Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I know the result that says "There is a bijection between distribution functions and characteristic functions".

So I was wondering if there are necessary or sufficient conditions on the characteristic functions so that it is a char. function of a valid random variable. I need conditions that are easy to verify.

For instance, is $\varphi_{X_1,X_2}(x_1,x_2) = (1 + |x_1 x_2| e^{-\frac{(x_1^2 + x_2^2)}2 \sigma^2} )$ a valid characteristic function?

Thank you.

share|cite|improve this question
You might find this interesting – Sasha Sep 29 '11 at 5:16
up vote 4 down vote accepted

Bochner's theorem characterizes if a characteristic function corresponds to the random variable, but it's not an easy criterion.

It says that for all $x_1, \ldots, x_m \in \mathbb{R}^2$ and $c_1,\ldots,c_m \in \mathbb{C}$ the characteristic function $\phi$ should verify $$ \sum_{i=1}^m \sum_{j=1}^m \bar{c}_i c_j \phi(x_i - x_j) \ge 0 $$

For the characteristic function in question, choose $m=2$, and $c_1 = -1$, $c_2=1$, and $x_1 = \{1, \frac{1}{2}\}$ and $x_2 = \{ \frac{1}{2}, 1 \}$. Then the resulting sum is $-\frac{1}{2} \exp(-\frac{\sigma^2}{4} )$.

Here is Mathematica code:

In[210]:= \[Phi][{w1_, w2_}] := 
 1 + Abs[w1*w2]*Exp[(-(w1^2 + w2^2)/2)*sigma^2]

In[211]:= With[{cvec = {-1, 1}, xvec = {{1, 1/2}, {1/2, 1}}}, 
  Conjugate[cvec] . Outer[\[Phi][#1 - #2] & , xvec, xvec, 1] . cvec]]

Out[211]= -(1/2)/E^(sigma^2/4)
share|cite|improve this answer
Thank you. I wanted to make a pair of r.vs that are marginally Normal but not jointly Normal. I know there are standard examples, but I wanted to fabricate one through char functions. Anyway the positive definite condition in Bochners theorem is one hard nut to satisfy :( – Isomorphism Sep 29 '11 at 5:36
@Isomorphism, try $\varphi_{X_1,X_2}(x_1,x_2)=\mathrm e^{-(x_1^2+x_2^2)/2}\cosh(x_1x_2)$. – Did Sep 29 '11 at 5:59
@DidierPiau: Thanks! I was looking for that :) – Isomorphism Sep 29 '11 at 19:28

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.