According to the Wiener-Hopf factorization, the characteristic function
φ of any probability distribution μ on R can be
decomposed in a unique way as
1−sφ(t)=[1−χ−(s,it)][1−χ+(s,it)],∣s∣≤1,t∈R,
where χ−(eiu,it) and χ+(eiu,it) are the characteristic
functions of possibly defective distributions in
Z+×(−∞,0) and Z+×[0,∞),
respectively.
We prove that μ can be characterized by the sole data of the upward
factor χ+(s,it), s∈[0,1), t∈R in many cases including
the cases where:
- μ has some exponential moments;
- the function t↦μ(t,∞) is completely monotone on
(0,∞);
- the density of μ on [0,∞) admits an analytic continuation on
R.
We conjecture that any probability distribution is actually characterized by
its upward factor. This conjecture is equivalent to the following: {\it Any
probability measure μ on R whose support is not included in
(−∞,0) is determined by its convolution powers μ∗n, n≥1
restricted to [0,∞)}. We show that in many instances, the sole knowledge
of μ and μ∗2 restricted to [0,∞) is actually sufficient to
determine μ. Then we investigate the analogous problem in the framework of
infinitely divisible distributions.