Joy Christian's colourful exploding balls experiment

[gill1109]

Given the completely transparent proof I have linked above and my repeated explanations, it is quite disingenuous of both Gill and Zen to continue the false propaganda about the universal quantifier. It is a cheap trick to discredit. I do not want to hear any more bogus claims about my model. It is time to grow up.

Yes, the completely transparently wrong proof. If you want to discuss it: start a new thread. Here we discuss the experiment.

[Joy Christian]

Given the completely transparent proof I have linked above and my repeated explanations, it is quite disingenuous of both Gill and Zen to continue the false propaganda about the universal quantifier. It is a cheap trick to discredit. I do not want to hear any more bogus claims about my model. It is time to grow up.

Yes, the completely transparently wrong proof. If you want to discuss it: start a new thread. Here we discuss the experiment.

You started this nonsense here. Stop making bogus claims about my work. The proof I have provided is trivially true. Anyone who has any knowledge of differential geometry can see that. Anyone who has any knowledge of Riemannian geometry can see that. Anyone who has any knowledge of quaternionic algebra can see that. Anyone who has any knowledge of geometric algebra can see that. Here is the proof again: http://libertesphilosophica.info/blog/w ... mplete.pdf. It is elementary.

[gill1109]

Joy said

You started this nonsense here.

No. Zen and Joy together started the nonsense here. Here's the proof:

Code: Select all

good <- abs(ca) > f & abs(cb) > f  ## Select the 'states'

Indeed: "Select the states." Not "Reject the data."
The initial or complete state of the system in the model is the pair (e, theta), NOT just the vector e by itself.

Not only nonsense, but off-topic nonsense to boot.

This thread is about an experiment. An experiment is described in kitchen-garden classical terms. It's a set of instructions to some lab assistants and computer programmers, who don't need to know anything at all about the theories being tested by the experiment. It has to be reproducible - other people in other labs can do it too. Quite independently of which theories they know, which theories they believe in. In fact, better that they don't know, and better that they don't have strong prior beliefs one way or the other, because we don't want any subconscious bias on the part of the guys who do all the hard nuts and bolts work to spoil the experiment.

[FrediFizzx]

It's not off topic because the experiment is about testing Joy's model macroscopically.

[Joy Christian]

I am reproducing my recent comments from another thread, because this is a very important point for my proposed experiment:

All I'm saying is that each correlation be calculated on a separate set of particles. You can generate one list of vectors but you must sample without replacement so that no pair of vectors contributes to more than one correlation. Only then will it be equivalent to 4 separate sets.

I agree, as you already know.

And as you have kindly noted, this is what I have been arguing for years:

For the record, let me repeat that equation (16) of my attached
experimental paper describes exactly how the expectation values
E(a, b), E(a', b), E(a, b'), and E(a', b') are to be computed in my
proposed experiment. Four separate sums are to be calculated as
follows

E(a, b) = 1/N Sum_j A_j B_j ,

E(a, b') = 1/N Sum_j A_j B'_j ,

E(a', b) = 1/N Sum_j A'_j B_j ,

and

E(a', b') = 1/N Sum_j A'_j B'_j .

It is a matter of indifference whether N here is chosen to be the same
or different for each of the four alternatives.

The experimental procedure described in my paper is unambiguous.

[Ben6993]

After working through Michel's posts, I think that the experiment as described at the end of Joy's paper gives Joy most chance of success. (Not that I think that the experiment will yield an average correlation exceeding 0.6.)

Joy declared that outcomes to be correlated are based on two paired fragments per explosion.

The number of balls exploded is to be large. The averaging is done over N trials, which I assume is N exploding balls, where N is large.

That implies that one single fragment gives rise to only one outcome in the experiment. It implies that couterfactual estimates of outcomes will not be used, nor will multiple outcomes be used from a single fragment. Michel's posts indicate to me that multiple outcomes from the same fragment are akin to counterfactual data which Joy needs to avoid.

[gill1109]

After working through Michel's posts, I think that the experiment as described at the end of Joy's paper gives Joy most chance of success. (Not that I think that the experiment will yield an average correlation exceeding 0.6.)

Joy declared that outcomes to be correlated are based on two paired fragments per explosion.

The number of balls exploded is to be large. The averaging is done over N trials, which I assume is N exploding balls, where N is large.

That implies that one single fragment gives rise to only one outcome in the experiment. It implies that couterfactual estimates of outcomes will not be used, nor will multiple outcomes be used from a single fragment. Michel's posts indicate to me that multiple outcomes from the same fragment are akin to counterfactual data which Joy needs to avoid.

If you want to know what the experiment is, you first read Joy's experimental paper carefully, and then you read this thread looking for agreements between Joy and Richard. What anybody else thinks is irrelevant, unless it led Joy or myself to request movement of the goal-posts, and said movement was agreed by the other party.

So far Joy has not moved the goal-posts one inch, nor have I.

Per explosion (run, trial) k, k = 1, ..., N, there are two paired fragments (hemispheres).
One fragment gives rise to one outcome in the experiment ... which is a direction.

The video cameras record the explosion and produce video film.
Computer software calculates two directions u_k and - u_k and stores them in two computer files.
At the end of the experiment there are two computer files each with N directions. One set is the negative of the other set.

Afterwards we *calculate* the number - E(0, 45) + E(0, 135) - E(90, 45) - E(90, 135) and compare it to the number 2.4 according to http://rpubs.com/gill1109/Bet

[Joy Christian]

A recent exchange worthy of reproducing as a record on this thread (the last line is mine):

I predict that Joy's experiment is going to falsify Joy's theory.

Only macroscopically if E(a, b) != -a.b for the macroscopic experiment. The quantum experiments themselves support Joy's theory microscopically.

So far, no experiment has disproved local hidden variables theory.

So far, the experiments do not discriminate between quantum theory, Joy's theory, and local hidden variables theories.

Hallelujah!!! You have earned my respect for being the first Bell-believer to acknowledge this fact about my local-realistic framework (it is not yet a full theory).

[gill1109]

I don't know what it a Bell-believer is. I doubt I am a Bell believer.

I do believe QM is seriously inadequate.

Bell offered four *alternative* positions which one might like to take in view of his analysis. Later he admitted there exists a fifth, which I christened "Bell's fifth position" in a paper more than 10 years ago.

I think his analysis is correct but I don't know which of the five positions consequently needs to be adopted. This is partly a metaphysical issue and partly a matter of experiment.

Experiment is inconclusive. Metaphysics is partly a matter of taste.

I keep an open mind on all this.

I do think that Christian's analysis is incorrect.

[Joy Christian]

I do think that Christian's analysis is incorrect.

And what qualification do you have to make this false claim? In my opinion none whatsoever.

You are neither a physicist, nor a philosopher. You are not even a mathematician. You are simply a statistician. And my analysis is hardly about statistics.

[gill1109]

And my analysis is hardly about statistics.

I believe that this is one of the main problems with your analysis. I believe it is also a main problem in the understanding of various contributors to this forum here.

The difference between population and sample?

Error bars, standard errors, and p-values?

Statistical proof versus logical proof?

Theoretical physicists, experimental physicists, pure mathematicians, applied mathematicians, mathematical statisticians, applied statisticians, philosophers, and logicians all need one another. Science needs all of them. Science is a collaborative enterprise of all of humanity.

[FrediFizzx]

Ok, you guys. Let's get back on topic.

[gill1109]

Thanks Fred. I deleted my remarks about my credentials.

[Joy Christian]

An interesting suggestion has been made by Jonathan J. Dickau in a comment at the FQXi blog:

“One place the torsional effects Joy describes CAN already be observed, is in the slow-motion footage of atomic bomb blasts. So no colorful balls are needed; the experiment has already been done.”

To which I replied:

Why didn't I think of that? The raw data of the bomb experiment I have been urging for since 2008 may already exist! Of course, one has to analyze the data carefully to construct the 3D maps of spins I describe in my experimental papers. But you may have reduced the cost of my proposed experiment by 4 or 5 orders of magnitude!

[FrediFizzx]

There may be ordinary bomb blasts that data exist for that may show the torsion effect also.

[Joy Christian]

There may be ordinary bomb blasts that data exist for that may show the torsion effect also.

Indeed. We need lots of data. Footage of just atomic bombs may not be enough.

[gill1109]

We need not construct 3D maps. Just getting the correlations at four points is enough.

E(0, 45) = E(90, 45) = E(90, 135) = - 0.7
E(0, 135) = + 0.7

This is the most efficient experiment. The most bang for your buck. See the paper by Van Dam, Gill, Grunwald in IEEE-IT (2005), on arXiv in 2003. http://arxiv.org/abs/quant-ph/0307125

And 10 000 pairs of particles (hemispheres, ...) is plenty. The standard errors will be so small that you won't need a statistician to tell you if it is significant or not. The number of standard errors between the two theories will be better than in the experiment which clinched the existence of the Higgs.

Rutherford: if you need a statistician you did the wrong experiment.

Already, Zeilinger, Gisin and all the big guys are using the insights of Caroline Thompson, Jan-Ake Larsson, an other statisticians in order to get unequivocal experimental results. Rutherford was right. You need to consult with a statistician before you do your experiment, in order to make him or her superfluous after the experiment. At last the quantum optics guys are learning this ...

PS did Michel do the Python translation of the bet-resolution program yet? Is anyone doing a Mathematica version? Progress is stagnating.

[gill1109]

Where is Michel, and where is the translation of http://rpubs.com/gill1109/Bet into Python and Mathematica and Java? My R implementation of Joy's instructions for calculation of the four correlations, and for the subsequent determination of who wins the bet. Preparations for the experiment and the bet are stagnating. I can't only offer a piece of R code authored by me to the adjudicating committee. Who don't speak R anyway. We need versions in several languages on public display on internet, and authorized both by Joy and me. Everyone can read it in the language they prefer, and test them all (same inputs, same outputs).

This work *has* to be taken in hand by Joy's supporters.
Some of Joy's supporters need to support him by helping to get this done.

[Joy Christian]

For the record, I am reproducing here my recent reply to Richard Gill from the FQXi blog:

You are moving the goalpost because you are beginning to realize that you are going to loose when the actual experiment is done, and we will see the strong correlations, just as I have predicted. Stick to the original bet. I am not interested in any other bet. I will not be responding to yours posts unless they are about the actual experiment I have proposed.

As for the R code you have written for the bet, it is deeply flawed, as I have already pointed out elsewhere. It does not respect my condition of calculating each correlation separately---i.e., each correlation must be calculated on a separate set of particles. You can generate one list of vectors but you must sample without replacement so that no pair of vectors contributes to more than one correlation. Only then will it be equivalent to 4 separate sets, as both Michel and I have been insisting:

For the record, let me repeat that equation (16) of my experimental paper describes exactly how the expectation values E(a, b), E(a', b), E(a, b'), and E(a', b') are to be computed in my proposed experiment. Four separate sums are to be calculated as follows

E(a, b) = 1/N Sum_j A_j B_j ,

E(a, b') = 1/N Sum_j A_j B'_j ,

E(a', b) = 1/N Sum_j A'_j B_j ,

and

E(a', b') = 1/N Sum_j A'_j B'_j .

It is a matter of indifference whether N here is chosen to be the same or different for each of the four alternatives.

[gill1109]

I shall rewrite the R script for settling the bet between J. Christian and R. Gill, according to the new instructions from J. Christian. Strange. First he fully agreed with it. First it was a matter of complete indifference to him whether the four correlations were based on the whole data set or only on four disjoint subsamples. Then M. Fodje got him worried. The same M. Fodje who refuses to experimentally run my simulation of what happens when we follow his new instructions - my simulation which compares the two scenarios.

You see, Fodje already knows everything he needs to know, he has no need to learn anything new. We have agreed that the only correct upper bound to CHSH when based on disjoint samples is 4. That no (normal) experiment can violate that bound. That "4" is the only bound which cannot be violated by experiment. Since then, discussion has ceased.

Apparently Christian knows that Fodje knows a few things which Christian doesn't know, hence his retraction from a previous agreement, based on following Christian's written instructions in the experimental paper, to the letter.

Christian's experimental paper makes completely clear that every correlation is based on the whole sample. Will he revise that paper? If he wants to move the goalposts, he had better do just that. Then we can talk again about a bet on the outcome of the new experiment based on new instructions in a revised paper. For the time being, the only bet I have signed up for is the one based on the explicit instructions in Christian's paper. Lots of people have read that paper and agreed with my interpretation of the written words. Apparently Christian doesn't always mean what he writes.

To be honest, it's actually a matter of pretty near complete indifference to me whether E(0, 45) and the other three correlations are based on all N pairs of exploding balls or just on four disjoint random samples of about one quarter of them. I happen to know a thing or two about statistical sampling theory. As long as N is reasonably large, say 10 000 or more, I have no worries at all. But then I am merely a statistician.