As “BilevelJuMP is not able to solve problems with quadratic constraints in the lower level.” Is there a recommended way to solve bilevel optimizations where the lower level has quadratic constraints and the upper level constraints are convex? (I am all that not familiar with this stuff).
Are you asking in general, or for BilevelJuMP.jl in particular?
This is a question for @joaquimg.
Whichever is simpler to use with 15 lower vars and 9 upper vars, where some vars or lowe var expressions appear in the upper level. Or both.
It seems like BilevelJuMP doesn’t support quadratic constraints in the lower level, but it does support second-order cones.
https://joaquimg.github.io/BilevelJuMP.jl/stable/tutorials/conic_lower/
So reformulate your quadratic constraint into an SOC.
If you get stuck, post a reproducible example, and we can help.
1 Like