Pages Menu
Categories Menu

Posted by on Jun 3, 2013 in EPR paradox | 5 comments

Quantum Archives censorship: & more about Computer Simulation

Now that I am waiting to hear if my paper on the Einstein Podolsky Rosen paradox will be accepted or dinged, I thought that I might try to post the paper on the Cornell University Library of archives: I used to post on the quantum archives, ArXiv.  Those early attempts were more learning events for me, and certainly I had a long way to go, but my basic ideas were sound and borne out as my model for spin evolved. However that is why I (used to) view the ArXiv as a good place to try out new ideas and get feedback.

Nonetheless, the Moderators blacklisted me in 2009, denying me the exposure that others have, and for unknown reasons too.  When I asked for the reasons, I get a note with no name, other than signed by  “the Moderators”  stating that in order to find out why they blacklisted me, I should submit the paper for publication.

By the way, most of the papers on the archives never get published. All I can infer is being Local Realist is enough for them to censor me. I am not the only one.  I have no idea how many people have been similarly censored but I have personal knowledge of two others.

I went to submit and got the following:


I am not endorsed.  I needed, therefore, and endorser.  I found someone whose work I respect and who I have always believed to be fair, and thought I would give it a try.  I will keep his name confidential.  ArXiv sent me an email to forward to XXX:


Could I impose on you to endorse me on the quantum archives, see below? I have finished my paper and want to post it there.

A local realistic reconciliation of the EPR paradox.


I recall you said to me that only a computer simulation would be taken seriously.  I have it and my model works, but it has some interesting differences.

Let me know either way if you can do this 

Best wishes,


B. C. Sanctuary requests your endorsement to submit an article to the quant-ph section of arXiv. To tell us that you would (or would not) like to endorse this person, please visit the following URL:

If that URL does not work for you, please visit

and enter the following six-digit alphanumeric string:

Endorsement Code: XXXXX

The reply I got was:

Hi Bryan,

I am not against endorsing you, but I want to look at what you wrote. Since I am completely overloaded with work for the next three weeks I won’t be able to do it any sooner.

The “simulation” however is not what I keep asking for. This has fixed settings. There is no facility to provide lists of settings to two independent instances of the program. I have no doubt that this simulation can achieve almost any value.



I replied:


Maybe I can suggest that one of your students looks at it

My simulation is based upon a model of spin. I have no control over the outcome. You are jumping to conclusions. I will be very surprised if you will find any conceptual errors in the paper.

Since the paper is submitted to Phys Rev A, I should get the acceptance within three weeks, and with that the archives cannot refuse me.  But your endorsement would get it up quicker.  I would think that opposing ideas are good for the field.  It does not matter anyway because I can post it elsewhere, just the archives would be nice, and I am pretty sure that even if you endorse me, the moderators will reject it based upon its title alone until it is accepted.


Best wishes


He replied:

Hi Bryan,

unfortunately you chose to ignore my central point:

The simulation needs to be local, i.e. two separate program parts and externally, randomly specified setting choices for each particle pair.



My reply bought in the Java Programmer, Chantal Roth, who wrote the program I used in my simulation:


The author of the java program I am using has agreed, and will comment publicly soon, that the simulation is completely local and no rules are broken (no signalling etc). Your requirement to use two computers is effectively done on one, simply by taking a product state and choosing the filter angles independently.  Your fight here is not with me, but with the programmer, Chantal Roth available at

 The problem with EPR is

1. The act of measurement changes the system and

2. The classical CHSH Bell’s Inequalities does not take into account non-commutation.

Conclusions are based, of course, on my model which, because of Heisenberg cannot simulate more than one half the quantum correlation in one simulation (one experiment). 

CHSH is not violated by my model (nor your experiment when spin has two axes).  Bell’s original inequalities are violated.

I have not made any errors in the simulation unless two computers, you want to use, satisfy some other “loophole” –so why are two computers necessary?  Roth’s program does the job.

You are again jumping to conclusions likely based on past attempts. The paper is readable (ideas are simple) and uses only quantum and no classical notions, and a loophole free simulation.


Reply from XXX

Dear Bryan,

I have no fight with the programmer, in fact I could not find any contact information. Sourceforge forces me to register to send a message.

In any case the simulation definitely has no facilities that I could provide a list of settings for each particle pair. So whatever you claim this is not enough for me.

Cheers XXX

After I introduced XXX to Chantal,  chantal wrote:


I originally wrote the program to understand the problem, and to try out different models. The basic setup is totally simple and can be verified by anyone easily who knows a little bit of programming:

1) it generates pairs of particles, and at this point hidden variables are shared

2) particles move to detectors A and B — from now on no more sharing of data of any kind —

3) detector orientation is set randomly

4) measurement takes place, record angle between AB, and if measurement yielded the same result or not

5) repeat 1-5 a gazillion times

6) compute Rab and so on, create plot

To allow the modeling of different theories, there are 2 classes, which are separate from the main program, where one can put in the formulas. One class for the hidden variables, one for the measurement (let me know if you need more detail :-).

(So this corresponds to the external settings you refer to I assume.)

What I can say is that my original, unmodified program satisfies all rules in the game and there are no cheats (and I am sure you would agree if you would take a closer look, and I’d be happy to show in detail – there is no hidden sharing of objects or other “tricks”), and at least so far there has been no way to get cos(a) without breaking those rules.

(The only way one can do is it of course if not all pairs are measured… but that is already breaking one of the rules, and that already requires an “illegal” modification).

From reading the modifications done to the program, I can confirm that each half of the experiment is done within the rules.

Best wishes and please feel free to ask any question….


PS: re external settings:

If you like I can change the program so that it takes a text file or similar as input for the filter angles (and particle orientation if desired) Is that maybe what you mean ?

But if you look at the code, it uses the standard random number generator… anyone would do and you could also switch to a totally different random generator.

Or if you like, the random generator could also be moved to another class so that you for instance could pick your own :-).

And XXX replied:

Dear Chantal,

thank you for your detailed response. The requirements are simple and you have already made some good suggestions.

1)      The program needs to be able to take a list of settings, one for each pair, e.g. a text file with one angle (or even just an index for the choice of either of two fixed angles that one could set in the user interface)

2)      To one instance of the program I will only provide the setting of one side so that locality is guaranteed.

3)      The program simply needs to generate the measurement result of that side for each pair (one side), i.e. a text file with a single number (+/-1) per line and as many lines as I provided settings.

I would then simply run two instances separately and afterwards calculate the correlation externally from the generated text files.

You can use any built-in correlation you want, e.g you can run the random number generator with the same seed every time.

I know that this is extremely simple for you. Bryan will probably need to adjust his implementation of the measurement functions.

All the best


Even though I did not get endorsed, the discussion brought out a number of points and gave more insight into how Computer Simulations are done.  Chantal has the final word today:

Dear XXX,

1) no problem

2/3) The problem I see with two separate instances is:
initially when the pairs are generated, the data is shared (the hidden “model”). So this needs to happen in the same instance (VM in case of Java).

If you really wanted several instances, then you would really need three programs, one that generates the pairs and shares the model, and two for measurements.
This would make the implementation complicated (the data of each particle would have to be save to a file and then read again by each measurement instance etc…)

I am not sure what you gain with all these changes – I know that you want to make sure there is no hidden communication going on :-), but you can see from the code itself what it does (it is by all means not complicated code at all), and it does not take much time to read the program :-). I’d be happy to remove any plotting code and strip it down to its bare minimum – at the end it is really just a few lines of code, basically this part for one “pair” (formula is where the user can put in his formula for measurement, and hiddenVars contains the model that is shared between the particles)


        Particle pA = new Particle("A");
        Particle pB = new Particle("B");
        hiddenVars.shareHiddenVariables(pA, pB);
        // set the angles of Filters A and B
        double angleA_deg = filterA.selectFilterSetting(Setup.A_ANGLES);
        double angleB_deg = filterB.selectFilterSetting(Setup.B_ANGLES);
       // now we send them to the detector and measure the probability for measuring spin +1
        spinA = filterA.measure(pA);
        spinB = filterB.measure(pB);  
        DataPoint result = new DataPoint(angleB_deg - angleA_deg, spinA, spinB, pA.getTheta());

And the class contains the comutation for the Rab and CHSH, such as this:

public double getCHSH() {
        double RA1B1 = computeRawCorrelations(Math.abs(Setup.A1 - Setup.B1));
        double RA1B2 = computeRawCorrelations(Math.abs(Setup.A1 - Setup.B2));
        double RA2B1 = computeRawCorrelations(Math.abs(Setup.A2 - Setup.B1));
        double RA2B2 = computeRawCorrelations(Math.abs(Setup.A2 - Setup.B2));

        double chsh = Math.abs(RA1B1 - RA1B2 + RA2B1 + RA2B2);
        return chsh;


/** compute raw product moment correlations */
public double computeRawCorrelations(double angle_in_degrees) {
        if (angle_in_degrees == 360) {
            angle_in_degrees = 0;
        int bucket = getBucket(angle_in_degrees);
        //well, eq are the number of ++ or -11...
        double eq = getEq(bucket);
        double neq = getNeq(bucket);
        // the total (usually eq + neq)
        double tot = getTotal(bucket);
        double Rab = 0;
        // we don't want division by 0 🙂
        if (tot > 0) {
            Rab = (eq - neq) / (eq + neq);
        } else {
            err("Not enough data points for " + angle_in_degrees);

        return Rab;

If you let me know where you see a problem with this I’d be happy to change it…

Best wishes,


Wonder what is next.


  1. Bryan writes “Conclusions are based, of course, on my model which, because of Heisenberg cannot simulate more than one half the quantum correlation in one simulation (one experiment). CHSH is not violated by my model (..,). Bell’s original inequalities are violated.”

    Bell’s original inequalities assume local realism and perfect anti-correlation (when equal settings are used in the two wings of the experiment). In one simulation we would find r(A,A) = -1/2, not -1. So Bell’s original inequality simply does not apply.

  2. Hi Richard,

    Thanks for your comments but you should read my paper as well as the blog.

    Since CHSH is derived from Bell’s original expression, then how can you say it is wrong, or not applicable?

    Recall Bell’s inequalities are classical and my model is quantum. The partitioning of LHVs in the CHSH derivation is ok except you have to understand that those two regions are quantum mechanically incompatible and cannot be measured simultaneously.

    Please look at the part of the paper with subsection:

    Coherence, filter setting and consistency with Bell’s Inequalities

    For a spin with two axes, each spin carries exactly the same polarization as seen in Bell’s expressions.

  3. Dear Bryan

    Perfect anti-correlation plus violation of Bell implies violation of a corresponding CHSH inequality. You say that you violate Bell but not CHSH. Therefore you do not obtain perfect anti-correlation. Therefore you do not reproduce the singlet correlations.

    All these correlations, by the way, are of course to be understood as the directly observed correlations between actual measurement outcomes of actual measurements.

    If your computer simulation is thought of as a simulation of actual measurement outcomes of actual measurements, and if moreover the code is “separable” (and I believe your programmer, that it truly is), then it cannot violate CHSH, and therefore it cannot reproduce the singlet correlations. As you said “Bell’s inequalities are classical and my model is quantum”. A “separable” computer simulation is also classical.


  4. Richard. I do indeed value your comments but you are thinking in the usual box and that is the reason people have failed so far to resolve EPR.

    In my model, NO experiment can violate any of Bell’s Inequalities. You have a choice: usual spin, violation; two axes, no violation.

    You cannot beat Heisenberg.

    Consider it this way: you are thinking of a quantum state. I am thinking of one particle that makes up that quantum state which is defined by a unique set of LHV.

    If I then look at the correlation for such a specific EPR pair in my model, my results show that for a fixed set of LHV, Bell’s original inequality is violated, but the CHSH is not.

    Since I am agreeing that no product state will violate Bell’s Inequalities, you should be happy. That is what you have been saying for years, and I agree. However the reason we agree is different. You believe quantum mechanics is complete and its non-locality comes out in Bell’s Inequalities. I find spin carries the quantum correlation in beable states half of which is destroyed by the act of measurement.

    I agree with all you say because you are using the usual spin as a point particle vector operator. I believe that if you think from the point of view of my beables states, you will agree with me.

    So if you think in my model, then I do have perfect anti correlation which exists until you measure it.

  5. So are you saying that experimenters who claimed to violate Bell’s inequality or the CHSH inequality were in fact mistaken? They did not observe r(a,b) = – a.b at all? They only actually observed r(a,b) = – a.b / 2 ?

Post a Reply

Your email address will not be published. Required fields are marked *