Benoît Legat

> We're assuming that the solver implements a file reader for CBF and that it doesn't implement any support for this transition? Does the solver need special support for CBF...

> So one sticking point is that when writing CBF files, we don't have a standard form that maps neatly onto CBF. Their PSDVar's can't be used in integrality, and...

This is part of the reason MosekTools is not using MatrixOfConstraints

I think it can with Mosek, that's what the bar{F}_{ij} matrices are for in https://docs.mosek.com/latest/capi/tutorial-sdo-shared.html}}

This seems possible indeed. That's a limitation of CBF then, I don't think we should do anything special to work around it. If the CBF writer sees PSD variables in...

If the CBF writer decides not to support PSD variables then `supports_add_constrained_variables` should return `false` and the transformation should be done by a functionize bridge so that it's explicit and...

Isn't the first comment of this issue saying that PSD variables are currently ignored ? This means that any code writing a model with PSD variables is currently incorrect so...

I did not try, I was just hoping it could serve as a test

What about now ?

As for the support of nonlinear functions, you could use Convex.jl to rewrite the JuMP nonlinear model once v0.16 is released, see https://github.com/jump-dev/Convex.jl/?tab=readme-ov-file#using-with-jump, I should write an example how to...