CATEGORY:
Blog
[back]
TOPIC:
Erasing the Quantum Past to Avoid Future Catastrophe
[refresh]
FQXi Administrator Zeeya Merali wrote on Apr. 21, 2009 @ 12:37 GMT
Well done to Nature of Time contestant Saibal Mitra. His essay “
Can we change the past by forgetting?” which you can download here (
pdf) has hit the headlines, appearing in
New Scientist.
The article by Marcus Chown sums up Mitra’s idea that one might (in very idealised circumstances) be able to switch to a different parallel universe to avoid an impending disaster, like an asteroid impact.
 |
| Ouch |
From Chown’s article:
“To understand how the many-worlds scenario could allow a future disaster to be avoided, says Mitra, consider a hypothetical machine intelligence that regularly backs up its memory. If it encountered a glitch, for example, it could reset its memory to, say, the previous day's state.
“Imagine that on learning of an impending disaster - perhaps a catastrophic asteroid strike on its planet - the machine resets its memory. Now, an observer sat next to the machine can verify that the "same machine" will still face disaster after the reset. But from the perspective of the machine's reset memory, the state of the universe in the many-worlds scenario becomes "undetermined". After all, for all the machine knows, the reset probably occurred for a mundane reason, such as a crash of its operating system.
“The next part defies our natural instincts: according to the many-worlds interpretation, all of these undetermined possibilities actually exist and open up to the machine. Even though it followed one particular history up to its resetting, it can be dealt a new card, says Mitra. So, from its unwitting perspective, the machine could "switch" to a parallel universe.”
 |
| If R2D2 reset his memory, could he have saved us from ever meeting the Ewoks? |
Hmm... I’ll be honest, I’m not sure I buy that all the undetermined possibilities of the machine’s measurements re-open just because the machine forgot that it made the measurements. (Would that be the equivalent of Schrodinger’s cat reverting to being both dead and alive in my universe, because I forgot I’d already observed its fate?) Why does this amnesia change the machine’s entire reality, when it’s already entangled on one timeline? And even if it did, on re-measuring, I don’t see it gains very much—the machine just re-splits again, with some future versions escaping disaster while others experience disaster, which seems pretty much the same as before.
But our
Max Tegmark has commented positively on the paper in the article, saying that it could potentially apply to “future beings whose minds are quantum computers.” (He may well be thinking back to a recent FQX discussion about how in the future
humans may meld with machines.)
Anyway perhaps I’m missing something. I’m willing to
re-read Mitra’s essay to try and dis-entangle (pun intended) what it means. Graeme Stemp-Morlock’s article on parallel universe pioneer (and Cold War “Dr Strangelove”)
Hugh Everett’s tumultuous life may help too.
this post has been edited by the forum administrator
post approved
John Merryman wrote on Apr. 22, 2009 @ 02:03 GMT
Why do I keep getting the feeling this is one of those gag universes that doesn't quite make the cut?
post approved
Brian Beverly wrote on Apr. 22, 2009 @ 04:18 GMT
Did the asteroid take out the laws of thermodynamics too?
post approved
Eckard Blumschein wrote on Apr. 22, 2009 @ 15:19 GMT
Given Mitra was correct, who would benefit, who would be hurt? While I ignored his essay, I could change my mind if quantum computers will actually work as promised.
So far I dislike any attempt to deny what Claude Shannon pointed out: Only the past can definitely not be influenced while any prediction of the future is more or less uncertain.
I am suffering form implications of belief in Einstein's belief that irreversibility is exclusively based on reasons of probability and therefore "the division into past, presence and future is merely an albeit obstinate illusion". Experts of signal processing do accordingly not believe me that future time does not matter in spectral analysis.
My method as well as physiology are not by chance superior. This hurts those a lot who cannot imagine even canonical theories being wrong.
post approved
John Merryman wrote on Apr. 22, 2009 @ 15:56 GMT
The reality is that time is not a dimension along which we travel from a singular past to multiple possibilities of the future. It is a process by which the possibilities of the future collapse into the series of events which are past. The earth doesn't travel along a meta dimension from yesterday to tomorrow. Tomorrow becomes yesterday because the earth rotates.
All the physicists in the world, working out the most complex mathematical theories imaginable, are not going to change that simple fact. They are only going to look that much more ridiculous when their house of theoretical cards collapses under the weight of its own incongruities.
post approved
amrit wrote on Apr. 23, 2009 @ 10:41 GMT
John Earth moves in timeless space, clocks also.
Time is run of clocks in eternity.
yours amrit
post approved
John Merryman wrote on Apr. 23, 2009 @ 16:18 GMT
amrit,
Yes, but explain that to the people who insist it is a dimension going from past to future and feel compelled to argue a many worlds scenario to explain how the singular events of the past transition to the multiple possibilities of the future. Since they are the gatekeepers in physics and the science media are their enablers, the only check will be how credulous those who finance them remain.
post approved
Lawrence B. Crowell wrote on Apr. 24, 2009 @ 02:37 GMT
The resurrection of a Schrodinger cat (kitten) has been performed with quantum optical states. This rests on the fact that the reduction of a quantum state is not perfect. The outcome is 99.99...9% certain to be, while the other .00...1% remaining is some small amplitude which persists. By some fancy entanglement procedures the wave function can be restored.
I am not a big fan of many world interpretation, or any interpretation for that matter. The extension of this appeals to an MWI sort of thing. Clearly this can't be done on a practical basis.
Lawrence B. Crowell
post approved
Brian Beverly wrote on Apr. 24, 2009 @ 08:27 GMT
Lawrence,
Thank you! The goal is always to explain experiments and provide unique testable predictions. Often the context of the MWI in physics is sensationalism without accomplishment (the Paris Hilton of quantum interpretations). If the MWI does not explain the experiments then it is another dead end.
However, I like the FQXI because they are hedging their bets and supporting MWI researchers and others. Declaring a null hypothesis too early would be a shame because it may need more time to reach maturity.
post approved
Jason Wolfe/wulphstein@gmail.com wrote on Apr. 24, 2009 @ 08:31 GMT
How do find out more about entanglement methodolgy, how it's done? I have a feeling it's going to be important.
post approved
John Merryman wrote on Apr. 24, 2009 @ 09:47 GMT
The energy doesn't collapse, or expand. Entanglement is a function of its configuration. What collapses are the possibilities of that configuration. Since they are possibilities, not actualities, there is no energy loss. If multiple possibilities were to emerge as actualities, that would require expanding overall energy to manifest them. Time is not a dimension along which events exist. It is a process by which events emerge and are replaced, as the energy remains constant.
post approved
Lawrence B. Crowell wrote on Apr. 25, 2009 @ 03:03 GMT
I will have to look this up. I read about this experiment a year ago or so. I think it was in the AAAS Science.
Lawrence B. Crowell
post approved
Jason Wolfe/wulphstein@gmail.com wrote on Apr. 25, 2009 @ 08:03 GMT
John,
If a wave amplitude Psi*Psi describes a particle, and every particles is connected to every other particle (mostly), then don't the wave amplitudes and the connections (superstrings/spinors/something) have to be made out of the same thing?
post approved
Lawrence B. Crowell wrote on Apr. 25, 2009 @ 11:59 GMT
The idea is that while a quantum wave function might so called collapse, it does not do so completely. So the amplitude which is not selected by the measurement still persists, though very very small. Yet if you have an entangled state with the original wave function that entangelement still exists after the selection. Thus by a pi-pulse on your entanglement state you can "resurrect" the pre-measurement state.
Don't expect any practical result from this. The conditions in the set up here are pretty specialized. A generic measurement puts off diagonal superposition or entanglement portions of a wave function into the environment of reservoir states in a random fashion. So this can't be done generically. Don't expect to be able to navigate through an MWI plurality of worlds any day soon --- or ever for that matter.
Lawrence B. Crowell
post approved
John Merryman wrote on Apr. 26, 2009 @ 01:29 GMT
Jason, Lawrence,
So it would seem there is some underlaying state, vacuum, entanglement, etc. from which all this fluctuation, particles, waves, positive/negatives, all the details we measure/observe, etc. emerge, but are still attached.
post approved
Lawrence B. Crowell wrote on Apr. 26, 2009 @ 11:10 GMT
A decoherence of a wave function is the loss of off diagonal terms in the density matrix,
rho_{ij} = |Y_i)(Y_j|,
where these phases are not destroyed but rather ranomly taken up by a reservoir of states. This reservoir of states constitutes the environment. So the standard measurement is one where this is completely lost. Yet if our state of ineterest is entanbled with some other system we have control over this lost entanglement phase can be restored.
This does lead to some questions about the qauantum-classical correspondence. This implies that the classical world is really an approximation, where the classical-like einselected quantum state is really still quantum mechanical with a superposition with very small amplitudes.
Lawrence B. Crowell
post approved
John Merryman wrote on Apr. 26, 2009 @ 12:48 GMT
Chaos and order, like yin and yang.
post approved
John Merryman wrote on Apr. 26, 2009 @ 18:09 GMT
Space blob' baffles astronomers
It might not look like much, but this image represents one of the most distant objects astronomers have ever seen, 12.9 billion light years away.
It is a "Lyman-alpha blob" and is 55,000 light years across - as large as present-day galaxies.
Though younger such blobs have been found, "Himiko" confounds the idea that such large objects grew more recently by the merger of smaller ones.
http://news.bbc.co.uk/2/hi/science/nature/8007844.stm
post approved
John Merryman wrote on Apr. 26, 2009 @ 21:25 GMT
Lawrence,
Order is subjective, but objectivity isn't chaotic, it just doesn't have a single frame of reference. We just can't think in this infinite multi-dimensionality.
So we are stuck in this cycle of expanding awareness and collapsing order, where our knowledge increases exponentially, as our understanding of the logical foundation fades. The future expands, while the past contracts.
post approved
FQXi Administrator Zeeya Merali wrote on May. 2, 2009 @ 12:59 GMT
Lawrence, I wrote an article for Nature last year about "uncollapsing" a quantum measurement: http://www.pas.rochester.edu/~jordan/Nature-news-uncollapse.
pdf
Is that what you're referring to? Those experiments have successfully brought systems back from the brink of collapse. If fully collapsed, I don't think it works, so as you've mentioned, extending this to saving machine intelligences from asteroids in some MW scenario isn't practical.
post approved
Saibal Mitra wrote on May. 2, 2009 @ 15:52 GMT
Nice to see my essay featured here! After I submitted the essay and after the deadline had passed, I noted a few errors. I corrected these and put a corrected version on the arXiv:
http://arxiv.org/abs/0902.3825
The arXiv version is a bit more rigorous. I also omitted the discussion about Tegmark's quantum suicide experiment.
It is true that the probability of the observer...
view entire post
Nice to see my essay featured here! After I submitted the essay and after the deadline had passed, I noted a few errors. I corrected these and put a corrected version on the arXiv:
http://arxiv.org/abs/0902.3825
The arXiv version is a bit more rigorous. I also omitted the discussion about Tegmark's quantum suicide experiment.
It is true that the probability of the observer having to face a disaster cannot be changed. Ultimately this is due to the fact that unitary time evolution conserves probabilities.
The issue raised about the observer already being entangled with the rest of the universe is adressed in the article. Decoherence leads to a system evolving into a global superposition. You cannot demonstrate quantum inteference effects involving two decohered branches. Probabilities derived by assuming that the different branches exist will thus always agree with the predictions of collapse interpretations that assume that only one branch really exists. Nevertheless, the interpretation of the probabilities is different in the MWI.
One can object to this by claiming that the observer after memory resetting is located on the same branch as he was in before memory reseting. In the article I explain why this is not the case. One has to define precisely what the observer is in terms of the degrees of freedom described by the theory. Of course, one can only that in general terms.
In the article I focus on classical machine observers. I assume that you can have a classical computer that can run some program which will implement a conscious observer. Whatever the macine is aware of is then fully specified by some string of zeroes and ones that specifies the states of the internal memory.
Then, because we are treating the whole situation using quantum mechanics, we need to also need to include the classical machine states in the quantum description. This is done as follows. One can specify the quantum state of the universe by specifying the (classical zero/one) states of each bit and then supplement that with whatever extra information is needed to specify the exact quantum state of the computer and the rest of the universe.
In the article I refer to the classical one/zero description of the bits the (macro) state of the machine. the extra information contained in the quantum state of machine is considered to be part of the "rest of universe". The fact that the machine is "classical" then means that its macro state as defined by the bitstring specifying the memory state will always be entangled with the rest of the universe.
Is it now reasonable to assume that this machine observer can be aware of any information that is not included in the macrostate, i.e. the bitstring defining its memory state? The answer must be no, because we defined the memory state such that it would contain everything the machine is aware of.
This then implies that reseting the the memory state to some bitstring will lead to a state of the form:
|bitstring) [|memory reset for reason 1) + [|memory reset for reason 2) + ...]
So, the bitstring specifying everything the machine can possibly be aware of factors out of the superposition. It is true that if you look at the precise quantum state of the transitors of the computer then the information of why the memory was reset would already be found in there, as the exact quantum state of the computer will be completely entagled with the rest of the universe. But, again, that information does not exist at the level of the bitstring and thus the computer cannot be aware of that.
From the perspctive of the machine, the rest of the universe exists in a superpostion of different states in which the memory was reset for different reasons. The outcome of a measurement to find out that reason is then not pre-determined.
One can now argue that since the prior probability of ultimately having to face a disaster does not change, it doesn't really matter if one performs memory resetting or not. But one would still have to admit that once you are faced with a disaster, you should expect to almost surely escape the disaster, at the expense of your other copies who are doing memory resetings for other reasons.
You can also look at this from the opposite perspective. Suppose you are capable of doing memory resettings and that from time to time you have to do such memory resstings because of glitches. Then, you know that when faced with some other type of problem or disaster, you could do a memory reseting as well in a attempt to escape that problem. This means that if you program yourself to never do that, then you can be sure that a memory resetting in case of a glitch is safe, in the sense that you are sure that after the resetting you won't find yourself having to face some kind of disaster.
view post as summary
post approved
FQXi Administrator Zeeya Merali wrote on May. 5, 2009 @ 18:33 GMT
Thanks for the explanation Saibal. It might take me a while to digest it, but then I'll get back to you if I have more questions.
post approved
paul valletta wrote on Jul. 12, 2009 @ 23:09 GMT
Hold on, can we change the paet by forgetting, or negative recall?..well there is nothing in a far off past, the past is only local, on very very short timescales?..I have to remember a FUTURE appointment, say next tuesday at 10:45am, when this moment arrives, if I forget my appointment I have actually altered my future, by not remembering my appointment?
Forgetting a little of your past is fundemental to relativity, a total recall would mean you could not distinguish a past event from a present event?..forgeting pasts actually alters future's !
post approved
Add a New Post