Another simulation for QM Local functions

Foundations of physics and/or philosophy of physics, and in particular, posts on unresolved or controversial issues

Re: Another simulation for QM Local functions

Postby jreed » Sun Aug 04, 2019 1:16 pm

FrediFizzx wrote:Upon advice from counsel, I've decided to release the Mathematica code for the Complete States Selection section. Apparently you can't sufficiently protect an idea that involves the laws of nature. So here it is,

EPRsims/QMlocal_CS_no0s3D.pdf

At least you saw the idea from the code here first. :D This version has the A and B measurement functions simplified as in the previous post.
.


Here's what I found when I checked your Complete States Selection code:

This is what's in your notebook (slightly rearranged for display):

λ4=(2/√(1+3 (RandomReal[{0,π},samples])⁄π)-1);

and here's Pearle's equation for the detection loophole from his paper, according to Gill:

λ3=(2/√(1+3RandomReal[{0,1},samples])-1);

As you can see these are identical except for the factor of pi that is the limit of the random selection in your algorithm, and a division to make the random selection equal to a limit of 1. You have rediscovered Pearle's result for the detection loophole.
jreed
 
Posts: 176
Joined: Mon Feb 17, 2014 5:10 pm

Re: Another simulation for QM Local functions

Postby FrediFizzx » Sun Aug 04, 2019 1:23 pm

jreed wrote:
FrediFizzx wrote:Upon advice from counsel, I've decided to release the Mathematica code for the Complete States Selection section. Apparently you can't sufficiently protect an idea that involves the laws of nature. So here it is,

EPRsims/QMlocal_CS_no0s3D.pdf

At least you saw the idea from the code here first. :D This version has the A and B measurement functions simplified as in the previous post.
.


Here's what I found when I checked your Complete States Selection code:

This is what's in your notebook (slightly rearranged for display):

λ4=(2/√(1+3 (RandomReal[{0,π},samples])⁄π)-1);

and here's Pearle's equation for the detection loophole from his paper, according to Gill:

λ3=(2/√(1+3RandomReal[{0,1},samples])-1);

As you can see these are identical except for the factor of pi that is the limit of the random selection in your algorithm, and a division to make the random selection equal to a limit of 1. You have rediscovered Pearle's result for the detection loophole.

:D There is no rediscovery going on here. It is intentionally Pearle's formula because Joy found out it is a result of 3-sphere topology.

Perhaps you should be asking questions instead of making statements?
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Sun Aug 04, 2019 4:57 pm

FrediFizzx wrote:Nope. It doesn't work. Nature is not going to conform and certainly doesn't have to conform to man-made rules.
.

I take it back again. :D The states that don't exist in S^3-land should be able to be eliminated in the data analysis stage. But why go to all that trouble? Better to just get rid of them from the start.

But in theory, what Albert Jan is suggesting should work. If someone want to program it up, go for it.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Sun Aug 04, 2019 9:45 pm

FrediFizzx wrote:
FrediFizzx wrote:Nope. It doesn't work. Nature is not going to conform and certainly doesn't have to conform to man-made rules.
.

I take it back again. :D The states that don't exist in S^3-land should be able to be eliminated in the data analysis stage. But why go to all that trouble? Better to just get rid of them from the start.

But in theory, what Albert Jan is suggesting should work. If someone want to program it up, go for it.
.

A thanks and tip o' the hat goes to Albert Jan. His idea does work. The states that don't exist in S^3-land can be eliminated in the data analysis stage.

Image

2 million trials one degree resolution. The code is a mess; I will post it tomorrow morning after cleaning it up.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Mon Aug 05, 2019 12:36 pm

Here is the Mathematica code for the new Complete States simulation. Big thanks again to Albert Jan for telling us how it should really work.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf

This could be run on three separate computers. The A and B computers just need a,s, lambda and b,s, lambda respectively. The CA and CB tables could be exported as .csv files and imported into the 3rd computer for data analysis. This is how Nature works via S^3 topology.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby ajw » Tue Aug 06, 2019 2:55 am

FrediFizzx wrote:Here is the Mathematica code for the new Complete States simulation. Big thanks again to Albert Jan for telling us how it should really work.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf

This could be run on three separate computers. The A and B computers just need a,s, lambda and b,s, lambda respectively. The CA and CB tables could be exported as .csv files and imported into the 3rd computer for data analysis. This is how Nature works via S^3 topology.
.

Fred, It seems to me that the random angle picking on each separate computer for Alice and Bob isn't there yet, or am I overlooking something? This point is important if you want to prove the simulation works while maintaining locality. The current model seems to use the index to pick the angles for Alice and Bob when running on the separate computers, so it works as if these filter angles where created when producing the entangled particle pair.
As mentioned earlier one is allowed to create states of angles for Alice and Bob and pass these lists as hidden variables to each separate computer. But the separate computers have to pick the random angles for each run (and then perform the measurement) to mimic the experimental situation.
ajw
 
Posts: 45
Joined: Sat Sep 05, 2015 2:04 pm

Re: Another simulation for QM Local functions

Postby FrediFizzx » Tue Aug 06, 2019 8:00 am

ajw wrote:
FrediFizzx wrote:Here is the Mathematica code for the new Complete States simulation. Big thanks again to Albert Jan for telling us how it should really work.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf

This could be run on three separate computers. The A and B computers just need a,s, lambda and b,s, lambda respectively. The CA and CB tables could be exported as .csv files and imported into the 3rd computer for data analysis. This is how Nature works via S^3 topology.
.

Fred, It seems to me that the random angle picking on each separate computer for Alice and Bob isn't there yet, or am I overlooking something? This point is important if you want to prove the simulation works while maintaining locality. The current model seems to use the index to pick the angles for Alice and Bob when running on the separate computers, so it works as if these filter angles where created when producing the entangled particle pair.
As mentioned earlier one is allowed to create states of angles for Alice and Bob and pass these lists as hidden variables to each separate computer. But the separate computers have to pick the random angles for each run (and then perform the measurement) to mimic the experimental situation.

I guess you are talking about this,

aliceDeg[[j]] = ArcTan[x1, y1];

And the same for Bob. That is part of determining the ArcCos selection later in the analysis. But yeah, that can and should be moved to the data analysis section. Thanks.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby gill1109 » Tue Aug 06, 2019 8:41 am

External parties need to be able to pick angles. Any angles you like. There should be three computer programs: one for the source, one for Alice's detector, one for Bob's detector. But anybody else should be able to supply angles (or directions, or names of settings, or whatever you like) for each of the detector computers, separately. There must be no constaints whatever on the content of those lists of settings, beyond the requirements that they are "legal settings which will definitely be accepted, and result in a measurement outcome".

For instance, we could think of settings as being angles in the plane, specified to a whole number of degrees. So the input lists are lists of whole numbers between, say, 0 and 359. The length N of the two lists is agreed on in advance. The settings in the left and right wings are supplied one at a time, and two binary outputs collected. This is repeated N times.

One of the recent "avoidable-loophole free experiments" had settings taken from the binary file of a maximally compressed copy of the movie "Back to the Future". N was huge and Bell's inequality was violated by a tiny amount, about 10 to the minus 5, but a tiny amount which was statistically overwhelmingly significantly different from zero - ie by a huge number of standard deviations. More precisely, they used the Eberhard inequality. If there is "no signalling", at least, statistically speaking, there is no significant signalling, then Eberhard and Bell-CHSH are very close to one another. But the formulas for the standard deviations of the two things are different. They combine some summary statistics of the experiment in different ways. I recently figured out the optimal test statistic. Yet another combination of the same basic summary statistics. The deviation of S from zero is insignificantly different, physically, but it increase the number of standard deviations away from zero by 50% or so. Thus tremendously improved the p-value. But who cares if the p-value is 10 to the minus 6 or 10 to the minus 9? It's better than the particle physicists threshold of 5 standard deviations.

If we want to simulate a Bell experiment free of avoidable loopholes" then each list should contain just two different settings. It would be reasonable for the local realist who writes the three programs to insist that each of the four setting combinations are repeated a lot of times, e.g. at least N/8 times. If N is large and the two lists are created by tossing fair coins then there will be close to N/4 of each setting combination and the probability than any of the four combinations occurs less than N/8 times is astronomically small.
gill1109
Mathematical Statistician
 
Posts: 2812
Joined: Tue Feb 04, 2014 10:39 pm
Location: Leiden

Re: Another simulation for QM Local functions

Postby gill1109 » Tue Aug 06, 2019 8:48 am

jreed wrote:
FrediFizzx wrote:Upon advice from counsel, I've decided to release the Mathematica code for the Complete States Selection section. Apparently you can't sufficiently protect an idea that involves the laws of nature. So here it is,

EPRsims/QMlocal_CS_no0s3D.pdf

At least you saw the idea from the code here first. :D This version has the A and B measurement functions simplified as in the previous post.
.


Here's what I found when I checked your Complete States Selection code:

This is what's in your notebook (slightly rearranged for display):

λ4=(2/√(1+3 (RandomReal[{0,π},samples])⁄π)-1);

and here's Pearle's equation for the detection loophole from his paper, according to Gill:

λ3=(2/√(1+3RandomReal[{0,1},samples])-1);

As you can see these are identical except for the factor of pi that is the limit of the random selection in your algorithm, and a division to make the random selection equal to a limit of 1. You have rediscovered Pearle's result for the detection loophole.

They didn't rediscover it. I told it to them. They adopted it. I can prove it, if you like! Of course, they are clever enough to correct Pearle's mistakes themselves, but I was the first ever to document those mistakes. I discussed it with Pearle himself.

Look: I *gave* my mathematical corrections of Pearle's formulas to Joy and his friends, and I gave the first-ever computer implementation of the corrected Pearl's model, coded by me, to Joy and his friends, including those friends of his who knew how to program a computer. It was a present from me. I wanted them to have it and to use it. But I cannot accept anyone claiming that they discovered it before me. Maybe Joy didn't know that his computing friends had got it from me. I could understand that. I would like to hear some confirmation from Joy and his friends, or alternatively, I would like to hear an alternative scenario, together with supporting copies of emails etc, etc.

Honestly, I am not going to speak at a conference together with Joy while this little "plagiarism" matter is between us! At least, I cannot speak at a conference together with Joy, without raising the matter in public.

Of course, you can say that "Pearle's model" is just a version of Michel Fodjes, or Caroline Thompson's or whatever. Yes, the basic principles (rejection!) are the same. They all result in close approximations to the cosine curve. In an experiment you wouldn't know the difference. But Pearle's model is crafted so as to give *exactly* the cosine. The others do not. They were not intended to give *exactly* the cosine. And Pearle was *before* the others. The others did not know it. Actually, Caroline did - the point of her model is that there was a physical story behind it, of spinning balls, which showed that there were simple physical mechanisms which could do this, in reality.
gill1109
Mathematical Statistician
 
Posts: 2812
Joined: Tue Feb 04, 2014 10:39 pm
Location: Leiden

Re: Another simulation for QM Local functions

Postby Joy Christian » Tue Aug 06, 2019 9:18 am

***
Both Michel Fodje and I have expressed our views on this supposed "priority dispute" before on this forum, many years ago: viewtopic.php?f=6&t=111&p=4672#p4666

I have no further comments to add to what has already been noted in the above thread by both Michel Fodje and myself. For me, the matter ended at that time, years ago.

***
Joy Christian
Research Physicist
 
Posts: 2793
Joined: Wed Feb 05, 2014 4:49 am
Location: Oxford, United Kingdom

Re: Another simulation for QM Local functions

Postby ajw » Tue Aug 06, 2019 9:22 am

FrediFizzx wrote:
ajw wrote:
FrediFizzx wrote:Here is the Mathematica code for the new Complete States simulation. Big thanks again to Albert Jan for telling us how it should really work.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf

This could be run on three separate computers. The A and B computers just need a,s, lambda and b,s, lambda respectively. The CA and CB tables could be exported as .csv files and imported into the 3rd computer for data analysis. This is how Nature works via S^3 topology.
.

Fred, It seems to me that the random angle picking on each separate computer for Alice and Bob isn't there yet, or am I overlooking something? This point is important if you want to prove the simulation works while maintaining locality. The current model seems to use the index to pick the angles for Alice and Bob when running on the separate computers, so it works as if these filter angles where created when producing the entangled particle pair.
As mentioned earlier one is allowed to create states of angles for Alice and Bob and pass these lists as hidden variables to each separate computer. But the separate computers have to pick the random angles for each run (and then perform the measurement) to mimic the experimental situation.

I guess you are talking about this,

aliceDeg[[j]] = ArcTan[x1, y1];

And the same for Bob. That is part of determining the ArcCos selection later in the analysis. But yeah, that can and should be moved to the data analysis section. Thanks.
.


No, I am talking about the start of the loop that is supposed to run on 2 computers for a=aa[[j]] and b=bb[[j]]

Code: Select all
Do[a=aa[[j]]; b=bb[[j]] ...


By using the same value for j on each computer for each particle pair, and thereby choosing the angle from a lookup table, you in fact synchronize the angle choice for Alice and Bob. That is why I think you should pick a random angle for Alice and also for Bob within this loop.

But in my memory I have explored this type of code when first working with the R-version model from Joy. I think it is a variation of the flatlanders implementation of the model, and therefore perhaps can be used to simulate Joy's proposal, but cannot be used to prove that it is working while maintaining locality.
Last edited by ajw on Tue Aug 06, 2019 9:26 am, edited 1 time in total.
ajw
 
Posts: 45
Joined: Sat Sep 05, 2015 2:04 pm

Re: Another simulation for QM Local functions

Postby FrediFizzx » Tue Aug 06, 2019 9:23 am

Joy Christian wrote:***
Both Michel Fodje and I have expressed our views on this supposed "priority dispute" before on this forum, many years ago: viewtopic.php?f=6&t=111&p=4672#p4666

I have no further comments to add to what has already been noted in the above thread by both Michel Fodje and myself. For me, the matter ended at that time, years ago.

***

I'm declaring the controversy about the "priority" OFF-TOPIC for this thread. Any further mention of it will be deleted. It is not that important for the actual physics. Start a new topic if you want to hash that out.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Tue Aug 06, 2019 9:27 am

ajw wrote:No, I am talking about the start of the loop that is supposed to run on 2 computers for a=aa[[j]] and b=bb[[j]]

Code: Select all
Do[a=aa[[j]]; b=bb[[j]] ...


By using the same value for j on each computer for each particle pair, and thereby choosing the angle from a lookup table, you in fact synchronize the angle choice for Alice and Bob. That is why I think you should pick a random angle for Alice and also for Bob within this loop.

But in my memory I have explored this type of code when first working with the R-version model from Joy. I think it is a variation of the flatlanders implementation of the model, and therefore can be used to simulate Joy's proposal, but cannot be used to prove that it is working while maintaining locality.

??? Those don't exist in the new code. Perhaps you need to delete your browser cache.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby ajw » Tue Aug 06, 2019 9:59 am

FrediFizzx wrote:
ajw wrote:No, I am talking about the start of the loop that is supposed to run on 2 computers for a=aa[[j]] and b=bb[[j]]

Code: Select all
Do[a=aa[[j]]; b=bb[[j]] ...


By using the same value for j on each computer for each particle pair, and thereby choosing the angle from a lookup table, you in fact synchronize the angle choice for Alice and Bob. That is why I think you should pick a random angle for Alice and also for Bob within this loop.

But in my memory I have explored this type of code when first working with the R-version model from Joy. I think it is a variation of the flatlanders implementation of the model, and therefore can be used to simulate Joy's proposal, but cannot be used to prove that it is working while maintaining locality.

??? Those don't exist in the new code. Perhaps you need to delete your browser cache.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf
.

Ah, sorry. I have read the newest one earlier, but when I started responding I accidentally opened the earlier one.
But I think the argument stays the same. There is no random angle picking within the loop, potentially performed by 2 computers.
ajw
 
Posts: 45
Joined: Sat Sep 05, 2015 2:04 pm

Re: Another simulation for QM Local functions

Postby FrediFizzx » Tue Aug 06, 2019 10:17 am

ajw wrote:
FrediFizzx wrote:??? Those don't exist in the new code. Perhaps you need to delete your browser cache.

EPRsims/QMlocal_CS_no0s3Dsep2.pdf
.

Ah, sorry. I have read the newest one earlier, but when I started responding I accidentally opened the earlier one.
But I think the argument stays the same. There is no random angle picking within the loop, potentially performed by 2 computers.

Sure for the particular way that I did it, but it is easy to see that you could have 3 computers each independently generating a, b and s + lambda. Then s and lambda are sent to Alice's and Bob's computers. And then they do their polarizer functions and measurement functions independently. Alice produces the CA table and Bob produces the CB table which are then sent to the fourth computer for analysis. It all works.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Tue Aug 06, 2019 10:54 am

OK, so you are talking about doing 3 separate Do loops,

Code: Select all
Do[s = Normalize@RandomVariate[NormalDistribution[], 3];  (*3D uniform random unit vectors*)
 \[Lambda] = RandomChoice[{-1, 1}], {j, trialcs}]
Do[a = Normalize@RandomVariate[NormalDistribution[], 3];
 If[a.s > 0, s1 = a, s1 = -a];  (*Polarizer Function*)
 A = \[Lambda] (a.(-s1));  (*Measurement Function*)
 CA[[j]] = {a, A, s}, {j, trialcs}]
Do[b = Normalize@RandomVariate[NormalDistribution[], 3];
 If[b.s > 0, s2 = b, s2 = -b];
 B = \[Lambda] (b.s2);
 CB[[j]] = {b, B}, {j, trialcs}]


Good. Now, I just have to get s and lambda into the A and B loops.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Tue Aug 06, 2019 2:43 pm

OK folks, now we are talking about the primo simulation thanks again to Albert Jan's persistence. For particle data generation we have 3 independent Do Loops for three independent computers. Lambda and s are sent to Alice and Bob and then the CA and CB tables and s are sent to the data analysis station.

EPRsims/QMlocal_CS_no0s3Dsep3.pdf

Enjoy!
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Wed Aug 07, 2019 5:55 pm

FrediFizzx wrote:OK folks, now we are talking about the primo simulation thanks again to Albert Jan's persistence. For particle data generation we have 3 independent Do Loops for three independent computers. Lambda and s are sent to Alice and Bob and then the CA and CB tables and s are sent to the data analysis station.

EPRsims/QMlocal_CS_no0s3Dsep3.pdf

Enjoy!
.

This also works without the polarizer functions by using Sign functions on the measurement functions. This is the geometric algebra version for Joy's S^3 Complete States model. A bit simplified of course for the measurement functions.

EPRsims/GAlocal_CS_no0s3Dsep.pdf

Perhaps John Reed can try this one using quaternions?
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby FrediFizzx » Thu Aug 08, 2019 8:55 pm

FrediFizzx wrote:
FrediFizzx wrote:OK folks, now we are talking about the primo simulation thanks again to Albert Jan's persistence. For particle data generation we have 3 independent Do Loops for three independent computers. Lambda and s are sent to Alice and Bob and then the CA and CB tables and s are sent to the data analysis station.

EPRsims/QMlocal_CS_no0s3Dsep3.pdf

Enjoy!
.

This also works without the polarizer functions by using Sign functions on the measurement functions. This is the geometric algebra version for Joy's S^3 Complete States model. A bit simplified of course for the measurement functions.

EPRsims/GAlocal_CS_no0s3Dsep.pdf

Perhaps John Reed can try this one using quaternions?
.

Actually the 3D quaternion version is no big deal. All it does is flip the signs on A and B measurement functions at the same time. Did it. If anyone is interested I will post the Mathematica code.
.
FrediFizzx
Independent Physics Researcher
 
Posts: 2905
Joined: Tue Mar 19, 2013 7:12 pm
Location: N. California, USA

Re: Another simulation for QM Local functions

Postby jreed » Fri Aug 09, 2019 6:16 am

FrediFizzx wrote:
FrediFizzx wrote:
FrediFizzx wrote:OK folks, now we are talking about the primo simulation thanks again to Albert Jan's persistence. For particle data generation we have 3 independent Do Loops for three independent computers. Lambda and s are sent to Alice and Bob and then the CA and CB tables and s are sent to the data analysis station.

EPRsims/QMlocal_CS_no0s3Dsep3.pdf

Enjoy!
.

This also works without the polarizer functions by using Sign functions on the measurement functions. This is the geometric algebra version for Joy's S^3 Complete States model. A bit simplified of course for the measurement functions.

EPRsims/GAlocal_CS_no0s3Dsep.pdf

Perhaps John Reed can try this one using quaternions?
.

Actually the 3D quaternion version is no big deal. All it does is flip the signs on A and B measurement functions at the same time. Did it. If anyone is interested I will post the Mathematica code.
.


Please post the code. I want to take a look at it.
jreed
 
Posts: 176
Joined: Mon Feb 17, 2014 5:10 pm

PreviousNext

Return to Sci.Physics.Foundations

Who is online

Users browsing this forum: No registered users and 69 guests

cron
CodeCogs - An Open Source Scientific Library