Inconcistent variable count In lazy bridge optimizer

[matbesancon]

This PR: #2695 broke FrankWolfe.jl's tests by making NumberOfVariables and ListOfVariableIndices inconsistent.

julia> length(MOI.get(lmo_moi.o, MOI.ListOfVariableIndices())), MOI.get(lmo_moi.o, MOI.NumberOfVariables())
(300, 301)

# underlying model is okay
julia> length(MOI.get(lmo_moi.o.model, MOI.ListOfVariableIndices())), MOI.get(lmo_moi.o.model, MOI.NumberOfVariables())
(301, 301)

I don't have a minimum reproducible example yet, but it seems to happen with a first group of variables created, and then one last add_constrained_variable

[matbesancon]

Confirmed that the issue is caused in the changes of the latest patch release

[blegat]

Does calling MOI.Utilities.final_touch(lmo_moi) changes anything ?

[matbesancon]

I don't have access to an index_map, final_touch takes two arguments

[odow]

This PR: #2695 broke FrankWolfe.jl's tests

Oops. I'll add it to solver-tests.yml.

[odow]

Here's a smaller example:

julia> import Hypatia

julia> import MathOptInterface as MOI

julia> import FrankWolfe

julia> import SCS

julia> function main(optimizer)
           model = MOI.instantiate(optimizer; with_bridge_type = Float64)
           lmo_moi = FrankWolfe.convert_mathopt(
               FrankWolfe.NuclearNormLMO(10.0),
               model;
               row_dimension = 15, col_dimension = 5, use_modify = false,
           )
           n = MOI.get(lmo_moi.o, MOI.NumberOfVariables())
           length_n = length(MOI.get(lmo_moi.o, MOI.ListOfVariableIndices()))
           return n == length_n
       end
main (generic function with 1 method)

julia> main(Hypatia.Optimizer)
false

julia> main(SCS.Optimizer)
true

[odow]

Excluding FrankWolfe:

julia> import Hypatia

julia> import MathOptInterface as MOI

julia> model = MOI.instantiate(Hypatia.Optimizer; with_bridge_type = Float64)
MOIB.LazyBridgeOptimizer{MOIU.CachingOptimizer{Hypatia.Optimizer{Float64}, MOIU.UniversalFallback{MOIU.Model{Float64}}}}
├ Variable bridges: none
├ Constraint bridges: none
├ Objective bridges: none
└ model: MOIU.CachingOptimizer
  ├ state: EMPTY_OPTIMIZER
  ├ mode: AUTOMATIC
  ├ model_cache: MOIU.UniversalFallback{MOIU.Model{Float64}}
  │ ├ ObjectiveSense: FEASIBILITY_SENSE
  │ ├ ObjectiveFunctionType: MOI.ScalarAffineFunction{Float64}
  │ ├ NumberOfVariables: 0
  │ └ NumberOfConstraints: 0
  └ optimizer: Hypatia.Optimizer{Float64}
    ├ ObjectiveSense: unknown
    ├ ObjectiveFunctionType: unknown
    ├ NumberOfVariables: unknown
    └ NumberOfConstraints: unknown

julia> x = MOI.add_variable(model)
MOI.VariableIndex(1)

julia> t, _ = MOI.add_constrained_variable(model, MOI.LessThan(10.0))
(MOI.VariableIndex(-1), MathOptInterface.ConstraintIndex{MathOptInterface.VariableIndex, MathOptInterface.LessThan{Float64}}(-1))

julia> MOI.get(model, MOI.NumberOfVariables())
2

julia> MOI.get(model, MOI.ListOfVariableIndices())
1-element Vector{MathOptInterface.VariableIndex}:
 MOI.VariableIndex(1)