Nonlinear subexpressions about jump.jl

[odow]

@common_expression(model, y, f(x))

@variable(model, y)
@constraint(model, y :== f(x))

[Robbybp]

What is stopping @expression(model, y, f(x)) from adopting the common-subexpression behavior you're describing? Would this require breaking changes to MOI?

[odow]

What is stopping @expression(model, y, f(x)) from adopting the common-subexpression behavior you're describing?

Yes, it's a breaking change. We discussed this on the monthly developer call. @mlubin is in favor of looking into a :subexpression node. I need to just try out some options.

[odow]

Some related discussion from jump-dev/MathOptInterface.jl#2488:

Some possible options:

1. Create a single unified expression graph (as opposed to the current expression tree). Change quite a lot of how ReverseAD processes functions etc.
2. Don't do anything. Add a new decision variable with y = f(x) for expressions. This makes things very explicit.
3. Add something new type/index to JuMP/MOI. But we're breaking the abstraction quite a bit.
4. Add a univariate :subexpression operator to MOI. So: @expression(model, expr, subexpression(sin(x) + 1)). Solvers would see this during parsing, and could choose to ignore, or store it in a common expression.

I don't have a strong opinion yet. But our current approach requires us to smack the modeler on the hand and tell them they're holding it wrong and that they should do (2): #3729 (comment)