Backward SDE StochasticDiffEq

Hello,

I am interested in solving backward SDEs as in these papers (https://arxiv.org/pdf/2011.13456.pdf, https://papers.nips.cc/paper/2021/file/0a9fdbb17feb6ccb7ec405cfb85222c4-Paper.pdf). They talk about standard Wiener process in the reverse-time direction. Is it sufficient to flip the integration time span in StochasticDiffEq?

using StochasticDiffEq

f(u,p,t) = -0.5*u
g(u,p,t) = 1.0

tf = 10.0
tspan = (0.0, tf)
prob = SDEProblem(f, g, zeros(Nx), tspan)
reverse_prob = SDEProblem(f, g, zeros(Nx), reverse(tspan))

Yes.

1 Like

Thank you Chris.