FTL Comms: Sending messages back in time.

[Cheetah]

I know next to nothing about quantum physics, but just having read “The Fabric of the Cosmos” by Brian Greene feel that I now have a good layman’s understanding of the principles. While pondering the weird and wonderful quantum world I came up with the following setup:



A photon emitter (PE) sends a beam through a beam splitter (BS1) and then towards 2 down converters (DC1&2). The signal photons travel on to create an interference pattern via SPP1&2 (Signal Photon Path) while the idler photons end up via IPP1&2 (Idler Photon Path) at one of two detectors (DT1&2). Since BS2 destroys any indirect which-path information of the signal photons (and idler photons) an interference pattern should form.
Removing BS2 will yield indirect which-path information of the signal photons and destroy the interference pattern. So the interference pattern can be “switched” on or off by removing/inserting BS2.

Now lengthen IPP1&2 considerably so that the idler photons take 1 hour, for argument sake, before reaching BS2. Since SPP1&2 is very short an interference pattern will form, or not, depending on whether BS2 is there or not, one hour from now when the idler photons reach the detectors.

So you can send a sort of Morse code an hour into the past by inserting/removing BS2. I believe this to be impossible as it would violate cause and effect. Where did I go wrong?

 

[blobru]

Hi, Cheetah.

Looks like a version of a quantum delayed-choice eraser you've got there.

So it's my understanding that the signal photons' interference patterns (bottom right-hand corner of your diagram) are only apparent after you have information about what happened to the idler photons (did they pass through BS2 or not? did they pass into DT1 or DT2?) That is, the interference pattern is deduced from information gathered at DT1 & DT2, which then allows you to isolate signal photons whose idlers have passed through BS2 into either DT1 or DT2. That's the only interference pattern you'll see, indirectly. Taken together, the signal photons won't show an obvious interference pattern. The interference patterns show up for the signal photons whose idlers passed through BS2 into DT1, and the signal photons whose idlers passed through BS2 into DT2. And we won't have that information, to isolate the corresponding [entangled] signal photons and deduce those interference patterns, until the idler photons have passed through BS2 into either DT1 or DT2. So we haven't necessarily caused an event in the past (though some might interpret it that way); however, we have affected how it is deciphered (which subset of the total set of signal photons a particular signal photon belongs to).

Warning: that's my convoluted understanding only! I'm not a physicist; there are several who post here who can explain it much better, and correct the mistakes I'm bound to have made.

 

[shadron]

So you can send a sort of Morse code an hour into the past by inserting/removing BS2. I believe this to be impossible as it would violate cause and effect. Where did I go wrong?

Cheetah, you may not be wrong. See this article: http://everything2.com/title/thiotimoline

 

[INRM]

blobru,

Why would some interpret it as an event occuring in the past and why would others not?

 

[blobru]

blobru,

Why would some interpret it as an event occuring in the past and why would others not?

Well, again in my limited understanding, detection of an interference pattern indicates the quantum is behaving like a wave; a smooth distribution of hits (lack of interference), like a particle. In QM, the behavior of the signal particle is entangled with the idler particle's behavior -- whether it passes through BS2 and interferes with itself (wave-like) or not (particle-like) -- an hour later (in Cheetah's set-up). Some interpret this as the idler particle retroactively causing the signal particle to behave like a particle or a wave (in the past); others as the idler particle causing us to presently categorize the signal as "particle" (non-interfering) or "wave" (interfering): only the "waves" to be counted in the calculation of the interference pattern.

 

[Cheetah]

Thanks for your informative reply blobru (and shadron, still puzzling on that one ). I had a look at the delayed choice quantum eraser on Wikipedia. Quite confusing and their setup is slightly more complicated, but I think it is equivalent to the following:



As I understand it if the idler is detected at DT3 or 4 the corresponding signal photons will not produce an interference pattern, but if detected at DT1 or 2 it will. The interference pattern will not be visible as it will be swamped by the non-interference pattern and can only be extracted by correlating the idlers detected at DT1&2 with their corresponding entangled signal partners. You therefore need the results of the 4 detectors to extract the pattern.
That should not be necessary in my proposed setup.
I'm sure my understanding is too simplistic and I must be missing something.

 

[blobru]

Thanks for your informative reply blobru (and shadron, still puzzling on that one ). I had a look at the delayed choice quantum eraser on Wikipedia. Quite confusing and their setup is slightly more complicated, but I think it is equivalent to the following:

View attachment 16824

Nice diagram, Cheetah. I agree it seems logically equivalent (minus the prisms and coincidence counter) to wiki's Delayed choice quantum eraser^WP (Kim, etal).

As I understand it if the idler is detected at DT3 or 4 the corresponding signal photons will not produce an interference pattern, but if detected at DT1 or 2 it will. The interference pattern will not be visible as it will be swamped by the non-interference pattern and can only be extracted by correlating the idlers detected at DT1&2 with their corresponding entangled signal partners. You therefore need the results of the 4 detectors to extract the pattern.

That's my understanding as well.

That should not be necessary in my proposed setup.
I'm sure my understanding is too simplistic and I must be missing something.

In your setup, when BS2 is removed, DT1 & DT2 function as DT3 & DT4 in the wiki setup. The idler photons that don't pass through BS2 will tend to swamp the interference pattern of those which do. So you'll need to identify these and ignore them to extract the interference pattern. But until they have definitely not passed through BS2 on their way to DT1 & DT2, you won't know which to ignore. (The role of DT3 and DT4 in the wiki setup is to identify which signal photons to ignore).

I think that's how it works. (Still, some interesting implications: will you be able to affect the shape of the interference pattern by selectively filtering signal photons? hmm...)


ETA: (correction to post #5)

Well, again in my limited understanding, detection of an interference pattern indicates the quantum is behaving like a wave; a smooth distribution of hits (lack of interference), like a particle. In QM, the behavior of the signal photon is entangled with the idler photon's behavior -- whether it passes through BS2 and interferes with itself (wave-like) or not (particle-like) -- an hour later (in Cheetah's set-up). Some interpret this as the idler photon retroactively causing the signal photon to behave like a particle or a wave (in the past); others as the idler photon 'causing' us to presently categorize the signal as "particle" (non-interfering) or "wave" (interfering): only the "waves" to be counted in the calculation of the interference pattern.

 

[Cheetah]

I don't think you have to filter anything, in the DCQE the two patterns form simultaneously, but consecutively in my setup. It just depends how long it takes for a recognizable pattern to form (should be all but instantaneous?). It also does not depend on where the idlers are detected, just whether BS2 is present or not.

Any experts around? This is really bugging me.

 

[blobru]

I don't think you have to filter anything, in the DCQE the two patterns form simultaneously, but consecutively in my setup. It just depends how long it takes for a recognizable pattern to form (should be all but instantaneous?). It also does not depend on where the idlers are detected, just whether BS2 is present or not.

Any experts around? This is really bugging me.

Not "simultaneously". In wiki's DCQE (see diagram), there's a gap of at least 8 nano-seconds (experiment pdf here) between the signal photon reaching D₀ (where the interference pattern forms) and the idler reaching one of the other detectors. In your setup, the gap is 1 hour. Same principle, though.

(I think.) Hopefully, an expert will drop by to sort it out for us laymen.

 

[Cheetah]

Darnit I wasn't clear .
I did not mean between the idler and signal photons reaching their targets, I am only talking about the signal photons.
In the DCQE some idlers are detected at DT1&2 and some at DT3&4 and which ones end up where is random, I suppose a 25% chance of ending up at any given detector. The corresponding signal photons trace out an interference pattern, or not, depending on where the idlers were detected. The two patterns obscure each other since each signal photon could be interfering with itself or not and you will never know until you find out where the corresponding idler was detected.
In my setup run the experiment with BS2 in place for 1 min. All the signal photons will trace out an interference pattern. Then remove BS2 and run for 1 min again, non of the signal photons will trace out an interference pattern, no matter where the idlers were detected. The two patterns cannot obscure each other as they are formed consecutively, not concurrently as in the DCQE.

Ha I'm building myself one this evening in the basement and sending myself the lottery numbers . Anyone want in?

 

[JoeyDonuts]

The new Mass Effect game gets around this by "Quantum Bit Entanglement" secure communications. The idea is that two bits are entangled at the quantum level, and will change states respectively no matter how much distance is involved.

I like to call it the McGuffin Microphone.

 

[blobru]

... Ha I'm building myself one this evening in the basement and sending myself the lottery numbers . Anyone want in?

Wow. Either you've got one awfully big basement, or a whole lotta mirrors. Might need to win a few lotteries just to pay for it all (of course, that won't be a problem).

Umm... 4-17-19-32-35-40, bonus number 27, if you wouldn't mind sidling a little quantum luck your thread buddy's way?

 

[Cheetah]

No problem, any idea which is the richest lottery, gotta get in quickly before the rest catch on.

 

[blobru]

Win Powerball^WP!!! (thanks, dglas)

Largest single ticket payout ever: $177,270,519.67 (before taxes -- exactly 4 years ago today)!

That kinda dough could buy a fella a lotta mirrors, and a lot bigger basement... just sayin'.

 

[Cheetah]

Hmmmmm poooowerbaaalll........

That's 1,360,881,689 ZAR! Wow!

 

[Cheetah]

Is there no one here that can shed some, ahem......photons on this issue? Surely there must be someone with a better understanding of QM on this forum. I know believe it's impossible to send messages back in time but cannot find fault with my layman's understanding and logic. Help me out here, please.

 

[Fredrik]

Cheetah, I think you made a small mistake (which is unrelated to the main issue here) when you designed this experiment. It seems to me that the pattern on the photographic plate in the lower right will be the same with or without BS2 (and with or without BS1). This isn't like a double slit experiments where a photon taking the left path is more likely to end up on the left side of the screen. Both beams on the lower right should contribute in exactly the same way to the total pattern. There's no interference at all.

So I think you have to replace the photographic plate with the same setup you have in the upper left, with a beam splitter and two detectors, and argue that the path lengths on the lower right can be adjusted so that only one of the detectors on the lower right will ever click, as long as the beam splitter on the upper left is in place. Now you seem to be able to make that detector click (with 50% probability) by removing the beam splitter on the upper left after it has already clicked. (It's definitely not possible that this is correct, but your question is about finding out what's wrong with an argument of this type).

With this modification, your question becomes identical to one I asked at Physics Forums 5 years ago. I don't think anyone who participated in the discussion then (including me) really understood the complete answer. I still don't, but I think the bottom line is that you won't ever get any interference of the type I just described in the lower right (interference that's supposed to ensure that one of the detectors on the lower right never clicks). I think that what this means is that the fact that "which way information" is available somewhere along the paths to the upper left is sufficient to destroy the interference on the lower right. The beam splitter that recombines the beams is making that information inaccessible, but that's not enough to restore the interference on the lower right. (I'm not aware of any other argument for this other than "if the recombination restores the interference, then we're sending messages to the past").

So the detectors on the lower right will both click 50% of the time, regardless of path lengths. This seems to make the experiment kind of dull, but we can make it interesting again by removing the beam splitter that I said you should insert on the lower right (but keep the detectors). Every time we run this experiment without BS2, the detector clicks on the upper left are correlated with the ones on the lower right. If you know which detector clicked on the upper left, you know which one clicked on the lower right. But if we run the experiment with BS2 in place, the detector clicks won't be correlated. (Each detector still clicks 50% of the time though. They do this regardless of whether BS2 is in place).

Now consider the following variant of the experiment: Every time a photon leaves the emitter, BS2 is not in place. It's put in place 59 minutes later. The insertion of BS2, made 59 minutes after one of the detectors clicked in the lower right, is sufficient to destroy the correlations between the detectors.

This looks very strange if we have a mental picture of photons as localized entities "flying" from the emitter to the detectors, splitting in two at BS1 and recombining two halves at BS2. It seems that the insertion of BS2 has prevented the split at BS1 which was made 59 minutes earlier. So in this mental picture of what's going on, there's still an element of retrocausation.

I would advice against using mental pictures like this one. The moment we start to think of QM as a description of what actually happens, then we're already deep into many-worlds territory. If you're not going to dig deeply into the subject of realistic interpretations of QM (and I suggest that you don't, because the literature is terrible), than you should think of QM as just a set of rules that tells us how to calculate probabilities of possibilities, and...I'll just let Rudolf Haag say it (from "Local quantum physics", page 2):

Take the example of a position measurement on an electron. It woud lead to a host of paradoxa if one wanted to assume that the electron has some position at a given time. "Position" is just not an attribute of an electron, it is an attribute of the "event" i.e. of the interaction process between the electron and an appropriately chosen measuring instrument (for instance a screen), not of the electron alone. The uncertainty about the position of the electron prior to the measurement is not due to our subjective ignorance. It arises from improperly attributing the concept of position to the electron instead of reserving it for the event.​