[Bridges] support ConstraintConflictStatus

[vitorlppf]

An infeasible JuMP model with a vectorized inequality constraint yields an error when trying to copy_conflict (example at the bottom). The error message points towards creating a MathOptInterface bug report, but since the bug arises when using the JuMP interface, it was not clear to me where to report it.

To be specific, the error message is:

ArgumentError: Bridge of type ScalarizeBridge does not support accessing the attribute MathOptInterface.ConstraintConflictStatus(1). If you encountered this error unexpectedly, it probably means your model has been reformulated using the bridge, and you are attempting to query an attribute that we haven't implemented yet for this bridge. Please open an issue at https://github.com/jump-dev/MathOptInterface.jl/issues/new and provide a reproducible example explaining what you were trying to do.

The example below was lightly modified from the one given in the documentation of copy_conflict. Replacing the constraint definition by the commented constraint yields no error.

using JuMP

import Gurobi

model = Model(Gurobi.Optimizer)

@variable(model, x[i=1:1] >= 0)
@constraint(model, c1[i=1:1], x <= [-2])
# @constraint(model, c1[i=1:1], x[i] <= [-2][i])
optimize!(model)
compute_conflict!(model)
if get_attribute(model, MOI.ConflictStatus()) == MOI.CONFLICT_FOUND
    iis_model, reference_map = copy_conflict(model)
    print(iis_model)
end

[odow]

I need to think about this. I hadn't really considered the interaction of an IIS and bridges.

I guess if any of the constraints within a bridge are part of the conflict, then the outer constraint is too.

As a near-term work-around: don't use vectored constraints with Gurobi 😄 try to write a model that uses only the constraint types supported by Gurobi. You can see which bridge are used with print_active_bridges(model).