ComponentArray and generation of JumpSet

[paolo-mgi]

Hello,

I wish to integrate a system of jump differential equations with JumpProblem. Variables in the system behave differently and I discrimiate between them using ComponentArray. The problem requires the definition of multiple jumps. I encounter the following problem. If I define the affect! functions manually one-by-one the code works as expected. If instead I create them with a function I get the error type NamedTuple has no field u. I spent some time inquiring the error but to no avail. I would be very interested to know what I am doing wrong.
Many thanks for your attention.

Below the two codes. The first works as expected

   du.st_vec .= 0.0
   du.mpr = 6.0.*c(t)*u.mpr/4.0
   du.jpr = 0.0 
end
sse_prob = ODEProblem(psi!, u0, (0.0, tf),p)
function rate!(u,p,t)
   return c(t)/4.0 
end 
# "Manual" definition of the affect functions
function affect1!(integrator)
      v = copyto!(integrator.p.tmp_psi, integrator.u.u.st_vec) # copy this to avoid aliasing later
      mart =  integrator.u.u.mpr 
      jump =  integrator.u.u.jpr 
      jump_prob = real(dot(v,v))
      if jump_prob > 0.0
         normv = sqrt(jump_prob)
      else
         normv=1.0
      end
      mul!(integrator.u.u.st_vec, integrator.p.jump_ops[1], v, 1/normv,false)
      integrator.u.u.mpr = -2.0*mart 
      integrator.u.u.jpr = -2.0*jump 
      nothing
    end

function affect2!(integrator)
      v = copyto!(integrator.p.tmp_psi, integrator.u.u.st_vec) # copy this to avoid aliasing later
      mart =  integrator.u.u.mpr 
      jump =  integrator.u.u.jpr 
      jump_prob = real(dot(v,v))
      if jump_prob > 0.0
         normv = sqrt(jump_prob)
      else
         normv=1.0
      end
      mul!(integrator.u.u.st_vec, integrator.p.jump_ops[2], v, 1/normv,false)
      integrator.u.u.mpr = -2.0*mart 
      integrator.u.u.jpr = -2.0*jump 
      nothing
    end

function affect3!(integrator)
      v = copyto!(integrator.p.tmp_psi, integrator.u.u.st_vec) # copy this to avoid aliasing later
      mart =  integrator.u.u.mpr 
      jump =  integrator.u.u.jpr 
      jump_prob = real(dot(v,v))
      if jump_prob > 0.0
         normv = sqrt(jump_prob)
      else
         normv=1.0
      end
      mul!(integrator.u.u.st_vec, integrator.p.jump_ops[3], v, 1/normv,false)
      integrator.u.u.mpr = -2.0*mart 
      integrator.u.u.jpr = -2.0*jump 
      nothing
    end
# Construction of the JumpSet
jump1=VariableRateJump(rate!, affect1!)
jump2=VariableRateJump(rate!, affect2!)
jump3=VariableRateJump(rate!, affect3!)
jump_prob = JumpProblem(sse_prob, DirectFW(), jump1,jump2,jump3 );
ensemble_sse = EnsembleProblem(jump_prob)
jump_sol_sse = DifferentialEquations.solve(ensemble_sse, Tsit5(), EnsembleSerial(), trajectories=num_traj; save_everystep=true, dense=false,abstol=1e-8,reltol=1e-8,maxiters=Int(1e6));

As said the above code gives the expected result.
If I now replace the second part of the above with

function create_jump_affect(affected_id::Int) 
    function jump_affect_by_id!(integrator)
      v = copyto!(integrator.p.tmp_psi, integrator.u.u.st_vec) # copy this to avoid aliasing later
      mart =  integrator.u.u.mpr 
      jump =  integrator.u.u.jpr 
      jump_prob = real(dot(v,v))
      if jump_prob > 0.0
         normv = sqrt(jump_prob)
      else
         normv=1.0
      end
      mul!(integrator.u.u.st_vec, integrator.p.jump_ops[affected_id], v, 1/normv,false)
      integrator.u.u.mpr = -2.0*mart 
      integrator.u.u.jpr = -2.0*jump 
      nothing
    end
    return jump_affect_by_id!
end
# Generate individual jumps by putting them in an array
jumps = VariableRateJump[]   
for i in 1:3
    push!(jumps,VariableRateJump(rate!, create_jump_affect(i)))
end
# Create a JumpSet with the provided constant rate jumps and empty/nil for other jump types
function create_jump_set(variable_jumps::Vector{VariableRateJump})
    return JumpSet((), tuple(variable_jumps...), nothing, nothing)
end
# JumpSet
jump_prob = JumpProblem(sse_prob, DirectFW(), jumpset )
ensemble_sse = EnsembleProblem(jump_prob)
jump_sol_sse = DifferentialEquations.solve(ensemble_sse, Tsit5(), EnsembleSerial(), trajectories=num_traj; save_everystep=true, dense=false,abstol=1e-8,reltol=1e-8,maxiters=Int(1e6));

I get the error

type NamedTuple has no field u

Many thanks for your patience!!