ERROR: MethodError: no method matching view(::MTKParameters) while solving an optimization problem via MOI #3637
Description
I am trying to solve a constrained nonlinear optimization problem (optimal power flow) using ModelingToolkit.jl. For compatibility with the other part of the project, I need to implement the problem using MTK rather than Optimization.jl or JuMP.
While I am able to construct the optimization system, instantiate the particular problem I want to solve, and pass it to the MOI solver, I am getting an error concerning an internal view function:
ERROR: MethodError: no method matching view(::MTKParameters{Vector{Float64}, Tuple{}, Tuple{}, Tuple{}}, ::UnitRange{Int64})
The function `view` exists, but no method is defined for this combination of argument types.
An example is provided below:
using ModelingToolkit, Optimization, OptimizationMOI, Ipopt
# system matrix
Y = [1.9802-19.802im -0.990099+9.90099im -0.990099+9.90099im
-0.990099+9.90099im 1.9802-19.802im -0.990099+9.90099im
-0.990099+9.90099im -0.990099+9.90099im 1.9802-19.802im]
n = size(Y, 1)
@variables begin
v[1:n], [bounds = (0.95, 1.05)]
θ[1:n], [bounds = (-π, π)]
p_g[1:n], [bounds = (0.0, 3.0)]
q_g[1:n], [bounds = (-2.0, 2.0)]
end
@parameters begin
p_d[1:n] = [0.0, 0.0, 0.8]
q_d[1:n] = [0.0, 0.0, 0.6]
c[1:n] = [1.0, 2.0, 0.0]
end
# constraints
cons = []
# power flow constraints
for i=1:n
# active power balance
push!(cons, p_g[i] - p_d[i] ~ sum(v[i]*v[j]*abs(Y[i, j]) * cos(angle(Y[i, j]) - θ[i] + θ[j]) for j=1:n))
# reactive power balance
push!(cons, q_g[i] - q_d[i] ~ -sum(v[i]*v[j]*abs(Y[i, j]) * sin(angle(Y[i, j]) - θ[i] + θ[j]) for j=1:n))
end
# slack bus and load bus constraints
push!(cons, θ[3] ~ 0.0);
push!(cons, p_g[3] ~ 0.0);
push!(cons, q_g[3] ~ 0.0);
# cost function
cost_fcn = sum(c[i] * p_g[i] for i=1:n)
@named sys = OptimizationSystem(cost_fcn, [v, θ, p_g, q_g], [p_d, q_d, c], constraints=cons)
sys = complete(sys)
x0 = [
v => [1.0, 1.0, 1.0]
θ => [0.1, 0.1, 0.1]
p_g => [0.1, 0.1, 0.8]
q_g => [0.1, 0.1, 0.6]
]
p = [
p_d => [0.0, 0.0, 0.8]
q_d => [0.0, 0.0, 0.6]
c => [1.0, 2.0, 0.0]
]
prob = OptimizationProblem(sys, x0, p, grad=true, hess=true, cons_j=true, cons_h=true)
nlp_solver = OptimizationMOI.MOI.OptimizerWithAttributes(Ipopt.Optimizer,
"print_level" => 0
)
sol = solve(prob, nlp_solver)