As the title says, if we take a time slice on any Ito diffusion - are we guaranteed that the data is always Normally distributed?

enter image description here

  • This seems like a useful property for computer generalization and when the distribution of the forward propagation is hard to solve.

Note: My question does not include fractal diffusion.

  • $\begingroup$ This seems trivially false since Ito diffusions are closed under "nice" functions (like the conditions of Ito's lemma, say). So if $X_t$ is a Brownian motion (therefore an Ito process), then $Y_t = e^{X_t}$ is also an Ito process, and has lognormal distribution at a fixed time. Was there a particular reason you expected this to be always normal? $\endgroup$ – Chris Haug Jun 1 '18 at 12:10
  • $\begingroup$ @ChrisHaug Thank you - that makes sense. The reason I ask is because I am writing a web-app and instead of transferring all the path data from one point to another - I wanted to return an array with mean and sd for each time index - which would be much smaller in size. $\endgroup$ – Edv Beq Jun 1 '18 at 12:59
  • $\begingroup$ @ChrisHaug A numerical schema for the forward kolmogorov would be helpful too if one exists. $\endgroup$ – Edv Beq Jun 1 '18 at 13:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.