Lazy constraint and user cut


In the past, “lazy constraints” and “user cuts” could only be used for mixed-integer LINEAR problems. In the recent AMPL documentation, I see that the “lazy” suffix is said to be usable for MIP problems.
I would like to know if the “lazy” suffix can now be used for mixed-integer NONLINEAR problems, such as MIQP or MISCOCP, and, if so, with which solvers this can be done.
Thank you in advance.

The .lazy suffix is recognized by Gurobi and by CPLEX, to specify lazy constraints and user cuts. You can use .lazy in any model, but .lazy values are only recognized for a model’s linear constraints. (If the .lazy suffix is defined on other kinds of constraints, the solver ignores it.)