# I Are entanglement correlations truly random?

Tags:
1. Dec 6, 2017

### entropy1

From the data of my computer code
Well, I can reassure you that was not my angle of approach. The cherry-picking part I introduced to illustrate how improbable it is to get a correlation out of pure randomness.
That's not entirely fair - I think it is a matter of starting point.

2. Dec 6, 2017

### Staff: Mentor

But you still haven't really explained what you mean by this.

3. Dec 6, 2017

### entropy1

Well, I am afraid I can't do better than this currently. I will ponder some more.

4. Dec 6, 2017

### Mentz114

You should stsrt by finding out the customary meanings of randomness and also how to calculate a correlation.

5. Dec 7, 2017

### Zafa Pi

A major problem in getting your question answered is that your terminology is sloppy, in fact truly sloppy.
You failed to make a definition. The terms "random" and "truly random" are neither used nor defined in probability texts. And after reading more of your posts it is not clear to me what you mean.

Let me give a simple concrete QM example:
Given an entangled pair from state √½(|00⟩ + |11⟩), we let A measure one of the pair at angle 0º, i.e. with measurement operator/observable $Z =\begin{pmatrix}1&0\\0&-1 \end {pmatrix}$.
We let B measure the other at 30°, i.e. with observable $½Z + √¾X =\begin{pmatrix}½&√¾\\√¾&-½\end{pmatrix}$.

The joint probability density of (A,B) is (1,1) with prob ⅜, (1,-1) with prob ⅛, (-1,1) with prob ⅛, (-1,-1) with prob ⅜. (1 & -1 are eigenvalues of the observables)
We see A and B agree with prob = ¾ = cos²30º, as usual.
The correlation coefficient is ½.
The marginal density of A is 1 with prob ½, -1 with prob ½. Same for B. A and B are not independent.

All of this is justified by repeated trials in the lab.

6. Dec 7, 2017

### entropy1

I thought I remembered the basics, but this morning my meds demand my brain, so I would have to look it up.

7. Dec 7, 2017

### entropy1

If the term is not defined in scientific literature, then why are you asking me, a layman, to define it? By the way, I gave one:
Anyway, I will think about what I mean by 'truly' random.

Last edited: Dec 7, 2017
8. Dec 7, 2017

### Staff: Mentor

Because you used it. We need to know what you meant by it.

9. Dec 7, 2017

### Zafa Pi

I don't know what you mean by fragment. Do you have a link?
If a1,a2, ... ,al with l=20 is a binary sequence, what is a fragment of 10 bits? Is it a subset of size 10? Is it a contiguous subset like a7,a8, ... ,a16? Or what?

10. Dec 8, 2017

### Zafa Pi

I now think you were trying to define a binary normal sequence, but failed.

"A sequence of bits is random if there exists no Program shorter than it which can produce the same sequence." ~ Kolmogorov
So obviously it is impossible to exhibit a Kolmogorov random sequence.

Neither normality or K-random imply one another. But all of this should be in the Probability section of PF. And none of this is relevant to QM.

11. Jan 13, 2018

### entropy1

Suppose I walk down the street, and each time I look to my right, a red car is passing. If I don't look, I don't know which color the passing cars have.

So the correlation between me looking and a red car passing is 100%.

So I assume the moments I look are random (A) and the cars passing have FAPP random colors (B).

So, in this case, with the correlation manifesting, are (A) and (B) "truly" random?

Since we generally do not see correlations like this always and everywhere, it should be, however not impossible, improbable to see this. So, I cannot determine whether there is a red car convention in town or not, since I don't know the counterfactual measurements (looking). So, would a string of red cars passing me still be random? After all it would require a red car convention. And if there is NO red car convention, would the string of cars passing still be truly random if the correlation with my looking direction would be 100% red cars? (Or, for that matter, would my peeking be random?)

The problem I see, is that if (A) and (B) are truly random, the measurements should be typical for what is reality. For example, based on my perceptions, I might say that in this street probably only red cars are allowed, while the counterfactual data is in contradiction with that.

You could also see it the other way round: I see typical cars passing, while when I'm not looking only red cars pass which I wouldn't know of. My assessment of the data might lead me to faulty conclusions.

So I think "randomness" is required to accurately assess reality.

Last edited: Jan 13, 2018
12. Jan 13, 2018

### stevendaryl

Staff Emeritus
There was a different thread in the "Set Theory, Logic, Probability and Statistics" forum on this topic. Random is relative to a model or theory. You can't know whether something is "truly" random unless you know what theory is correct. Which, of course, you can never know.

According to QM, the results of certain types of measurements are random, in the sense that QM doesn't propose any means of determining the values ahead of time. According to a different theory (maybe Bohmian mechanics), the results may not be random.

The facts you describe above is consistent with multiple explanations:
1. All the cars are red.
2. There are cars of other colors, but for whatever reason, you only have an impulse to look at a car when the car is red.
3. There are cars of other colors, but just by coincidence, you happened to look at the moments a red car is passing.
4. Etc.

13. Jan 14, 2018

### Stephen Tashi

The standard mathematical treatment of probability (which uses measure theory) says nothing about events actually happening. It doesn't have any axioms that say you can take random samples. It does not have a model of time as that notion is used in physics. So the standard mathematical theory does not deal with questions about a probability "before" or "after" some time or a probability that changes with the "actual" occurance of an event.

The standard techniques for applying probability theory to real life problems do assume that it is possible to take random samples and that events actually happen (or don't happen). In applications of probability theory the indexing set used in the abstract definition of "stochastic process" is often interpreted to be time in the physical sense.

The distinction between mathematical probability theory and interpretations that people make when applying it is blurred by the fact that only the most advanced texts on mathematical probability theory confine themselves to discussing that theory. The typical textbook on probability theory tries to be helpful by teaching both probability theory and its useful applications. For example, the "conditional probability" P(A|B) has a very abstract mathematical definition. However, typical textbooks present P(A}B) by interpreting it to mean "The probability of event A given that the event B has (actually) happened".

In mathematical probability theory, a specific sequence of numbers can be assigned a probability and it can be a member of a "sample space" on which a probability measure is defined. But there is no definition for a particular sequence of numbers being "random" or "not random". In mathematical probability theory, there is a definition for two random variables to be correlated However there is no definition for two specific sequences of numbers to be correlated. In this thread, there is the usual confusion involving numerical calculations done on specific sets of numbers to estimate mathematical correlation versus the mathematical definition of correlation.

Attempts have been made to create mathematical notions of randomness for specific sequences of numbers. These attempts are not "standard" mathematical probability theory.

When discussing physics, people are making their own interpretations of mathematical probability theory.