seed given by solver keyword argument does not work for sparse noise_rate_prototype

[sho7014]

Hello,

seed argument in solve() does not seem to work when noise_rate_prototype is specified using sparse arrays. Instead. seed can be appropriately set using Random.seed!().

You can see this behavior using the following code.

using StochasticDiffEq, PyPlot, SparseArrays, Random

function f!(du,u,p,t)
    du[1] = 0.0
end

function g!(du,u,p,t)
    du[1,1] = 1.0
    du[1,3] = 1.0
end

dt = 1e-2
u0 = [0.0]
tspan = (0.0,0.1)

#seed does not work
sde_prob = SDEProblem(f!, g!, u0, tspan, noise_rate_prototype=sparse([1.0 0.0 1.0]))
sol = solve(sde_prob, EulerHeun(), dt=dt, seed=1)

#seed works
sde_prob = SDEProblem(f!, g!, u0, tspan, noise_rate_prototype=[1.0 0.0 1.0])
sol = solve(sde_prob, EulerHeun(), dt=dt, seed=1)

#seed works
sde_prob = SDEProblem(f!, g!, u0, tspan, noise_rate_prototype=sparse([1.0 0.0 1.0]))
Random.seed!(1)
sol = solve(sde_prob, EulerHeun(), dt=dt)