In the 60's, I. Gel'fand introduced the concept of generalized stochastic processes (Ch. III, Vol. 4 of his work on Generalized functions). For a generalized stochastic process $\Phi$, he defines the concepts of stationarity ($\Phi(\varphi)$ and $\Phi(\varphi(\cdot - t_0))$ have the same law) and of independence at every point (the random variable $\Phi(\varphi_1)$ and $\Phi(\varphi_2)$ are independent if $\varphi_1$ and $\varphi_2$ have disjoint supports).

Gel'fand especially introduces the complete class of Lévy white noises as generalized stochastic processes with characteristic functional of the form $$L(\varphi) =\exp\left( \int f(\varphi(t)) \mathrm{d}t \right),$$ with $f$ a function that has a L\'evy-Khintchine representation.

Obviously, white noises are not the only stationary and independent at every point processes (ex: the weak derivative of a white noise). I am interested by a characterization of stationary and independent at every point processes. Especially, Gel'fand conjectured the following result.

Conjecture: If a generalized stochastic process $\Phi$ is stationary and independent at every point, then the characteristic function $L$ of $\Phi$ has the form $$L(\varphi) =\exp\left( \int f(\varphi(t),\varphi^{(1)}(t),\cdots, \varphi^{(n)}(t)) \mathrm{d}t \right),$$ with $f$ a continuous function from $\mathbb{R}^{n+1}$ to $\mathbb{C}$ with $f(0)=0$.

Is that result true? In order to express a kind of reciprocal result, can we characterize the functions $f$ such that the previous functional is a characteristic functional (main problem: its positive-definiteness)?

Some people are extensively studied the positive-definiteness of functionals but wasn't able to find references that are clearly answering my question.

Thanks for your attention.


1 Answer 1


I would suggest the reference [1], where J.N. Pandey addresses this problem.

[1] J.N. Pandey, ``On the positive definiteness of a functional'', Canad. Math. Bull., Vol 22 (2), 1979

  • $\begingroup$ This note is clearly related with my second question (condition of positive-definiteness), but doesn't say anything on the conjecture. Indeed, Pandey is focusing on the case we already know that $L(\varphi)$ has the form expressed in my conjecture. He shows that a sufficient condition on $f$ expressed by Gel'fand is not necessary (with a counter-example) and gives himself a necessary (but not sufficient) condition. At least, it shows that one conjecture of Gel'fand was false! Thank you. $\endgroup$
    – Goulifet
    Mar 7, 2014 at 16:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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