# Variable rate jumps diffeq?

I’m trying to get a variable rate jump to work. I am using

``````
## jump in price
#rate1(u,p,t) = λ0 # constant jump rate
rate1(u,p,t) = λ0.*exp(u) # volatility dependent jump rate
# jump is normal with st. dev. equal to λ1 times current st. dev.
affect1!(integrator) = (integrator.u = integrator.u.+randn(size(integrator.u)).*λ1.*exp(integrator.u./2.0))

# this works:
jump1 = ConstantRateJump(rate1,affect1!)
# this does not
#jump1 = VariableRateJump(rate1,affect1!)
jump_prob = JumpProblem(prob,Direct(), jump1)

``````

The rate, rate1, depends on the second integrator, so I understand that it is a variable rate jump. However, I get a dimension error if I try to specify it as variable rate (jump1, commented out). The solver gives a solution if I specify it as constant rate.

Am I making an error in the syntax for a variable rate jump?
Thanks!

Here’s the environment, and the whole file, if it helps

``````(JD) pkg> st
Status `~/Desktop/JD/Project.toml`
[a077e3f3] DiffEqProblemLibrary v4.8.0
[0c46a032] DifferentialEquations v6.14.0 `https://github.com/SciML/DifferentialEquations.jl.git#master`
[91a5bcdd] Plots v0.29.9
[e6cf234a] RandomNumbers v1.4.0

(JD) pkg>

``````
``````using DifferentialEquations, Plots

function MyProblem(μ0,μ1,κ,α,σ,ρ,u0,tspan)
f = function (du,u,p,t)
du = μ0 + μ1*(u-α)/σ # drift in prices
du = κ*(α-u) # mean reversion in shocks
end
g = function (du,u,p,t)
du = exp(u/2.0)
du = σ
end
Γ = [1 ρ;ρ 1] # Covariance Matrix
noise = CorrelatedWienerProcess!(Γ,tspan,zeros(2),zeros(2))
sde_f = SDEFunction{true}(f,g)
SDEProblem(sde_f,g,u0,tspan,noise=noise)
end

μ0 = 0.0
μ1 = 0.0
κ = 0.05
α = 0.2
σ = 0.2
ρ = -0.7
λ0 = 2.0
λ1 = 3.0
u0 = [0;α]
dt = 0.01
prob = MyProblem(μ0, μ1, κ, α, σ, ρ, u0, (0.0,1000.0))

## jump in price
#rate1(u,p,t) = λ0 # constant jump rate
rate1(u,p,t) = λ0.*exp(u) # volatility dependent jump rate
# jump is normal with st. dev. equal to λ1 times current st. dev.
affect1!(integrator) = (integrator.u = integrator.u.+randn(size(integrator.u)).*λ1.*exp(integrator.u./2.0))

# this works:
jump1 = ConstantRateJump(rate1,affect1!)
# this does not
#jump1 = VariableRateJump(rate1,affect1!)
jump_prob = JumpProblem(prob,Direct(), jump1)

``````

An incidental question, is there a way to visualize exactly when jumps occur?

Xref: Specify jumps in a Heston-like model? for the solution

1 Like

Hi @mcreel

I ran into the same problem today. The example from the tutorial

works fine for the constant rate jump, but not for the variable rate jump

The error message I’m getting from the variable rate jump is

``````MethodError: no method matching CartesianIndices(::Array{Float64,1}, ::Int64, ::Int64)
``````

BTW I updated to the latest version of DifferentialEquations.jl and the problem still exists

I just ran the tutorial code:

``````using OrdinaryDiffEq, DiffEqJump
function f(du,u,p,t)
du = u
end

prob = ODEProblem(f,[0.2],(0.0,10.0))
rate(u,p,t) = u
affect!(integrator) = (integrator.u = integrator.u/2)
jump = VariableRateJump(rate,affect!)
jump2 = deepcopy(jump)
jump_prob = JumpProblem(prob,Direct(),jump,jump2)
sol = solve(jump_prob,Tsit5())
``````

It ran fine. What are you running?

Hi @tzxiong. Please see ChrisRackaukas’ answer in the other thread. The variable rate problem was more complicated than what I was interested in getting into, so I ended up working with constant rate jumps, and that all worked out well.

1 Like