Is GRW theory an interpretation of quantum mechanics or a rival theory?

[Jarvis323]

Sounds like you are talking QT. If a quantity is uniquely determined by another, this question isn't relevant. Correlation on the other hand always implies a certain type of dependence. If you assume something is sepererable, then you have to specify what that means and Bell's factorization is one type of separation. And it yields his inequality. Strongest separation is that of independence which implies null correlation.

We can have a reduced description of something based on correlation in the information we obtain about it and calculations exploiting those correlations. It doesn't mean that the reduced description represents less real stuff. The theory is about prediction. It is only necessary that we have sufficient information to make predictions. We even seek simplified models as preferable. It doesn't mean nature is simple and reduces in reality to the shortest description as we like our models to.

 

[Killtech]

We can have a reduced description of something based on correlation in the information we obtain about it and calculations exploiting those correlations. It doesn't mean that the reduced description represents less real stuff. The theory is about prediction. It is only necessary that we have sufficient information to make predictions. We even seek simplified models as preferable. It doesn't mean nature is simple and reduces in reality to the shortest description as we like our models to.

entirely different motivation. Don't think it makes sense to talk about ontology of an approximation. Approximations may work with various things like averages over many original values and you would be forced to declare such as ontic since the originals aren't available but then you can more or less declare anything as ontic since you could derive very different effective quantities and use those instead. But when you could declare almost any quantity as ontic, it makes the word loses all meaning.

Alternatively you could declare most quantities as non-ontic making some of the information a theory uses having no ontic source inside the theory.

 

[Morbert]

Continuing the ontology discussion from the many-worlds thread here as it's probably more approrpriate

In classical physics, the finest possible graining is unique. CH, on the other hand, says that the finest possible graining is not unique. A Bohmian would say that the finest possible graining corresponds to primitive ontology, so in this language one would say that in CH primitive ontology is not unique. A classical analogy would be using the fact that a classical field $\phi(x)$ can equivalently be represented by its Fourier transform $\tilde{\phi}(k)$ and saying that $\tilde{\phi}(k)$ is not any less real than $\phi(x)$. The claim that, in classical physics, $\tilde{\phi}(k)$ is as real as $\phi(x)$ is very seducing mathematically, but at the same time it's very hard to swallow it from an intuitive physical point of view. CH interpretation of QM is hard to swallow for exactly the same reason.

"Finest possible graining" might have two interpretations. If, by a finest possible graining, we mean for example alternative field configurations $\phi(x)$, CH would fail to ascribe any ontological meaning to them, primitive or otherwise, since these histories (four-dimensional configurations) would fail to decohere for general circumstances. We would have to introduce a coarse-graining that satisfies a decoherence criterion. E.g. Instead of alternative field configurations, we would take alternative averages of field configurations over causally consistent spacetime regions. Once we have these as our finest possible grainings, CH can offer an ontic meaning to them, in the sense that we can interpret measurement outcomes as revealing pre-rexisting properties that these averages represent. The necessity of coarse-graining means we would struggle to say these properties are primitive. It would therefore be hard to interpret CH as offering multiple representations of a primitive ontology. The ontic contribution of CH only extends to the interpretation of measurements as revealing pre-existing properties, regardless of the status of those properties as primitive. Using the example from the other thread: CH won't tell you if your airplane is a primitive element of reality, it will just tell you if it will crash.

There is an adjascent interpretation "Extended Probability Ensemble Decoherent Histories" which does try to ascribe an ontic meaning to maximally fine-grained histories. This interpretation selects a preferred set of histories, and if we are talking about a field theory, they would indeed be field configurations $\phi(x)$.

 

[physika]

the particle position and its mass we have all information we need so we can derive the force from there. Thus the force field itself merely holds redundant information.

Without field there is no particle.

 

[Demystifier]

You would still have to claim that "kinematics" only includes position, which is debatable. And then you would have to claim that, while things that change are "ontic" (kinematics), the things that cause them to change are not (dynamics). Which, again, does not seem to be a "basic intuition" that an ordinary person in the street would have.

I had a different idea in mind. I was suggesting to completely remove "ontic" from the vocabulary and talk about "kinematics" instead. Here is an example:
- What's wrong with Copenhagen interpretation?
- The problem is that it doesn't clearly specify what is its kinematics.
- But it does, it's wave function $\psi(x,t)$.
- But $\psi(x,t)$ is similar to $S(x,t)$ in Hamilton-Jacobi formulation of classical mechanics. And we know that $S(x,t)$ is not kinematics, but one of ways to formulate the dynamics. By analogy, it's possible that $\psi(x,t)$ in QM is also one of ways to formulate dynamics. If so, then it's not clear what in Copenhagen QM is kinematics.
- I think you are just trying to justify Bohmian mechanics, according to which quantum kinematics is $x(t)$.
- Not necessarily, for example MWI and GRW say clearly what the kinematics is, but they don't say that it's $x(t)$. Copenhagen, on the other hand, does not say clearly what's kinematics.

 

[PeterDonis]

I was suggesting to completely remove "ontic" from the vocabulary and talk about "kinematics" instead.

How does that help? Why should I just care about "kinematics" and not about "dynamics"? Or why should I care about making a precise distinction between them?

 

[Demystifier]

$E$ is force on unit charge. So forces are ontic after all.

Unless you believe that $E$ exists even when it does not act on a charge, for example during a propagation in EM wave.

 

[Demystifier]

Continuing the ontology discussion from the many-worlds thread here as it's probably more approrpriate
"Finest possible graining" might have two interpretations. If, by a finest possible graining, we mean for example alternative field configurations $\phi(x)$, CH would fail to ascribe any ontological meaning to them, primitive or otherwise, since these histories (four-dimensional configurations) would fail to decohere for general circumstances. We would have to introduce a coarse-graining that satisfies a decoherence criterion. E.g. Instead of alternative field configurations, we would take alternative averages of field configurations over causally consistent spacetime regions. Once we have these as our finest possible grainings, CH can offer an ontic meaning to them, in the sense that we can interpret measurement outcomes as revealing pre-rexisting properties that these averages represent. The necessity of coarse-graining means we would struggle to say these properties are primitive. It would therefore be hard to interpret CH as offering multiple representations of a primitive ontology. The ontic contribution of CH only extends to the interpretation of measurements as revealing pre-existing properties, regardless of the status of those properties as primitive. Using the example from the other thread: CH won't tell you if your airplane is a primitive element of reality, it will just tell you if it will crash.

There is an adjascent interpretation "Extended Probability Ensemble Decoherent Histories" which does try to ascribe an ontic meaning to maximally fine-grained histories. This interpretation selects a preferred set of histories, and if we are talking about a field theory, they would indeed be field configurations $\phi(x)$.

It seems to me that here you identify decoherent histories with consistent histories. My understanding is that the former is just one version of the latter.

 

[Demystifier]

Or why should I care about making a precise distinction between them?

I don't know why should you care, but in the context of classical mechanics many physicists care. In elementary school (7th grade) that was one of the first things we were taught in physics class.

Or for Copenhagenians, it can be a way to explain the difference between classical and quantum mechanics: classical physics treats kinematics and dynamics separately, while QM doesn't. From that point of view, BM, MWI and GRW are attempts to restore their separation in QM too.

Isn't it at least a bit less vague than a talk about ontic?

 

[PeterDonis]

In elementary school (7th grade) that was one of the first things we were taught in physics class.

I guess it depends on the school system and the curriculum. I took physics in high school and again in several classes in college and nobody ever mentioned kinematics vs. dynamics. I don't feel like I missed anything.

 

[Demystifier]

I guess it depends on the school system and the curriculum. I took physics in high school and again in several classes in college and nobody ever mentioned kinematics vs. dynamics. I don't feel like I missed anything.

I'm not saying that physics cannot be taught without that distinction. I'm just saying that kinematics vs. dynamics is less vague and less philosophic than ontic vs. non-ontic. So if we want to discuss quantum foundations at all (and it seems that we do), it makes a step forward by removing ontic and non-ontic from the vocabulary and replacing them with kinematics and dynamics. In fact, I plan to do exactly that in the future discussions. We shall see how that will work.

 

[martinbn]

$E$ being ontic is less problematic than force being ontic, because $E$ exists in space and time. The force resulting from Newton's law of gravity on the other hand does not have an independent existence in space and time, but can only exist where the particle position already exists. Therefore the particle position is the primary ontology in that case, and force is only one of many possible secondary variables.

I don't see the difference. You place a test object and it experiences a force. Same in both cases.

This existence in space and time is precisely one of the reasons why ontology with respect to "physical existence" has a slightly different character from ontology with respect to "mathematical existence". The "intuitive notion" of something which exists physically normally implies that it exists at some time and at some point, or at least in some region of space and time. This is one reason why the wavefunction should better not be ontic.

I agree with this. This is why for me it is strange to shift the ontology from the particles to the mathematical object $x(t)$.

 

[martinbn]

I don't know why should you care, but in the context of classical mechanics many physicists care. In elementary school (7th grade) that was one of the first things we were taught in physics class.

Or for Copenhagenians, it can be a way to explain the difference between classical and quantum mechanics: classical physics treats kinematics and dynamics separately, while QM doesn't. From that point of view, BM, MWI and GRW are attempts to restore their separation in QM too.

Isn't it at least a bit less vague than a talk about ontic?

For me it is less vague. I went to a similar school system. In fact it is obvious that position is a kinematics notion while momentum is a dynamics one.

Where i went to school/uni mechanics is always divided into statics, kinematics, dynamics.

 

[vanhees71]

Sorry,
Treated, but not ontic...

Ontic (from the Greek ὄν, genitive ὄντος: "of that which is") is physical, real, or factual existence.
Maybe both, particles as excitations of the fields.

.

So if this is the definition of "ontic", which I thought is indeed the meaning before reading this thread, after which it got a word without any meaning, then it's simply what a physicist would call an objectively observable phenomenon. Then it doesn't make sense to say the one or the other element of a theory were "ontic" or "not ontic", because then all theory is not ontic, because it's just a quantitative mathematical description of the observable phenomena.

 

[vanhees71]

$E$ being ontic is less problematic than force being ontic, because $E$ exists in space and time. The force resulting from Newton's law of gravity on the other hand does not have an independent existence in space and time, but can only exist where the particle position already exists. Therefore the particle position is the primary ontology in that case, and force is only one of many possible secondary variables.

This existence in space and time is precisely one of the reasons why ontology with respect to "physical existence" has a slightly different character from ontology with respect to "mathematical existence". The "intuitive notion" of something which exists physically normally implies that it exists at some time and at some point, or at least in some region of space and time. This is one reason why the wavefunction should better not be ontic.

I'd say the electromagnetic field is indeed not "ontic", because it is just a description of phenomena occurring, according to our description in terms of Maxwell's theory, when this field interacts with some charged matter. One ontic element, i.e., one observable and even directly perceivable by the human senses phenomenon is what we call "light", though here we also have to distinguish between the physical and the physiological meaning. Physically what's observable is an electric signal which we describe to be due to the photoelectric effect, i.e., the interaction of the em. wave field with the charged matter particles making up our retina, which already starts to process the signal, which then is propgated to our brain where it's further processed and so on until we preceive the phenomenon "light".

 

[physika]

So if this is the definition of "ontic", which I thought is indeed the meaning before reading this thread, after which it got a word without any meaning, then it's simply what a physicist would call an objectively observable phenomenon. Then it doesn't make sense to say the one or the other element of a theory were "ontic" or "not ontic", because then all theory is not ontic, because it's just a quantitative mathematical description of the observable phenomena.

Maybe wish to be.

 

[gentzen]

So if this is the definition of "ontic", which I thought is indeed the meaning before reading this thread, after which it got a word without any meaning, then it's simply what a physicist would call an objectively observable phenomenon. Then it doesn't make sense to say the one or the other element of a theory were "ontic" or "not ontic", because then all theory is not ontic, because it's just a quantitative mathematical description of the observable phenomena.

Demystifier's basic premise of this thread was that the meaning of the word "ontic" can be learned by observing how it is actually used. However, instead of actually doing this, he tried to shortcut that process by presenting some borderline examples. Worse, his refernce (Dürr) used the word "primitive ontology" (instead of "ontic") for the concept exemplified by those borderline cases.

A book by Detlef Dürr is probably not good enough to make you happy. So let me instead play this game with the book "Do we really understand quantum mechanics" by Franck Laloë (because I expect that it is easier for you to respect that author). The words "ontic" and "ontology" don't appear often in that book, and basically only in the places where they could not be avoided:

Many authors introduce the words “ontic” and “epistemic” in this context, even if they are more common in philosophy than in physics. An analogy with classical physics is useful to understand their meaning: in classical physics, the position and momenta of all particles of a physical system provide a direct description of its physical properties; this is an “ontic” description. But if the system is described only in a probabilistic way by a statistical distribution in phase space (as in the Liouville theorem) reflecting a partial knowledge of its state, this description is called “epistemic”.

Tumulka has proposed a relativistic version of the GRW theory [704], which he calls “flash ontology”, where the “local beables”(§11.8) are given by a discrete set of space-time points, at which the collapses are centered; these points have a random distribution that is determined by the initial wave function.

Does this help to reduce your impression that it is a "word without any meaning"?

 

[vanhees71]

I've nothing against Dürr. To the contrary his approach to Bohmian mechanics in his book "Understanding quantum mechanics" makes at least sense for non-relativistic single-particle QT, though I'm not a follower of Bohmian mechanics, as long as there's no convincing consistent version for relativistic QFT (and a formulation, where the basic causality principle of relativity is not fulfilled, is not at all convincing form me!).

The above quote also doesn't make sense to me, because for me the probabilistic behavior of quantum systems are also "direct descriptions of its behavior". It's precisely what's observed: E.g., if I shoot an electron with pretty well defined momentum through a double slit, I don't know very precisely, where it will be registered behind the slits, and that's what's observed. The electron will be registered somewhere on the slits. QT predicts a probability distribution which I can calculate as an interference pattern of a Schrödinger wave function, and then repeating the experiment with equally prepared electrons a lot of time, I indeed get the corresponding distribution of registered electrons at my slit. The randomness for me is an objective phenomenon as is the position of a classical particle in phase space within classical mechanics. For sure, a single electron is an example, where the classical "phase space ontology" fails to describe the phenomena, while the "probabilistic ontology" of quantum theory describes them very well.

 

[Morbert]

It seems to me that here you identify decoherent histories with consistent histories. My understanding is that the former is just one version of the latter.

Consistent histories and decoherent histories are ultimately the same formalism. As research projects, the former has concerned itself with minimal decoherence criteria, while the latter focuses on cases with strong decoherence. https://www.webofstories.com/play/murray.gell-mann/161

Ontologically speaking, Griffith's CH interpretation is closer to EPE-DH than to Omnes's CH. And you can see him equate CH with DH here.

 

[martinbn]

Demystifier's basic premise of this thread was that the meaning of the word "ontic" can be learned by observing how it is actually used. However, instead of actually doing this, he tried to shortcut that process by presenting some borderline examples. Worse, his refernce (Dürr) used the word "primitive ontology" (instead of "ontic") for the concept exemplified by those borderline cases.

A book by Detlef Dürr is probably not good enough to make you happy. So let me instead play this game with the book "Do we really understand quantum mechanics" by Franck Laloë (because I expect that it is easier for you to respect that author). The words "ontic" and "ontology" don't appear often in that book, and basically only in the places where they could not be avoided:Does this help to reduce your impression that it is a "word without any meaning"?

The quote includes momentum along with position as ontic.

 

[Killtech]

Without field there is no particle.

Newton's gravity is a special case, given how it is mathematically setup. You could just skip the force entirely and just describe how each particles accelerates another since gravity there works via an action at range. The old classical model could be understood that particles interact directly with each other at range with no intermediary field. This is why in that old model your statement doesn't apply.

 

[physika]

just describe how each particles

Right !

As your said:
"Just describe"
...no more than that.

 

[Killtech]

Right !

As your said:
"Just describe"
...no more than that.

$$\ddot x_1 = G \frac {m_2} {(x_1-x_2)^2}, \ddot x_2 = G \frac {m_1} {(x_1-x_2)^2}$$

 

[gentzen]

I've nothing against Dürr. To the contrary his approach to Bohmian mechanics ...

The above quote also doesn't make sense to me, because for me the probabilistic behavior of quantum systems are also "direct descriptions of its behavior".

If you have already trouble making sense of Franck Laloë's words, then my expectation would be that you should have even more trouble making sense of Detlef Dürr's words.

I'd say the electromagnetic field is indeed not "ontic", because it is just a description of phenomena occurring, according to our description in terms of Maxwell's theory, when this field interacts with some charged matter. One ontic element, i.e., one observable and even directly perceivable by the human senses phenomenon is what we call "light", though ...

I wrote that elaboration on "... the meaning of the word ontic can be learned by observing how it is actually used ..." to have some basis for answering your comment. But apparently that didn't work as planned. So instead I will answer from my own perspective (which is unrelated to this thread and Demystifier's goals):
Concepts like "ontology" and "ontic" depend on context, both on the theory being analysed (like classical mechanics, classical electrodynamics, special relativity, ...) and on the field of study (like mathematics, philosophy, physics, ...). In our context, classical electrodynamics has a field ontology, and classical mechanics has a particle ontology. Explanations why $E$ being ontic is less problematic than force being ontic are just elaborations of those basic facts about the typical usage of the words "ontology" and "ontic" in that context.

 

[vanhees71]

For me force is for sure much more ontic than the electric field. A force is directly having an observable consequence, i.e, the acceleration of a particle the force is acting on. Anyway, it doesn't have the slightest advantage for the understanding of the meaning of either quantity in physics to say it is ontic or not.

 

[Killtech]

For me force is for sure much more ontic than the electric field. A force is directly having an observable consequence, i.e, the acceleration of a particle the force is acting on. Anyway, it doesn't have the slightest advantage for the understanding of the meaning of either quantity in physics to say it is ontic or not.

Force being directly observable is a hard oversimplification. If we go into detail, it's way more complicated and nothing is really directly observable all on its own.

forces represent the interaction between different entities, in this case the field and a particle. But we can measure neither of them directly. For one we cannot measure what happens non-locally at all, so we are reliant on entities that carry such information over distances. Usually light waves and particles are regarded as such carriers.

But in any realistic depiction of a measurement, information changes hands quite a few times until it reaches us. Originally it may be carried by light in a scattering event, then it interacts with something and is then converted by additional interactions into an electric signal which is then converted into either something mechanical (like a moving needle visible to the human eye) or processed digitally. How is anything of this a direct observation? The only thing direct is local interaction and self interaction of entities but all actual measurement is very indirect.

You cannot leave out the intermediary entities that transport information. They have to be modeled to be able to make correct predictions. Same goes for their interaction with other entities, however that is different as it itself does not hold any information on it's own but merely mediates the changing hands of existing information between one entity to another.

I think the question of ontic is about who really holds the actual information.

 

[gentzen]

The quote includes momentum along with position as ontic.

I know. But my comment also pointed out that Dürr talked about "primitive ontology" when he highlighted the ontic importance of position.

You were the one who complained to Demystifier: "My problem is that you are not discussing like a scientist. ... You are discussing like a member of school debate team." And now behave like that and just try to score easy points. Such a behavior encourages others to be careful with what they write, and to not acknowledge the questions they did not address yet.

Do you then think that your engagement on the quantum foundations forum is a waste of time? Or if not, what's your motivation for that?

The motivation is the same as when I read theards (in other forums) about perpetual motion, logical errors in relativity and so on. It is clear that it is nonsense, but figuring out why or reading someone's response to such nonsense usually leads to a better understanding for me.

Do you find this funny? What do you expect how one should react?

What do you expect how I should feel, since I talked about ontological commitment and ontological overcommitment, and might be indirectly be responsible for Demystifier trying to find out how to explain such concepts?

If you analyse the ontological commitments of a theory relying on a particle ontology, the conclusion that particle position is the main ontological commitment seems justified to me. You can include momentum too in your ontology if you wish, but the particle ontology does not force this upon you.

 

[martinbn]

I know. But my comment also pointed out that Dürr talked about "primitive ontology" when he highlighted the ontic importance of position.

You were the one who complained to Demystifier: "My problem is that you are not discussing like a scientist. ... You are discussing like a member of school debate team." And now behave like that and just try to score easy points. Such a behavior encourages others to be careful with what they write, and to not acknowledge the questions they did not address yet.Do you find this funny? What do you expect how one should react?

What do you expect how I should feel, since I talked about ontological commitment and ontological overcommitment, and might be indirectly be responsible for Demystifier trying to find out how to explain such concepts?

If you analyse the ontological commitments of a theory relying on a particle ontology, the conclusion that particle position is the main ontological commitment seems justified to me. You can include momentum too in your ontology if you wish, but the particle ontology does not force this upon you.

Fair enough, i will stop.

 

[Demystifier]

For me it is less vague. I went to a similar school system. In fact it is obvious that position is a kinematics notion while momentum is a dynamics one.

Where i went to school/uni mechanics is always divided into statics, kinematics, dynamics.

Good, since now we have finally some common ground to start with, now we can start to address interesting questions. In your view, what's the kinematics of QM?

 

[physika]

$$\ddot x_1 = G \frac {m_2} {(x_1-x_2)^2}, \ddot x_2 = G \frac {m_1} {(x_1-x_2)^2}$$

a description.

 

[martinbn]

Good, since now we have finally some common ground to start with, now we can start to address interesting questions. In your view, what's the kinematics of QM?

I can only speculate. I've not seen the division into kinematics and dynamics outside of classical mechanics. For example the typical spin/polarization cases I would call kinematics.

 

[vanhees71]

Kinematics: States are represented by self-adjoint positive semidefinite operators with trace 1, the Statistical Operator $\hat{\rho}$ of the system. Observables are represented by self-adjoint operators. The possible values of observables are given by the spectrum of these operators. Expectation values of an observable represented by the operator $\hat{A}$ is given by
$$\langle A \rangle = \mathrm{Tr}(\hat{\rho} \hat{A}).$$

Dynamics: There exists an operator $\hat{H}$, the Hamilton operator of the system, such that the time derivative of an observable $A$ is represented by the operator
$$\mathring{\hat{A}}=\frac{1}{\mathrm{i} \hbar} [\hat{A},\hat{H}]+\partial_t \hat{A}.$$
The statistical operator obeys
$$\mathring{\hat{\rho}}=0.$$

One should note that this defines the physical/covariant time derivative. The mathematical time dependence of the statistical operator and the observable-representing observables are only defined modulo an arbitrary (time-dependent) unitary transformation. The time-dependence can be pretty freely chosen by choice of a picture (e.g., the Schrödinger picture lumps the entire time dependence to the statistical operator, and the observable-representing operators are time-independent, except for possible explicit time dependences; for the Heisenberg picture it's the other way; the general case is due to Dirac).

 

[Morbert]

If QM assigns a probability for some path $q(t)$ through configuation space between intial and final configurations $q_i$, $q_f$ as $$p\left[q(t)\right] = \mathrm{Re}\left[\psi^*(q_f,t_f) e^{i\mathcal{S}\left[q(t)\right]}\psi(q_i,t_i)\right]$$and we sufficiently coarse-grain these paths so that the probabilities obey standard probability calculus, we can presumably identify a classical limit where some coarse-grained path $\alpha$ has a probability close to $1$ and is characterised by classical kinematic equations.

 

[Fra]

Given the corroborated standard QM, IMO the "kinematics" refers to the states of the ensemble, ie. statistics, in line with what Vanhees writes.

But the interesting part is the, why is this not satisfactory to everyone? (and this dissatisfaction is what drives these discussions)

I will speak for myself, and one reason, is that the "ontics" at statistical levels, does not seem to have a weel understood correspondence with to the "ontics" at agent(=observer) level. And this is a serious issue.

One may object to this, and not that the physicists in the scientific commmunity indeed has created this ontic. This is true, but this is at a time scale that is HUGE compare to the interaction time scales! This is why THIS "understanding" is not the same as understanding the nature of interactions themselves. For the group of observes (human scientists) to infer these states and laws at statistical level, one need to prepare and repeat some experiments many many times. This is not a problem, or wrong per see, it's just that it does not correspond to an inside agent. And the "laws" that are created or discovered from the scientific process, have a different constructing principle than the reason why laws exist in the first place. An extension of this issue also leads to Smolins critique to the "Newtonian paradigm". The reason is that the decompisition of kinematics and dynamics is rather an artefact of the human perspective to atomic physics. The mathematics that fits well, seems to break down when it comes to cosmology and unification and finetuning questions.

For someone seeing from the agent perspective, these things are a major defect and can not be ignored. It's just obvious that this is incomplete.

Edit: Unless it's not clear to all. The use of the word AGENT instead of OBSERVER for me is motivated by that it tries to emphasize that the observer need to be neither human, a physicists or even be a biological system. AGENT for me puts emphasis on that the observer is a physical system, and is made of whatver all thing are made of (matter). So an AGENT does not need to have a brain, but it has to have some kind of ENERGY and if localized - MASS. The physical limits of the AGENT, also constrains the physically possible inferences, via throttling of resources. For example communication between agents of grossly different masses, is either necessarily lossy, or in exceptional cases a tuning is in place that allows a non-lossy compression. These abstractions are probably easier to motivate when one ponders about black holes. For atomic physics all this may seem irrelevant, but I still don't think it is. If information is "lost" wrt and agent, it might as well mean it's encrypted beyond decrypting capabilites of that agent.

/Fredrik

 

[physika]

"The decompisition of kinematics and dynamics is rather an artifact"

Similar ideas from Spekkens.

https://arxiv.org/abs/1209.0023

...