- 13,415
- 4,218
What is the operational interpretation of probability in QM? Choose whatever interpretation you favor.
The minimal statistical interpretation is neither minimal nor statistical
- Context: Graduate
- Thread starter Demystifier
- Start date
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
133 replies · 10K views
- 4,464
- 2,506
Yes, but the topic here is the minimal statistical interpretation.atyy said:
- 14,781
- 7,420
But why exactly it changes, in my opinion?martinbn said:
- 4,464
- 2,506
Are you asking me?! The point was that in the statistical interpretation you cannot say that.Demystifier said:
- 15,171
- 3,378
martinbn said:
Well, you can make up your own mind whether the "ensemble" interpretation is powerful enough to change the mathematics of quantum theory.
IIRC, when @bhobba and I discussed, although he is a fan of Ballentine while I am not, bhobba's own Ensemble interpretation includes a postulate equivalent to state reduction, which is that one can at the end of decoherence in a measurement take the reduced density operator to be ignorance interpretable.
Also, the argument against state reduction is incomprehensible if one takes an operational view of quantum theory. In an operational view, who cares what absurdities happen to the state, since it is not necessarily real and just a tool to calculate the probabilities of measurement outcomes.
That @vanhees71 cares about "causality" (about which he is confused) shows that he thinks the quantum state is real, which would also make the objection to state reduction sensible. Very few people would consider such an interpretation to be minimal.
- 4,464
- 2,506
Not sure what this has to do with the discussion here! All I said is that I disagree with Demystifier, because the minimal statistical interpretation doesn't need the reduction postulate (and it doesn't need to be agnostic to be minimal). And I wrote how I think it should be phrased in the language of the interpretation. I don't claim that the interpretation is problem free (it seems that none is), nor that one should subscribe to it, nor that I do, just what it says and that it doesn't say that it has a reduction postulate. If you think I am wrong somewhere comment on that. But I don't see that point of vague references to different discussions or opinions.atyy said:
- 4,464
- 2,506
Of course, if you take a different point of view from the statistical, the statistical will seem incomprehensible. That applies to any two different points of view.atyy said:
- 11,094
- 3,940
atyy said:
Just to be clear what the ignorance in my interpretation is. At the end of decoherence you have a mixed state. If it is a proper one things are more sensible as far as everyday experience goes (ie it actually is in one of the possible pure states of the mixed state - but we do not know which one - only the probability) or an improper one where the act of measuring the mixed state itself, somehow gives the result. Trouble is at least so far nobody has figured how to tell the difference or even if it is worth worrying about.
Thanks
Bill
Last edited:
- 14,781
- 7,420
That was certainly not my point.martinbn said:
- 4,464
- 2,506
That was my point! I started by saying that I didn't agree with you. What is your point, besides the one in the first post? And what do you find not ok with my argument?Demystifier said:
- 15,171
- 3,378
bhobba said:
Wouldn't it make a difference for a selective measurement, where there are subensembles with conditional states labelled by measurement outcomes?
Whereas for a non-selective measurement (where the measurement result is discarded), certainly there is no difference between purely unitary decoherence and measurement with state reduction with respect to the observables that can be measured.
- 14,781
- 7,420
None.martinbn said:
- 24,488
- 15,060
I don't care whether my statements or ontological or epistemic or whatever. These are just philosophical categories which don't mean much for the fact that we are able to observe things in Nature by our senses and using real-world apparati in our labs.Lord Jestocost said:
If there is any interpretation consistent with experiments (there are no proofs in natural sciences only the agreement between a theory and all experiments done so far ;-)) it's the statistical interpretation: What QT claims to be "random" given a "preparation" has turned out to be "random" in experiment, and hitherto nobody has come up with a way to make what's "random" in QT "determined" in a new theory beyond QT. So we have to live with QT for the time being.
That Fock argues against any statistical interpretation is clear. The poor Russian physicists in soviet times had to make arguments against any implication of their theories being not in accordance with the (philosophical!) doctrine of dialectic materialism. It's my favorite example for the "value" of philosophy in the sciences. You can find an amusing (but at the time dead serious) philosophical discussion in Blokhintsev's book on QM.
- 4,464
- 2,506
martinbn said:
So, you have not other point, and you find my argument ok. Did you change your mind? I guess not. Why not?Demystifier said:None.
Gold Member
- 1,106
- 1,009
bhobba said:
Decoherence cannot effect that a pure state evolves into a mixed state - the math is unambiguous. As Daniel Greenberger puts it - and that is - so to speak - "the state of the art":
"Decoherence merely scrambles up the phases so that it would be impossible, from a practical point of view, to unscramble them. But it does not solve the basic problem that a pure state does not evolve by the Schrödinger equation into a mixed one."
Gold Member
- 1,106
- 1,009
Fra said:
One should not mix up statements about the post-measurement situation with statements about the pre-measurement situation. Maybe, the following might help. As Maximilian Schlosshauer puts it in “Decoherence, the measurement problem, and interpretations of quantum Mechanics”, Section B. 1. Superpositions and ensembles (https://arxiv.org/abs/quant-ph/0312059):
“Put differently, if an ensemble interpretation could be attached to a superposition, the latter would simply represent an ensemble of more fundamentally determined states, and based on the additional knowledge brought about by the results of measurements, we could simply choose a subensemble consisting of the definite pointer state obtained in the measurement. But then, since the time evolution has been strictly deterministic according to the Schrödinger equation, we could backtrack this subensemble in time and thus also specify the initial state more completely (“postselection”), and therefore this state necessarily could not be physically identical to the initially prepared state on the left-hand side of Eq. (2.1).“
Last edited:
- 24,488
- 15,060
The ensemble interpretation is just the math of quantum theory and no additional assumptions like collapse which is outside of the math of quantum theory, claiming that one needs another "mechanics" or "dynamics" for the interaction of a particle with a measurement device (or in some flavors any macroscopic object) thanatyy said:
what's described in QT. There's no hint that such a quantum-classical cut is needed though.
To take an argument FOR state reduction is incomprehensible if one takes and operational view of quantum theory. What quantum theory describes about the "operations" done in an experiment is precisely this: Some object is prepared in some specific way (represented by the statistical operator) and after some time a measurement of an observable of the system is done, giving some (usually random) value with a probability distribution predicted by the theory, and you can check this prediction on an ensemble of equally prepared system. There is no need to know, in which state the system after the measurement might be. Usually it's not even present anymore after the measurement in the previous form (e.g., a photon usually gets absorbed when detected).
I may be confused, but local relativistic QFTs are, by construction, causal. There cannot be a causal influence across space-like distances in Minkwoski space. That's a mathematical feature of the theory and directly contradicted by the state-reduction/collapse assumption.
Fra
- 4,444
- 739
If this boils down to Ballentine's books (which i don't own), perhaps I miss some subtle point here.Lord Jestocost said:
Talking about mixing, my point was to not mix up how to infer a LAW vs how to apply it for expectations in a single situation. Fock seems to have an issues that the LAW that is inferred at statistical level, can't be applied for predicting the "potentially possible" in a single experiment. I still don't get the point.
Ensemble of "experiments already carried out" ~> statistical law that implies future distribution from past distribution (short time scales). This part of science creation is not about predictions, its about extracting laws from experimental data (timeless in this case). It's at this level that we can claim to have deterministic evolution of generalized probability distribution (represented via wavefunctions).
Next, applying this to NEW single experiments, similar to the situation in where the LAW was inferred, the task is not to extract new laws (we have it already), the task is just to predict the future, from the present, in an individual experiment; given the law. The rational expectation here of an individual expertiment, is just to used the probabilistic rules for the corresponding ensemble, right? This does not necessarily imply that the wavefunction has any ontological meaning for the system in a single individual experiment, one must not forget that the LAW is corroborated at the statistical level only. The "states" to which deterministic evolution applies, is a statistical state (represented mathematically by a vector). How can it not be a statistical state? what's the logic?
(The ponderings about subensembles and post-selection means we are considering a DIFFERENT system and different measurement setup and thus different observer in the abstract, that does not solve anything as i can see. From my perspective, it makes not sense to "transport" rules of reasoning between two different observers as if we are in a euclidian "reasoning space". This is done all the time, and is disturbing, and only makes the fog bigger. I don't think this has much to do with Focks objection?)
/Fredrik
- 15,171
- 3,378
vanhees71 said:
Well, if there is no state, there is no unitary evolution, so this contradicts your claim that there is only unitary evolution of the quantim state.
- 11,094
- 3,940
Lord Jestocost said:
Yes even the Microwave Background Radiation will do it. So valid point - if it was a proper mixed state it will soon be scrambled into a new mixed state - but is the new mixed state itself proper? That is how do you tell the difference between a proper and improper mixed state? I have no idea how to do that. If Atty can elaborate on his proposal it would help. Of course if it is a proper mixed state something else may also be going on we do not know at this epoch.
Thanks
Bill
Last edited:
- 4,464
- 2,506
Just to stress that this is not about the ensemble interpretation of quantum mechanichs. It is about an ensemble interpretation of superposition.Lord Jestocost said:
- 4,464
- 2,506
No, there is no state of the system! It is either absorbed, or it is no longer the same system. But the state of a given system evolves only unitarily.atyy said:
- 11,094
- 3,940
dextercioby said:
Formally if you read a math book on QM it is the Kolmogorov axioms eg:
https://www.amazon.com/dp/0387493859/?tag=pfamazon01-20
As far as the Ensemble interpretation goes operationally it is frequentist.
For Copenhagen it doesn't really specify, but it fits naturally with the Bayesian interpretation. And there is an interpretation called Bayesian which obviously uses the Bayesian view:
https://www.quantamagazine.org/quantum-bayesianism-explained-by-its-founder-20150604/
Thanks
Bill
- 11,094
- 3,940
atyy said:
Well if you take Gleason literally the state is just an aid to calculation. It is taken as unitarty because then it is normalised when you calculate probabilities making things easier. An actual pure state is really an operator |u><u|. It is just usually mapped to |u> again for convenience - but the 'phase' is irrelevant.
Thanks
Bill
- 14,781
- 7,420
If you say that I find your argument OK, who am I to question it?martinbn said:
- 4,464
- 2,506
If you don't find it ok, can you please tell me what you think is not ok.Demystifier said:If you say that I find your argument OK, who am I to question it?
ps: This is a discussion thread started by you. I would have thought that you'd be interested in the topic. So why do I have to pull one word answers from you. You are not a spy and this is not an interogation. If you are not interested fine. But it is hard for me to take siriously short "witty" responses with lots of emojies.
- 24,488
- 15,060
We argue in circles again. Unitary time evolution applies to the dynamics of a CLOSED system. For open systems you get an effective description, which is often even a classical theory (Kadanoff-Baym equations, (quantum-)transport or Langevin equations for particles and/or quasiparticles, hydrodynamics, point-particle mechanics...).atyy said:
In classical dynamics (point particles and/or fields) for open system also the fundamental dynamical laws, i.e., the Euler-Lagrange equations of the corresponding action principle, are substituted by corresponding effective descriptions. So this is nothing specific to quantum theory.
- 15,171
- 3,378
vanhees71 said:
But wouldn't you say that if you enlarge the system you would still get a closed system, and by partial tracing you can get the quantum state of the subsystem?
Fra
- 4,444
- 739
Is this what Demystifier means...?
"because the probability (conditional probability, to be more precise) suddenly changes when new information (about the measurement outcome) arrives."
But this conditional probability (the one that collapses), belongs not to the whole ensemble, but to the agent that is part in the interaction in one specific experiment. Is this what is corroborated? I woud say no.
"Therefore the state, which is nothing but the probability amplitude, suddenly changes as well, and this sudden change is called collapse (or state reduction)."
I this is your point, then i say the whole ensemble does not "collapse" just because of ONE experiment. To ask what happens after the measurement, means we have a new system. QM only describes what happens "in between" measurement, which is also when its "closed. We "open" the system for masurement, then close it, to open it later.
An ensemble is robust by construction, but an agent doing a randow walk will face "brownian like" collapses all the time, its part of the game. The only problem is that this interpretation where the probabiity is attached to one individual agent participating and updating his information, is sensible and plausible and IMO what a measurement theory SHOULD describe, but its when you think deeper about this I've realized that the mathematics of QM does not describe this. At best it describes a crippled picture with dominat agents sitting at infinit observing but not needing to face backreactions.
There are several ways to suggest how this situation can be improved but that involves radical speculation and explicit modification (or reconstruction) of QM. But until the, I think the statistical interpretation is the most accurate one. The problem for me is not the interpretation, but the theory itself when it comes to unification. Once the theory is reconstruted, obviously we need a "new interpretation" as well, and in my fantasies then the preferred one is the "inside agent" interpretation. But that interpretation fails for current formalism.
/Fredrik
"because the probability (conditional probability, to be more precise) suddenly changes when new information (about the measurement outcome) arrives."
But this conditional probability (the one that collapses), belongs not to the whole ensemble, but to the agent that is part in the interaction in one specific experiment. Is this what is corroborated? I woud say no.
"Therefore the state, which is nothing but the probability amplitude, suddenly changes as well, and this sudden change is called collapse (or state reduction)."
I this is your point, then i say the whole ensemble does not "collapse" just because of ONE experiment. To ask what happens after the measurement, means we have a new system. QM only describes what happens "in between" measurement, which is also when its "closed. We "open" the system for masurement, then close it, to open it later.
An ensemble is robust by construction, but an agent doing a randow walk will face "brownian like" collapses all the time, its part of the game. The only problem is that this interpretation where the probabiity is attached to one individual agent participating and updating his information, is sensible and plausible and IMO what a measurement theory SHOULD describe, but its when you think deeper about this I've realized that the mathematics of QM does not describe this. At best it describes a crippled picture with dominat agents sitting at infinit observing but not needing to face backreactions.
There are several ways to suggest how this situation can be improved but that involves radical speculation and explicit modification (or reconstruction) of QM. But until the, I think the statistical interpretation is the most accurate one. The problem for me is not the interpretation, but the theory itself when it comes to unification. Once the theory is reconstruted, obviously we need a "new interpretation" as well, and in my fantasies then the preferred one is the "inside agent" interpretation. But that interpretation fails for current formalism.
/Fredrik
Gold Member
- 1,093
- 822
As an aside I think the minimal ensemble interpretation can be readily extended to successive measurements without having to worry about conceptualising new ensembles throughout. E.g. If we have a preparation ##\rho## and carry out a measurement of observables ##\hat{A}=\sum a_i\Pi_{a_i}## and ##\hat{B}=\sum b_i\Pi_{b_i}## at times ##t_1## and ##t_2## respectively, we can construct a compound observable $$\hat{C} = \sum_{i,j}a_ib_j\Pi_{a_i}(t_1)\Pi_{b_j}(t_2)$$The user can then compute probabilities and conditional probabilities like
\begin{eqnarray*}
p(a_i\land b_j) &=& \mathrm{tr}\{ \rho \Pi_{a_i}(t_1)\Pi_{b_j}(t_2) \}\\
p(b_j | a_i) &=& \frac{\mathrm{tr}\{ \rho \Pi_{a_i}(t_1)\Pi_{b_j}(t_2) \}}{ \mathrm{tr}\{ \rho \Pi_{a_i}(t_1) \} }\\
\end{eqnarray*}Without having to conceptualise any intermediate ensemble ##\rho(t)##.
\begin{eqnarray*}
p(a_i\land b_j) &=& \mathrm{tr}\{ \rho \Pi_{a_i}(t_1)\Pi_{b_j}(t_2) \}\\
p(b_j | a_i) &=& \frac{\mathrm{tr}\{ \rho \Pi_{a_i}(t_1)\Pi_{b_j}(t_2) \}}{ \mathrm{tr}\{ \rho \Pi_{a_i}(t_1) \} }\\
\end{eqnarray*}Without having to conceptualise any intermediate ensemble ##\rho(t)##.