On 1/06/2017 4:43 am, Bruno Marchal wrote:
On 31 May 2017, at 04:01, Bruce Kellett wrote:
On 30/05/2017 9:35 pm, Bruno Marchal wrote:
On 30 May 2017, at 11:28, Telmo Menezes wrote:
I get your point with decoherence.
Again, I would say that it all depends on theories of mind. What does
mind supervene on? Perhaps it is true that every single coupling with
the environment prevents the current observer state to become
compatible with other branches. But can we be sure? I feel that such
certainties come from a strong belief in emergentism (which I cannot
disprove, but find problematic).
It is impossible to recohere the past, FAPP.
But only FAPP. To make the blue T-rex interfereing with the
red-T-rex, we must erase the trace of particle interaction between
the T-rex in its whole light-cone, and this without forgetting the
particles "swallowed" by the black-holes, etc. It is just completely
impossible, but to derive from that the unicity of the past, is, it
seems to me (and you if I understood well) is invalid.
I think the recoherence of paths that have completely decohered is
more than just FAPP impossible, I think it is impossible in
principle. One major problem with recoherence in general is that
information leaks from the paths at the speed of light (as well as
less slowly for other interactions). Since this vital information
goes out along the light cone, it can never be recaptured and
returned to the original interaction.
In QM + special relativity OK. (note that to me QM + special
relativity => no collapse (and even the Many dreams, but we have
agreed to disagree on this if I remember well).
Reasoning in QM without SR is not very profitable. Besides, QM + SR does
not particularly imply MWI -- it is perfectly possible to have a
consistent collapse model of QM+SR. I know you don't agree because of
the non-locality implied by EPR, but this non-locality is not removed in
MWI, regardless of what you might say.
Consequently, indispensable phase information is lost *in principle*,
so the recoherence is, in general, impossible.
OK. (I was reasoning in naive classical QM)
Of course, with carefully constructed systems, where the loss of
information along the light cone is prevented, recoherence is
possible in special circumstances, but not in general.
From this, the uniqueness of the past of any decoherent history is
assured. So deriving the unicity (if I understand this use of the
word) is by no means invalid -- it is proved.
In QM + SR. OK.
Even if one encounters one of those rare situations in which
recoherence is achieved, that still does not invalidate the
uniqueness of the past history -- recoherence, if it occurs, simply
means that no new branches are formed at that point, so the
decoherent history remains unique.
OK. That might suggest that we identify our indistinguishible past in
arithmetic, if we assume mechanism. I use the Y = II principle, or the
"quantum" linearity of the tensor product "@": we have that a @ (b +
c) = (a @ b) + (a @ c).
That makes sense.
Linearity is the heart of QM. It is linearity that allows
superpositions, and leads to all the "quantum weirdness".
FWIW, you
are expressing my own understanding of the situation: there can be no
superposition of red and green screens or dinosaurs, or dead and
live cats,
because there can be no quantum superposition of macroscopic objects.
Superpositions of wave functions are only possible for systems
isolated from
interaction with their environment, which is why quantum computers
are so
fricking hard to make: keeping aggregates of particles isolated from
interactions with the surrounding environment is exponentially more
difficult as the system grows in size.
The main question for me is this: can two branches hold different
observer states, if they differ only by things that are not
observable?
I would say no, intuitively. I would even say "no" just for the
things not observed, even when observable.
I previously answered Telmo's question in the affirmative, viz., two
fully decohered branches will hold different observer states, even if
the differences are not observed or observable. So if some trivial
physical event happens to your body, such as the decay of a K 40
nucleus in your foot, this would not be noticeable, or even
particularly observable even if you were looking for it. But such an
event causes at least two branches to form every instant -- one in
which the decay has occurred, and one in which it has not. And since
this is a beta decay, a neutrino is lost along the light cone in
every case of decay. Perfect recombination of the branches is, then,
according to the above argument, not possible. You might object that
this decay in my toe did not alter my conscious state -- that is
correct, but there are now two copies of the Moscow man as in step 3,
My mistake here - I misremembered step 3. Moscow is a target, not the
origin. I should have said two copies of the Helsinki man.
But those state difference are accessible to the observers, and
indeed, only this makes the analogy with step 3 working.
In MWI, the differences are not observable by anyone. Any observer has
access to only one branch, so only one copy. They can say nothing about
the other branch.
With computationalism, the two identical states at the correct
susbt-level, if never distinguishable, will not add ways to the
probabilities (that would entail that a computer with double cable
would have a higher probability). The measure remains on the first
person view, which eventually needs the self-referential definition.
usually I just say that if in step 3 you add a reconsitution box in
Moscow, exactly identical, the P(W) remains equal to 1/2, but a slight
difference in the future histories of the two guy in Moscow, will
changes the probability into 1/3 (but that is a typical open problem).
I suspect that this follows from my mistake about which man was
duplicated. I did not mean that there was a duplication *after* the
initial formation of two branches.
and these can evolve in different directions while each remains
unaware of the existence of the other. They can never recombine and
compare diaries!
OK.
But this has to be tempered by the fact that any interaction will
count as an observation, making super-exponentially hard to indeed
recover a macroscopic superposition in the past, even the very close
past. Of course, that might change the day we succeed in building a
fault tolerant (topological perhaps) quantum computer.
That will not help in the general case. Our future quantum computer
might be able to delay decoherence for some useful finite time, but
that still only retaines the superposition in the said computer, it
does not help with recombination of decohered branches in general.
In general no. But I was in a thought experiment in classical QM, with
some T-rex having isolated a quantum qubit.
Unfortunately, the T-rex missed them. yet, if a T-rex made a solid
topological quantum qubit, in the state 0+1, we would have a past
with 0, and a past with 1, as long as we don't look at it. I read,
already a long time ago, some experimental evidence of temporal
Bell's inequality going in this direction, and I think we don't even
need to test them, as we get them with the usual Bell's inequality
violation, if we accept special relativity (and some amount of
physical realism (not the full materialism, to be sure).
The temporal version of Bell's inequality simply shows that the
ubiquitous non-locality of quantum entanglement occurs even over
time. And despite your protestations to the contrary, it is now
generally accepted that MWI does not remove the essential
non-locality associated with entangled states.
Of course, it assures them in all branches, where indeed Aspect like
experiences can be made. It seems to me that we did agree on this:
that non-locality does not entail any physical influence in the past.
That does happen in the unique universe view though; even if there is
no possible communication of information is done.
Non-locality means that there is no physical transfer of information,
but that there is non-physical(non-local) transfer of information. But
this information transfer cannot be used for signalling. Signalling is
possible only with actual; physical transfer, a consequence of SR and
the fact that 'information' is physical.
This non-locality is even more evident in the more recent delayed
choice experiments that use entangled photons to manipulate photon
polarization states non-locally.
I have no problem with one-branch observable, apparent, non-locality.
I have a problem only with the action at a distance that you need in
case you assume one contextually well defined physical reality.
Non-locality is not removed in MWI as you appear to believe.
For me the abandon of the collapse is the solution of the EPR
"paradox", and Aspect experience is somehow the confirmation of our
belonging to macrosuperposition.
The non-local (paradoxical) nature of EPR remains even without collapse.
As on the previous occasion we discussed this, you were unable to
demonstrate where the notion of 'collapse' is used in Bell's theorem -
all Bell requires is that measurements give results, and that is what
the whole of physics is based on: in MWI as well as in any other
interpretation.
Bruce
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.
