Tiny fix of Coevolve() with the dependency graph check

[palmtree2013]

this pull request is related to #276 (comment)

[ChrisRackauckas]

Can you add a test which catches this? I guess it should be a case with mixed jumps?

[palmtree2013]

Here is a MWE

using JumpProcesses

maj_rate = [1.]
react_stoich_ = [Vector{Pair{Int, Int}}()]
net_stoich_ = [[1 => 1]]
mass_action_jump_ = MassActionJump(maj_rate, react_stoich_, net_stoich_; scale_rates=false)

affect! = function (integrator)
    integrator.u[1] -= 1
end
cs_rate1(u,p,t) = 0.2 * u[1]
constant_rate_jump = ConstantRateJump(cs_rate1, affect!)
jumpset_ = JumpSet((), (constant_rate_jump,), nothing, mass_action_jump_)

u0 = [0]
tspan = (0.0, 30.0)
prob_ = DiscreteProblem(u0, tspan)
alg = Coevolve()

jprob_ = JumpProblem(dprob_, alg, jumpset_, save_positions=(false, false))

where the JumpSet has one constant rate jump and one mass action jump, but the dependency graph is not provided. Thus, I suppose Coevolve() should show error with

JumpProcesses.jl/src/aggregators/coevolve.jl

Line 31 in 2afd123

error("To use Coevolve a dependency graph between jumps must be supplied.")

But instead, the dependency graph is generated by

JumpProcesses.jl/src/aggregators/coevolve.jl

Line 33 in 2afd123

dg = make_dependency_graph(length(u), maj)

But Coevolve() will still show error because the length of dependency graph does not fit the total reaction number

JumpProcesses.jl/src/aggregators/coevolve.jl

Line 45 in 2afd123

error("Number of nodes in the dependency graph must be the same as the number of jumps.")

So at the end, one still needs to provide a correct dependency graph.

[ChrisRackauckas]

Can you add that MWE as a test?

[ChrisRackauckas]

Use Catalyst to generate the problem so it has the dependency graph? Or use one of the hand-built ones and extend it?

[isaacsas]

I just added a test for this based on the MWE.

[isaacsas]

GTG if tests pass I think.

[isaacsas]

Thanks @palmtree2013

[palmtree2013]

Thank you for adding the test