JumpProcesses.jl solvers

General Usage Performance
1

I’m modelling a biochemical system using JumpProcesses.jl. I am coupling a continuous ODE to a jumpset which includes MassActionJumps, ConstantRateJumps and unbounded VariableRateJumps.

When I increase my tspan, or in certain parameter conditions I hit MaxIters. This system is stiff and when I solve this system deterministically I use Rodas4. I can’t see options to use anything other than Tsit5 when solving a jump problem which includes unbounded variable rate jumps, but Tsit5 is a solver for non-stiff problems, so I’m wondering if this is why I’m hitting MaxIters.

Ideally I’d like a solution that avoids having to increase the MaxIters to keep computation time down. Any tips on this would be much appreciated.

2

All of the ODE solvers can be used.

3

Thanks, I thought that would be the case, I get this error when trying to use any other solver than Tsit5 - ERROR: MethodError: no method matching strides(::ExtendedJumpArray{Float64, 1, Vector{Float64}, Vector{Float64}}). Do you have any suggestions as to what could be causing that?

4

It’s a missing dispatch for ExtendedJumpArrays which are what is used with VariableRateJumps. I opened a related issue on JumpProcesses: ODE-jump models with implicit solvers not working · Issue #459 · SciML/JumpProcesses.jl · GitHub, but if you could add a short example that illustrates the error you get that would be helpful to us.

5

Okay yeah that’s a bug but shouldn’t be hard to fix.