Coming Soon!

[FrediFizzx]

@gill1109 I suspect you're just being lazy. Ask what is the first thing you don't understand. Heck, I am way too lazy to write this in another code. Plus having more fun working on other things.
.

[gill1109]

@gill1109 I suspect you're just being lazy. Ask what is the first thing you don't understand. Heck, I am way too lazy to write this in another code. Plus having more fun working on other things.
.

The first thing I don’t understand is how you actually match, and the second thing I don’t understand is what the outcomes then actually are.

I am certainly lazy. Plus I too am having a lot of fun working on quite different things, too.

Anyway, if you think what you’ve done is important, I hope you’ll write it up for a journal publication. Even if you don’t want to go all the way, maybe other folk are interested too, but like me, struggle with reading Mathematica code and/or do not wish to support Wolfram’s business.

I’m impressed at your achievement. Please see my requests as positive encouragement.

[FrediFizzx]

@gill1109 I suspect you're just being lazy. Ask what is the first thing you don't understand. Heck, I am way too lazy to write this in another code. Plus having more fun working on other things.
.

The first thing I don’t understand is how you actually match, and the second thing I don’t understand is what the outcomes then actually are.

The original trial numbers are recorded as the third item in CA1 and CB1. Those are the raw data outputs of A and B of which there are 5 items in each all necessary for proper matching. Which are then spit into two lists of events each. One for the non-constrained events and one for the constrained events according to the HV. I'm actually going to move the splits to the matching section. Now, what happens when you split the data like that, you end up with "orphans" in the non-constrained events because their partner on the other side went to the constrained event list. So, we use the recorded trial numbers to figure out which events are orphans and then move them to the constrained event list where they actually belong. Then after some further processing and sorting we end up with outA5 and outB5. Those are the cleaned up matching outcomes which are then analyzed.
.

[FrediFizzx]

Here is the simulation that uses Joy's 3-sphere quaternion model. 400,000 trials at one degree resolution.



EPRsims/newCS-13-S3quat-forum.pdf
EPRsims/newCS-13-S3quat-forum.nb

I've updated the Mathematica programing by moving outA1, outA2, outB1 and outB2 to the matching section. And renamed the A and B raw data outputs to outA and outB.

Enjoy!!
.

[Joy Christian]

.
I have two questions:

(1) I am not sure whether quaternions are necessary or add anything to your model. In what way do they improve the model?

and

(2) Would the final plot be better, or get closer to the -cosine curve, with a larger number of trials; say, 2 or 3 million trials?
.

[FrediFizzx]

.
I have two questions:

(1) I am not sure whether quaternions are necessary or add anything to your model. In what way do they improve the model?

and

(2) Would the final plot be better, or get closer to the -cosine curve, with a larger number of trials; say, 2 or 3 million trials?
.

It is not a matter of improving the model. It is just a demonstration that the quaternion model also works.

I would think that it seems the model is converging to -a.b. It took 10 hours to run 400K trials so it would take days to do 2 or 3 million trials. I'm not that patient.
.

[FrediFizzx]

Of course we have to demonstrate that Joy's 3-sphere GA model works also. The Clifford package for Mathematica is a bit slow so I only did 200,000 trials at one degree resolution.



EPRsims/newCS-12-S3GA-forum.pdf
EPRsims/newCS-12-S3GA-forum.nb

I had a heck of a time figuring out the sine of the GA product angle but as you can see I finally got it.

Enjoy!!
.

[gill1109]

.
I have two questions:

(1) I am not sure whether quaternions are necessary or add anything to your model. In what way do they improve the model?

and

(2) Would the final plot be better, or get closer to the -cosine curve, with a larger number of trials; say, 2 or 3 million trials?
.

Once Fred (or someone else) has written down math formulas or pseudo code describing what he is doing, we can investigate analytically whether or not the simulated curves will converge to the negative cosine. I think that if he wants to publish his findings, he’ll have to do exactly that. He has to make his research ‘reproducible’ and transparant.

[FrediFizzx]

Once Fred (or someone else) has written down math formulas or pseudo code describing what he is doing, we can investigate analytically whether or not the simulated curves will converge to the negative cosine. I think that if he wants to publish his findings, he’ll have to do exactly that. He has to make his research ‘reproducible’ and transparent.

Hmm... More freakin' nonsense from the master nonsense maker. I already explained some of it to you and you still don't have a clue. Here is some more explanation.

The telltale lists.



This is from item 4 in the raw outputs. The f1 and f2 events are less than the HV, g1 and g2 are events more than HV. You can see right off the bat, the first event between A and B don't match.
.

[jreed]

.
I have two questions:

(1) I am not sure whether quaternions are necessary or add anything to your model. In what way do they improve the model?

and

(2) Would the final plot be better, or get closer to the -cosine curve, with a larger number of trials; say, 2 or 3 million trials?
.

Once Fred (or someone else) has written down math formulas or pseudo code describing what he is doing, we can investigate analytically whether or not the simulated curves will converge to the negative cosine. I think that if he wants to publish his findings, he’ll have to do exactly that. He has to make his research ‘reproducible’ and transparant.

I have been trying to keep up with Fred, but he keeps changing the program every few days. It has been modified a lot since I first translated Fodje's EPR Simple version. Now we have those previously lost detections saved and somehow brought back into the calculation. I don't understand what physical experiment this is meant to simulate. The code has also been upgraded through the use of some more advanced Mathematica routines such as pure functions, MemberQ, Intersection, Catenate and several others. This is fine for Mathematica users who can look them up, but will make the program harder to understand for those unfamiliar with Mathematica.

Once the program reaches a final stable state, I can try to give a description of what it's doing.

[FrediFizzx]

Once the program reaches a final stable state, I can try to give a description of what it's doing.

Most of those "updates" were different versions like CHSH, Joy's quaternion and GA models. It's stable. Here you go.

EPRsims/newCS-13-forum.nb

But most likely, Gill will never understand it unless you program it in R. And then, he probably still won't understand it and probably won't want to understand it because it will kill his "world".
.

[jreed]

Thanks Fred. I'll see if I can explain it in pseudo code. I don't program in R.

[FrediFizzx]

Thanks Fred. I'll see if I can explain it in pseudo code. I don't program in R.

You're welcome, John. Yeah, I don't program in R either even though I have R Studio on my computer. It just seems really lame compared to Mathematica. I started to learn Python but lost it when I had to get a new computer. But now that I can do geometric algebra in Mathematica using the Clifford package, I'm all set.
.

[gill1109]

Once the program reaches a final stable state, I can try to give a description of what it's doing.

Most of those "updates" were different versions like CHSH, Joy's quaternion and GA models. It's stable. Here you go.
EPRsims/newCS-13-forum.nb
But most likely, Gill will never understand it unless you program it in R. And then, he probably still won't understand it and probably won't want to understand it because it will kill his "world".

It will not kill my world. You and I live in different worlds, Fred. Completely uncommunicating, parallel, worlds.

Thanks Fred. I'll see if I can explain it in pseudo-code. I don't program in R.

Looking forward to seeing it! Python or C would also be fine. In fact, almost anything but Mathematica would be fine.

One cannot play with the code without buying Mathematica. Actually, not quite true, I get myself a trial version for a few week's free trial once in a while. I could try to do that again. Results in a stream of junk mail from Wolfram. But I'm not in a hurry this time. I think Fred will agree that if I run his program but with N = 1 repeatedly, independently (ie with new random seed each time), the collected results won't violate Bell's inequality.

[FrediFizzx]

It will not kill my world. You and I live in different worlds, Fred. Completely uncommunicating, parallel, worlds.

One cannot play with the code without buying Mathematica. Actually, not quite true, I get myself a trial version for a few week's free trial once in a while. I could try to do that again. Results in a stream of junk mail from Wolfram. But I'm not in a hurry this time. I think Fred will agree that if I run his program but with N = 1 repeatedly, independently (ie with new random seed each time), the collected results won't violate Bell's inequality.

What is this nonsense? Just like Michel said, changing the goal posts already. It's ridiculous! You can't collect the results of that. It's meaningless.
.

[FrediFizzx]

Well, actually that might work if the collected terms are averaged properly. But changing the seed each time is stupid. That is like having no seed at all.
.

[gill1109]

Well, actually that might work if the collected terms are averaged properly. But changing the seed each time is stupid. That is like having no seed at all.

Indeed, it should work if the collected terms are averaged properly.

Every time you start up Mathematica and take a new random sample, the seed is different from what it was the time before. Unless you deliberately set it to be the same, which is useful for testing purposes. Are you saying that Mathematica is stupid?

I'm not changing any goal posts.

QM makes predictions of the probabilities of what you will see when N = 1. You want to reproduce the predictions of QM in a local realist way? Then it must be possible, at least as a thought experiment, to run your program with N = 1 on 1 million computers and gather the data later. We certainly could stop and start Mathematica 1 million times on one computer and gather the data together afterwards. https://reference.wolfram.com/language/ ... ml#8730447

When you start Mathematica it sets its own seed using "the time of day and certain attributes of the current Wolfram Language session". All programming languages are like this. After the initial seed is set, subsequent calls to get a random number use a deterministic algorithm to get the next ones. That's why these things are called pseudo random number generators. I worked on these many years ago, inventing new ones for particular computational architectures.

[FrediFizzx]

Indeed, it should work if the collected terms are averaged properly.

Every time you start up Mathematica and take a new random sample, the seed is different from what it was the time before. Unless you deliberately set it to be the same, which is useful for testing purposes. Are you saying that Mathematica is stupid?

No, I'm saying you are stupid for saying to change the seed each time. Just don't have a seed.
.

[jreed]

Thanks Fred. I'll see if I can explain it in pseudo code. I don't program in R.

I've looked at the code for the final version you sent, and I'm confused. Here are some questions:

In the do loops to generate particle data, there is an inner If statement and an outer If statement. The inner If doesn't need to be inside the outer If as far as I can see.

In those same loops, there are several Sign[Cos... statements, and several Sign[Sin... statements. Why are the Sign[Sin statements there?

There is a lot of complex Mathematica stuff to match trial numbers. This was never necessary in previous versions. Why was it put there?

[FrediFizzx]

Thanks Fred. I'll see if I can explain it in pseudo code. I don't program in R.

I've looked at the code for the final version you sent, and I'm confused. Here are some questions:

In the do loops to generate particle data, there is an inner If statement and an outer If statement. The inner If doesn't need to be inside the outer If as far as I can see.

In those same loops, there are several Sign[Cos... statements, and several Sign[Sin... statements. Why are the Sign[Sin statements there?

There is a lot of complex Mathematica stuff to match trial numbers. This was never necessary in previous versions. Why was it put there?

Your right about the If[If. You can fix that. I saw that after I posted the code and fixed it. I'm not sure how that happened. Sorry about not telling you.

If you look at outA and outB, you will see 5 items each being recorded in lists. Those extra Cos[ and Sin[ statements are for recording items 4 and 5 that are necessary for proper matching of the events. Of course the 3rd item is just the original trial number also necessary for proper matching. IOW, if you just sent a, b, A and B to the analysis, you are matching constrained events (less than the HV) in A or B with non-constrained events in A or B and you will just get straight lines. The matching fixes that so that the partner to a constrained event is changed to be a constrained event. Hope that helps.
.