Ref JuliaOpt/MathProgBase.jl#91 and JuliaOpt/AmplNLWriter.jl#24, using the same wrapper approach as AmplNLWriter.
The two Linux failures are this error from #6 when using Couenne.
The failures on Mac are the same MIQP vectorized problem, but due to tolerance issues with the optimal solution, again when using Couenne.
It would be possible to make the tests pass by removing Couenne from quad_mip_solvers, if that was something we wanted to do, but otherwise everything seems fine.
Ref JuliaOpt/MathProgBase.jl#91 and JuliaOpt/AmplNLWriter.jl#24, using the same wrapper approach as AmplNLWriter.
The two Linux failures are this error from #6 when using Couenne.
The failures on Mac are the same MIQP vectorized problem, but due to tolerance issues with the optimal solution, again when using Couenne.
It would be possible to make the tests pass by removing Couenne from
quad_mip_solvers
, if that was something we wanted to do, but otherwise everything seems fine.