Coming Soon!

[FrediFizzx]

.
Congratulations, Fred. Someone is converging to eating their hat in public, as they boasted some years ago!
.

Thanks. This is just the beginning of even more refinement.
.

[FrediFizzx]

Ok folks, as promised here is the super-duper ultra local realistic model that Gill said was impossible. It is time for Gill to post a video somewhere showing him eating his hat. One million trials at one degree resolution and I went to a more simple model to help speed it up. Took one day of processing mainly because of the trial number matching functions. If you have one million events then they have to do one trillion calculations more than one time. Three separate Do loops. A and B are run at completely different times.



Here are the Mathematica PDF and notebook files.

EPRsims/newCS-10-forum.pdf
EPRsims/newCS-10-forum.nb

Enjoy!!!! There is absolutely no doubt now that Bell is dead, dead, dead and shot down for good.
.

[gill1109]

Ok folks, as promised here is the super-duper ultra local realistic model that Gill said was impossible. It is time for Gill to post a video somewhere showing him eating his hat. One million trials at one degree resolution and I went to a more simple model to help speed it up. Took one day of processing mainly because of the trial number matching functions. If you have one million events then they have to do one trillion calculations more than one time. Three separate Do loops. A and B are run at completely different times.



Here are the Mathematica PDF and notebook files.

EPRsims/newCS-10-forum.pdf
EPRsims/newCS-10-forum.nb

Enjoy!!!! There is absolutely no doubt now that Bell is dead, dead, dead and shot down for good.
.

I am looking forward to seeing this simulation performed on two separate computers, each receiving settings from external parties, data gathered for analysis afterwards. That is what various mathematical theorems say is impossible. If someone does it, I promise I will eat my hat in public. I also offer a prize of 65 000 Euro. Rules: outcomes +/-1; sample size at least, say, 10 000, for a Bell-CHSH type experiment; "S" must exceed 2.4. I have no objection to simulating the two separate computers on one, because it is easy to perform some tests to check that there is no cheating, as long as the program has "set-seed" and "restore-seed" facilities and allows the user to freely specify sample size and streams of settings.

Please supply a version of your program in Python or R. The program will take a lot less time to run if you only allow two settings for Alice and two for Bob, and only do ten thousand trials, not a million.

[FrediFizzx]

@gill1109
You have the Mathematica code so knock your lights out and do whatever you want. I'm not interested in doing anything else since I know what it is exactly and it works. Well..., except write up a paper about it and get it published. I was thinking of the title, "Bell's 'theorem' was always just a theory and now it is nothing".

Here is the CHSH version.

CHSH = 2.64946

Here are the files.

EPRsims/newCS-10-CHSH-forum.pdf
EPRsims/newCS-10-CHSH-forum.nb

Enjoy!!!
.

[Joy Christian]

@gill1109
You have the Mathematica code so knock your lights out and do whatever you want. I'm not interested in doing anything else since I know what it is exactly and it works. Well..., except write up a paper about it and get it published. I was thinking of the title, "Bell's 'theorem' was always just a theory and now it is nothing".

Here is the CHSH version.

CHSH = 2.64946

Here are the files.

EPRsims/newCS-10-CHSH-forum.pdf
EPRsims/newCS-10-CHSH-forum.nb

Enjoy!!!

Not surprisingly, the goalpost-shifting by Gill has begun, almost as a reflex action. This is how a believer in a belief system always reacts to any threat to his or her beliefs.
.

[jreed]

I just downloaded your latest version in Mathematica, and tried setting lambda = 0 in the do loop over angles. I got a CHSH result of 1.993. This indicates that lambda is that old detection loophole, and if it is zero, you will get the usual classic result.

[FrediFizzx]

I just downloaded your latest version in Mathematica, and tried setting lambda = 0 in the do loop over angles. I got a CHSH result of 1.993. This indicates that lambda is that old detection loophole, and if it is zero, you will get the usual classic result.

10,000 particles in and 10,000 detected. Check that. Of course we don't want the HV to ever be zero. No point to that.
.

[jreed]

I just downloaded your latest version in Mathematica, and tried setting lambda = 0 in the do loop over angles. I got a CHSH result of 1.993. This indicates that lambda is that old detection loophole, and if it is zero, you will get the usual classic result.

10,000 particles in and 10,000 detected. Check that. Of course we don't want the HV to ever be zero. No point to that.
.

You are using the same logic that I pointed out in an earlier post:

"I think I have the answer. I generated the classic triangle function using a Mathematica routine I wrote. Then I added this to a cosine function in the proportions I found for the detection loophole part (0.835) and the triangle part (0.165) that I got when I generated that cosine-like curve using your program. When this result is plotted on top of the cosine, it is difficult to find a difference between these curves. Those straight lines are in there, but in the proportions found by the program, can't be distinguished. What you have here is the detection loophole, but the missing events are not discarded, but hidden in the final result."

By the way, what does that hidden variable lambda represent? It determines whether the solution turns out to be quantum or classical.

[Joy Christian]

.
It turns out that Fred's measurement functions are exceedingly simple:



It should now be easy to translate Fred's simulation into other programming languages, such as Python or R.
.

[FrediFizzx]

Thanks Joy. Yes, we were able to simplify that a bit. Looks good. Most all the rest of the Mathematic programming is for getting the trial numbers matched up correctly.

Anyways, it is good that Gill is finally finished off for good. Wasted a good part of his life being a Bell fanatic for nothing. It's not like we didn't warn him.
.

[Joy Christian]

Thanks Joy. Yes, we were able to simplify that a bit. Looks good. Most all the rest of the Mathematic programming is for getting the trial numbers matched up correctly.

In fact, there is only one hidden variable in the model ---- the angle theta --- lambda is not needed as a separate variable:



And there is a one-to-one correspondence between what is emitted by the source and what is detected by the detectors. So no detection loophole!
.

[FrediFizzx]

That's correct because I made lambda a function of the singlet vector angle.
.

[Joy Christian]

.

.

[Austin Fearnley]

Hi Fred

IMO you need to check your code. I have just run your formulae in an excel spreadsheet and found abs correlation = 0.495650509
for a = 0 degrees and b = 45 degrees. I only used 360 pairs at evenly spaced values of theta.

I may of course have made a mistake as it came out like this first time and I have not checked afterwards. But it is not difficult to run on a spreadsheet. It only took one hour (today, while lunch was cooking!). I will check later if you do not find a problem at your end.

I tried the spreadsheet run as I like the cos^2 (theta/2) part of the formulae. With chimes of Malus. But I did not really expect it to work as I have run a variety of cos^2 (theta/2) formulae a few months back. So I tried in hope rather than expectation of success. I found in my methods that I could only make it work using retrocausality, as you know.

Best wishes
Austin

[FrediFizzx]

@Austin I have absolutely no idea of what you are describing. Try again.
.

[Guest]

R code using the formulas provided by Joy Christian in the image title "Fred's Disproof of Bell's Theorem".

Code: Select all

A <- function(a, th) {
    ifelse(abs(cos(a - th)) > 0.25 * cos(th / 2)^2,
           -sign(cos(a - th)),
           -sign(sin(a - th)))
}

B <- function(b, th) -A(b, th)

N <- 10^6
th <- runif(N, 0, 2*pi)

a <- 0
b <- pi/4

mean(A(a, th) * B(b, th))
-cos(a - b)


[Austin Fearnley]

Hi Fred

Not sure about the confusion?

Today, I have run your formulae (the ones very recently posted above in this thread) in an excel spreadsheet and found abs correlation to be 0.495650509
for a = 0 degrees and b = 45 degrees. I only used 360 pairs at evenly spaced values of theta.

This does not beat the classical Bell correlation of 0.5.

Note that IMO it is possible to use a spreadsheet for these formulae rather than a computer program (and I have done it).
(BTW I found the very recent formulae posted by you and Joy very useful whereas I was not inclined to try to interpret your posted computer code.)

Best, Austin

[FrediFizzx]

Hi Fred

Not sure about the confusion?

Today, I have run your formulae (the ones very recently posted above in this thread) in an excel spreadsheet and found abs correlation to be 0.495650509
for a = 0 degrees and b = 45 degrees. I only used 360 pairs at evenly spaced values of theta.

This does not beat the classical Bell correlation of 0.5.

Note that IMO it is possible to use a spreadsheet for these formulae rather than a computer program (and I have done it).
(BTW I found the very recent formulae posted by you and Joy very useful whereas I was not inclined to try to interpret your posted computer code.)

Best, Austin

You are going to have to post your code or your Excel spreadsheet.
.

[FrediFizzx]

R code using the formulas provided by Joy Christian in the image title "Fred's Disproof of Bell's Theorem".

Code: Select all

A <- function(a, th) {
    ifelse(abs(cos(a - th)) > 0.25 * cos(th / 2)^2,
           -sign(cos(a - th)),
           -sign(sin(a - th)))
}

B <- function(b, th) -A(b, th)

N <- 10^6
th <- runif(N, 0, 2*pi)

a <- 0
b <- pi/4

mean(A(a, th) * B(b, th))
-cos(a - b)


Well, the part you are missing is that you are not matching trial numbers. I suppose that is missing in what Joy did also.
.

[Joy Christian]

R code using the formulas provided by Joy Christian in the image title "Fred's Disproof of Bell's Theorem".

Code: Select all

A <- function(a, th) {
    ifelse(abs(cos(a - th)) > 0.25 * cos(th / 2)^2,
           -sign(cos(a - th)),
           -sign(sin(a - th)))
}

B <- function(b, th) -A(b, th)

N <- 10^6
th <- runif(N, 0, 2*pi)

a <- 0
b <- pi/4

mean(A(a, th) * B(b, th))
-cos(a - b)


I am not sure what this R code is showing? Is it agreeing or disagreeing with Fred's result?
.