Simulation with non-local behavior

[FrediFizzx]

Looks like it might be possible to predict individual outcomes event by event with grossly non-local behavior. It is a non-local hidden variable model with 720 degrees of data at 1 degree resolution.



Here is a PDF of the Mathematica simulation,

EPRsims/non-local.pdf

The only problem with this is that output A is a coin toss variable. So you have to know a, b, lambda and A to predict the correct outcome for B. Don't know if this actually relates to quantum mechanics in any way.
.

[FrediFizzx]

Well..., we can just do A the regular way implementing the polarizer function where e is the A particle vector. So if you know a, b, e and lambda, you can predict the A and B outcomes. Here is the updated code.

EPRsims/non-local2.pdf
.

[Joy Christian]

***
This is of course not surprising. In 1964 Bell already noted that nonlocal models can reproduce the strong correlation:


***

[FrediFizzx]

***
This is of course not surprising. In 1964 Bell already noted that nonlocal models can reproduce the strong correlation:


***

Not quite the same thing. We can predict the individual A and B outcomes event by event to produce -a.b. Plus this is a HV non-local model. I mainly did this simulation to see if there might be a clue for local behavior.
.

[Joy Christian]

***
This is of course not surprising. In 1964 Bell already noted that nonlocal models can reproduce the strong correlation:


***

Not quite the same thing. We can predict the individual A and B outcomes event by event to produce -a.b. Plus this is a HV non-local model. I mainly did this simulation to see if there might be a clue for local behavior.

It is the same thing. Bell's local model of 1964 he is referring to in the above paragraph predicts individual A and B outcomes event-by-event. About ten years ago someone simulated its nonlocal version for me to convince me that only nonlocal models can reproduce the -a.b correlations, and rather easily. I continued working on my 3-sphere model regardless.

***

[FrediFizzx]

… It is the same thing. Bell's local model of 1964 he is referring to in the above paragraph predicts individual A and B outcomes event-by-event. About ten years ago someone simulated its nonlocal version for me to convince me that only nonlocal models can reproduce the -a.b correlations, and rather easily. I continued working on my 3-sphere model regardless.

***

It looks like to me that with Bell's configuration, you will get straight lines event by event instead of the negative cosine curve. Bell's HV is just the particle spin vector. You need this function,



with lambda random 0 to 1 to produce the negative cosine curve. But it is complete nonsense anyways. Not only does station B know A's angle, it also knows A's outcome. Very un-Natural.
.

[Guest]

It looks like to me that with Bell's configuration, you will get straight lines event by event instead of the negative cosine curve. Bell's HV is just the particle spin vector. You need this function,

with lambda random 0 to 1 to produce the negative cosine curve. But it is complete nonsense anyway. Not only does station B know A's angle, but it also knows A's outcome. Very un-Natural.

I like your model, Fred. It's elegant, one of the most elegant of this type that I have seen.

Some other people with models like this, but much more complex, are Anthony Crofts (Illinois) and David Oaknin (Haifa).
http://www.life.illinois.edu/crofts/
http://www.life.illinois.edu/crofts/Bell_Ineq/
https://www.researchgate.net/profile/David_Oaknin
https://arxiv.org/abs/1411.5704

[FrediFizzx]

It looks like to me that with Bell's configuration, you will get straight lines event by event instead of the negative cosine curve. Bell's HV is just the particle spin vector. You need this function,

with lambda random 0 to 1 to produce the negative cosine curve. But it is complete nonsense anyway. Not only does station B know A's angle, but it also knows A's outcome. Very un-Natural.

I like your model, Fred. It's elegant, one of the most elegant of this type that I have seen.

Some other people with models like this, but much more complex, are Anthony Crofts (Illinois) and David Oaknin (Haifa).
http://www.life.illinois.edu/crofts/
http://www.life.illinois.edu/crofts/Bell_Ineq/
https://www.researchgate.net/profile/David_Oaknin
https://arxiv.org/abs/1411.5704

Thanks. It is probably the most simple non-local HV model that can produce the negative cosine curve event by event and completely predictable if you know all the 4 variables. The simple function is adapted from Vongehr's QRC quantum model. I had to divide (b - a) by 2 to get it to work properly. You can see from the plot that the tails of the curve didn't quite come in all the way. I suspect that is just due to there not being enough data for complete averaging as you can see from what I did at the end of the code. But this gives me some ideas to try for a local HV model.

Oaknin's model is ridiculously complex and it looks like the HV's depend somewhat on (b - a). I will check out what Crofts did.
.

[minkwe]

Remember that QM must obey no-signalling. So models like these must also pass no-signalling to be valid. Do they?

[FrediFizzx]

Remember that QM must obey no-signalling. So models like these must also pass no-signalling to be valid. Do they?

Probably not. It is pretty hokey anyways since only station B knows both A's angle and A's outcome. I mean, how could it know A's outcome with no-signaling? And why only B?
.

[minkwe]

Remember that QM must obey no-signalling. So models like these must also pass no-signalling to be valid. Do they?

Probably not. It is pretty hokey anyways since only station B knows both A's angle and A's outcome. I mean, how could it know A's outcome with no-signaling? And why only B?
.

It is possible to convert it into a variant that does not require knowledge of outcomes but just knowledge of settings. But that does not take away the difficulty of no-signalling which is not permitted by QM. Too often people present such non-local models thinking it reflects what is happening in QM. But that couldn't be further from the truth. For a while the Bell proponents have been asking for local simulations that match QM but nobody has asked them to produce non-local simulations that match QM and all the required conditions like no-signalling. I think looking at what they come up with and the difficulties they face will be instructive.

I'm in the process of coming up with a straight-forward full-proof test of no-signalling. Stay tuned.

[gill1109]

For a while the Bell proponents have been asking for local simulations that match QM but nobody has asked them to produce non-local simulations that match QM and all the required conditions like no-signalling. I think looking at what they come up with and the difficulties they face will be instructive.
I'm in the process of coming up with a straight-forward full-proof test of no-signalling.stay tuned

Exciting!

Want to see a *non-local* simulation that matches QM and all required conditions? It goes like this

Code: Select all

Repeat N times
     Pick two settings a, b however you like
     Pick two outcomes x, y from the joint probability distribution  p(x, y | a, b) according to QM's EPR-B model.


[minkwe]

For a while the Bell proponents have been asking for local simulations that match QM but nobody has asked them to produce non-local simulations that match QM and all the required conditions like no-signalling. I think looking at what they come up with and the difficulties they face will be instructive.
I'm in the process of coming up with a straight-forward full-proof test of no-signalling.stay tuned

Exciting!

Want to see a *non-local* simulation that matches QM and all required conditions? It goes like this

Code: Select all

Repeat N times
     Pick two settings a, b however you like
     Pick two outcomes x, y from the joint probability distribution  p(x, y | a, b) according to QM's EPR-B model.


Sorry, but in case you did not know, QM does not permit signalling like that. You can't use (a,b) together at any station to obtain outcomes. Otherwise, Alice and Bob can easily communicate using entanglement, which is forbidden. Go back to the drawing board.

[Heinera]

For a while the Bell proponents have been asking for local simulations that match QM but nobody has asked them to produce non-local simulations that match QM and all the required conditions like no-signalling. I think looking at what they come up with and the difficulties they face will be instructive.
I'm in the process of coming up with a straight-forward full-proof test of no-signalling.stay tuned

Exciting!

Want to see a *non-local* simulation that matches QM and all required conditions? It goes like this

Code: Select all

Repeat N times
     Pick two settings a, b however you like
     Pick two outcomes x, y from the joint probability distribution  p(x, y | a, b) according to QM's EPR-B model.


Sorry, but in case you did not know, QM does not permit signalling like that. You can't use (a,b) together at any station to obtain outcomes. Otherwise, Alice and Bob can easily communicate using entanglement, which is forbidden. Go back to the drawing board.

Nope. Even with Richard's model, there is no way that Alice can signal to Bob (or vice versa) by using detector settings. No matter what setting Alice chooses, the probability distribution on Bob's end will still be 50/50. And of course if you had known even a whiff about quantum mechanics you would know that entanglement requires a distribution that depends on both a and b, but still is consistent with the no-signalling theorem.

For this model implemented in R see https://rpubs.com/heinera/16727

[minkwe]

Nope. Even with Richard's model, there is no way that Alice can signal to Bob (or vice versa) by using detector settings. No matter what setting Alice chooses, the probability distribution on Bob's end will still be 50/50. And of course if you had known even a whiff about quantum mechanics you would know that entanglement requires a distribution that depends on both a and b, but still is consistent with the no-signalling theorem.

For this model implemented in R see https://rpubs.com/heinera/16727

Of course you don't understand what it means to send information. It means the mutual information between Alice's setting and Bob's outcome and vice versa is significantly non-zero. Which means Bob can learn something about Alice's settings just by looking at his outcomes and vice versa. Which is obviously the case for Richard's model and yours.

[Heinera]

Which means Bob can learn something about Alice's settings just by looking at his outcomes and vice versa. Which is obviously the case for Richard's model and yours.

And exactly how can Bob learn about Alice's setting just by looking at his outcomes? Can you show us a protocol Bob can follow to gain this information?

[minkwe]

I don't have to. You should learn some information theory. All I have to show is that the mutual information between Bob's settings and Alice's outcomes is significantly non-zero. This means information has been transmitted. The mutual information exactly quantifies the amount of information transmitted.

[Heinera]

I don't have to. You should learn some information theory. All I have to show is that the mutual information between Bob's settings and Alice's outcomes is significantly non-zero. This means information has been transmitted. The mutual information exactly quantifies the amount of information transmitted.

Dodging the question again? You wrote "Which means Bob can learn something about Alice's settings just by looking at his outcomes and vice versa. Which is obviously the case for Richard's model and yours." So, what "obvious" analysis does Bob have to do to learn something about Alice's settings just by looking at his outcomes?

I think that it is you who need to learn some information theory. There is no information about Alice's settings in Bob's results, and vice versa.

[minkwe]

Please, do you know what mutual information means? Do you understand what it means for the mutual information to be significantly non-zero?

Don't speak about things you don't understand. It is embarrassing.

[Heinera]

Please, do you know what mutual information means? Do you understand what it means for the mutual information to be significantly non-zero?

Do you understand anything about information theory at all, except what you've learned from pop-sci books? Since you are clearly incapable of answering the question of how Bob can extract information about Alice's settings "just by looking at his outcomes," this discussion has come to an end.