Using SDPA-sparse format

The function sdpa_sparse_to_problem can be used to read an SDPA-sparse file and create a Problem. In particular, this makes it possible to use JuMP to create a semidefinite program and write it to the SDPA-sparse format, which can then be read and solved with ClusteredLowRankSolver. The JuMP interface is not yet supported for direct conversion to a Problem.

ClusteredLowRankSolver.sdpa_sparse_to_problem — Function

sdpa_sparse_to_problem(filename,obj_shift = 0; T=Float64)

Define the Problem from the file filename assuming it is in SDPA sparse format, using the number type T. Optionally add an objective shift.

source

Since the default formulation in JuMP is different from the formulation in ClusteredLowRankSolver, it is recommended to use the package Dualization.jl before writing the problem to a file in JuMP.

using Dualization

model = Model()
% code to make the model

write_to_file(dualize(model), "sdp.dat-s")