StochasticDiffEq.jl
StochasticDiffEq.jl copied to clipboard
SciML
Dynamical integrators
Following discussion in #259 this is a proposed implementation of the ABOBA and OBABO integrators (GJ set of integrators coming later). I have also modified the BAOAB implementation to allow matrix as friction (the gamma parameter) in the equation.
That seems to work, but I am not sure to have done everything properly (in particular to verify is gamma is a matrix or not). So let me know if there are changes to do.
Also, I would like to add the integrator to the documentation, but I am not sure of where I should propose changes.
I think the implementation looks reasonable. @frankschae could you take a look through it as well?
I have added some correction to OBABO that actually require 2 noise processes.
I think the implementation looks pretty good to go? @frankschae what do you think should be done with the problem types here?
I just realized I forgot about these. Let's see what's going on with those convergence tests and get this merged.
I change the convergence tests (and separate also the tests into convergence and interface tests).
I did not update the c1 and c2 as function as @frankschae requested yet, waiting the reply to my comment
Unexpected Pass
Expression: abs(sim1.𝒪est[:weak_final] - 2) < 0.5
Got correct result, please change to @test if no longer broken.
:-D
By the way, did you know a way to reduce the frequency of the log output?
[ Info: Monte Carlo iteration: 90088/100000
The logs start to be very heavy when reaching 100000 iterations
No but I think a PR would be very welcome. You could add a verbose kwarg to
https://github.com/SciML/DiffEqDevTools.jl/blob/5a4439f33ddcc5c1ca561cad125f77e01015af72/src/convergence.jl#L121