Classical versions of q-Gaussian processes: conditional moments and Bell’s inequality

Since \tilde{X},\tilde{Y},\tilde{Z} are bounded random variables, to prove (?) we need only to verify that for arbitrary polynomials P,Q,

This is equivalent to

see (?). The latter follows from (?) and (?), proving (?).

To prove (?), we verify that for arbitrary polynomials P,Q we have

By definition (?), this is equivalent to

It suffices to show that (?) holds true when P=H_n and Q=H_m are the q-Hermite polynomials defined by (?). Formula (?) implies that the left-hand side of (?) is given by

and the right-hand side becomes

Using formulas from Lemma ? we can see that both sides are zero, except when m=n or m=n\pm 2. We now consider these three cases separately.

Case m=n+2: Using Lemma ?, (?) simplifies to

This equation is satisfied when coefficients A,B,C satisfy the equation

Case m=n-2: Using Lemma ?, (?) simplifies to

This equation is satisfied whenever

Case m=n: We use again Lemma ?. On both sides of Eq. (?) we factor out \langle f, h \rangle^{n-1}[n]_q!, and equate the remaining coefficients. (This is allowed since we are after sufficient conditions only!) We get

Now we use [n+1]_q=1+q[n]_q. Suppressing the correction to the constant term (i.e., the term free of n), we get

where c+\dots denotes the suppressed constant term corrections.

This equation holds true when the coefficients at [n]_q match, which gives

and the constant terms match: c+\dots=D. The latter holds true when the expectations are equal (n=m=0), and hence this condition is equivalent to (?). The remaining three equations (?), (?), and (?) have a unique solution given by the expressions (?), (?), (?).

Since the case q=-1 is well known, we restrict our attention to the case -1<q<1. Our starting point is expression (?). A computation shows that the conditional variance Var(\tilde{Y}|\tilde{X},\tilde{Z}):=E(\tilde{Y}^2|\tilde{X},\tilde{Z})-\left(E(\tilde{Y}|\tilde{X},\tilde{Z})\right)^2 is as follows.

The right-hand side of this expression must be non-negative over the support of \tilde{X},\tilde{Z}. It is known, see [Bryc98, Lemma 8.1] or [BKS97, Theorem 1.10], that \tilde{X},\tilde{Z} have the joint probability density function f(x,z) with respect to the product of marginals \nu_q. Moreover, f is defined for all -2/\sqrt{1-q}\leq x,z\leq 2/\sqrt{1-q} and from its explicit product expansion we can see that

is strictly positive. In particular, the right-hand side of (?) must be non-negative when evaluated at \tilde{X}=\sqrt{2}/\sqrt{1-q},\tilde{Z}=-\sqrt{2}/\sqrt{1-q}.

Using formulas (?), (?) with the above values of \tilde{X},\tilde{Y} we get the rational expression for the conditional variance which can be written as follows.

Therefore

Since the assumptions imply that 1-q\langle f, h \rangle^2+(1+q)\langle f, h \rangle\langle f, g \rangle\langle g, h \rangle\leq 1+\langle f, h \rangle^2, this implies (1-\langle f, h \rangle)^2 \geq \left(\langle f, g \rangle-\langle g, h \rangle\right)^2, proving (?).