite coarse-grained state $k$ different from $j$, then $\Pi_j |\psi\rangle = 0$.

Then the Born rule can be formulated as: The probability of the system being in coarse-grained state $j$ is given by:

$P(j) = \langle \psi|\Pi_j|\psi \rangle$

So the Born rule applies to coarse-grained projection operators.

The usual Born rule can be derived from this one. The usual formulation says that if you measure a property of a subsystem, then you will get an eigenvalue, with probabilities given by the square of the amplitude corresponding to the decomposition of the subsystem state into eigenstates. But if you interpret "measurement" as meaning: "A process whereby the value of the microscopic quantity is amplified to make a macroscopic difference", then different values of the microscopic property will lead to different coarse-grained states of the measurement device.So the Born rule on coarse-grained states implies that you will get results with the right probabilities.

But note: To have agreement with observation, you only need the Born rule to apply to coarse-grained projections, not to arbitrary (microscopic) projections. And furthermore, I don't know of a way to consistently extend the Born rule in terms of projections to microscopic properties. I don't think there is any way.

So the Born rule in my understanding requires a distinction between macroscopic coarse-grained descriptions (where the rule applies) and microscopic descriptions (where it does not).