MathOptChordalDecomposition.jl is a MathOptInterface.jl layer that implements chordal decomposition of sparse semidefinite constraints.
If you need help, please ask a question on the JuMP community forum.
If you have a reproducible example of a bug, please open a GitHub issue.
MathOptChordalDecomposition.jl is licensed under the
MIT License.
Install MathOptChordalDecomposition as follows:
import Pkg
Pkg.add("MathOptChordalDecomposition")To use MathOptChordalDecomposition with JuMP, use MathOptChordalDecomposition.Optimizer:
using JuMP, MathOptChordalDecomposition, SCS
model = Model(() -> MathOptChordalDecomposition.Optimizer(SCS.Optimizer))Change SCS for any other conic solver that supports semidefinite constraints.
The sdplib directory contains four semidefinite programming problems from the
SDPLIB library.
The function main, defined below, reads one of the problems and
constructs a JuMP.jl model.
For this example, it is significantly faster to solve the problem with MathOptChordalDecomposition than to use SCS by itself:
julia> using FileIO, JLD2, JuMP, LinearAlgebra, SCS
julia> import MathOptChordalDecomposition as MOCD
julia> function main(optimizer, name::String)
data = FileIO.load("./sdplib/$name.jld2");
F, c, m, n = data["F"], data["c"], data["m"], data["n"]
model = Model(optimizer)
set_silent(model)
@variable(model, x[1:m])
@objective(model, Min, c' * x)
@constraint(
model,
con,
LinearAlgebra.Symmetric(-F[1] + x' * F[2:end]) in PSDCone(),
)
optimize!(model)
return objective_value(model)
end
main (generic function with 1 method)
julia> @time main(SCS.Optimizer, "mcp124-1")
9.447474 seconds (154.70 k allocations: 12.313 MiB)
141.96561765120785
julia> @time main(() -> MOCD.Optimizer(SCS.Optimizer), "mcp124-1")
0.245992 seconds (170.72 k allocations: 15.103 MiB, 1.85% compilation time)
141.9887372030578