JuMP problem get conflicting constraints when using small values

Specific Domains Optimization (Mathematical)
, ,
2

It seems that the problem is not with small values, but with the fact that the equality constraint is quadratic. This works:

using JuMP
using Gurobi

model = Model(Gurobi.Optimizer)
@variable(model, x)
a = 5e-5
b = 3e-3

@constraint(model, c1, x == 0)
@constraint(model, c2, 1 == (a * x +b*x))

@objective(model, Min, x)

optimize!(model)

compute_conflict!(model)

MOI.get.(model, MOI.ConstraintConflictStatus(), c1)

MOI.get.(model, MOI.ConstraintConflictStatus(), c2)

Gives:
image

Can Gurobi always support quadratic equality constraints? The example works with inequality:

using JuMP
using Gurobi

model = Model(Gurobi.Optimizer)
@variable(model, x)
a = 5e-5
b = 3e-3

@constraint(model, c1, x == 0)
@constraint(model, c2, 1 <= (a * x^2 +b*x))

@objective(model, Min, x)

optimize!(model)

compute_conflict!(model)

# MOI.get.(model, MOI.ConstraintConflictStatus(), c1)

MOI.get.(model, MOI.ConstraintConflictStatus(), c2)
# # Without broadcasting
MOI.get.(model, MOI.ConstraintConflictStatus(), c2)

and gives:

image
image547×510 10.3 KB

I am unfortunately not familiar with Gurobi enough to offer more help.