search

FluxML / model-zoo

Please do not feed the models
https://fluxml.ai/
Other
915 stars 335 forks source link

Modelling OU process #139

Open doccosmos opened 5 years ago

doccosmos commented 5 years ago

I want to model a time series via an OU process but I'm not clear whether I explicitly declare the OU SDE as the thing to solve with no noise process:

f(u,t) = α*u g(u,t) = β prob = SDEProblem(f,g,...)

or declare the whole thing as an example of an OhrsteinUhlenbeckProcess:

prob = SDEProblem(null,noise=OhrsteinUhlenbeckProcess(...))

?

ChrisRackauckas commented 4 years ago

@doccosmos can you describe a bit more about what you're trying to do? Is it just direct simulation of OU, i.e. https://docs.juliadiffeq.org/latest/features/noise_process/#Direct-Simulation-of-the-Noise-Process-1 ?

@MikeInnes @dhairyagandhi96 this should probably be transferred to StochasticDiffEq.jl

doccosmos commented 4 years ago

I have an experimental time series (t, x, x_err) that I want to model with a OU. What is the best way to do this?

ChrisRackauckas commented 4 years ago

Bridge.jl may be a better fit. DiffEq focused more on the numerical methods for nonlinear functions, while Bridge had the analysis of classic systems.