Bug in vectorized constraints with SparseAxisArrays

[hdavid16]

I think the following should work (but is not currently supported):

@variable(model, u[i = 1:2, j = i:3])
Containers.@container(c[i = 1:2, j = i:3],rand())
@constraint(model, u .<= c)

ERROR: MethodError: no method matching _indices()
Closest candidates are:
  _indices(::JuMP.Containers.SparseAxisArray, ::Any...) at C:\Users\HD\.julia\packages\JuMP\lnUbA\src\Containers\SparseAxisArray.jl:217
  _indices(::Any, ::Any...) at C:\Users\HD\.julia\packages\JuMP\lnUbA\src\Containers\SparseAxisArray.jl:215

If the indices match, shouldn't this be allowed?

[odow]

I think it's intended that this works, because broadcasting two SparseAxisArrays over a function works. I'd have to dig in further to see what the issue is:

julia> using JuMP

julia> model = Model()
A JuMP Model
Feasibility problem with:
Variables: 0
Model mode: AUTOMATIC
CachingOptimizer state: NO_OPTIMIZER
Solver name: No optimizer attached.

julia> @variable(model, u[i = 1:2, j = i:3])
JuMP.Containers.SparseAxisArray{VariableRef, 2, Tuple{Int64, Int64}} with 5 entries:
  [1, 1]  =  u[1,1]
  [1, 2]  =  u[1,2]
  [1, 3]  =  u[1,3]
  [2, 2]  =  u[2,2]
  [2, 3]  =  u[2,3]

julia> Containers.@container(c[i = 1:2, j = i:3],rand())
JuMP.Containers.SparseAxisArray{Float64, 2, Tuple{Int64, Int64}} with 5 entries:
  [1, 1]  =  0.527184
  [1, 2]  =  0.557751
  [1, 3]  =  0.118786
  [2, 2]  =  0.683632
  [2, 3]  =  0.828271

julia> @constraint(model, u .<= c)
ERROR: MethodError: no method matching _indices()
Closest candidates are:
  _indices(::JuMP.Containers.SparseAxisArray, ::Any...) at /Users/oscar/.julia/dev/JuMP/src/Containers/SparseAxisArray.jl:225
  _indices(::Any, ::Any...) at /Users/oscar/.julia/dev/JuMP/src/Containers/SparseAxisArray.jl:223
Stacktrace:
 [1] _indices(::Base.RefValue{String}) (repeats 3 times)
   @ JuMP.Containers ~/.julia/dev/JuMP/src/Containers/SparseAxisArray.jl:223
 [2] copy(bc::Base.Broadcast.Broadcasted{JuMP.Containers.BroadcastStyle{2, Tuple{Int64, Int64}}, Nothing, typeof(add_constraint), Tuple{Base.RefValue{Model}, Base.Broadcast.Broadcasted{JuMP.Containers.BroadcastStyle{2, Tuple{Int64, Int64}}, Nothing, typeof(build_constraint), Tuple{Base.RefValue{JuMP.var"#_error#85"{Tuple{Symbol, Expr}, Symbol, LineNumberNode}}, JuMP.Containers.SparseAxisArray{AffExpr, 2, Tuple{Int64, Int64}}, Base.RefValue{MathOptInterface.LessThan{Float64}}}}, Base.RefValue{String}}})
   @ JuMP.Containers ~/.julia/dev/JuMP/src/Containers/SparseAxisArray.jl:182
 [3] materialize(bc::Base.Broadcast.Broadcasted{JuMP.Containers.BroadcastStyle{2, Tuple{Int64, Int64}}, Nothing, typeof(add_constraint), Tuple{Base.RefValue{Model}, Base.Broadcast.Broadcasted{JuMP.Containers.BroadcastStyle{2, Tuple{Int64, Int64}}, Nothing, typeof(build_constraint), Tuple{Base.RefValue{JuMP.var"#_error#85"{Tuple{Symbol, Expr}, Symbol, LineNumberNode}}, JuMP.Containers.SparseAxisArray{AffExpr, 2, Tuple{Int64, Int64}}, Base.RefValue{MathOptInterface.LessThan{Float64}}}}, Base.RefValue{String}}})
   @ Base.Broadcast ./broadcast.jl:883
 [4] macro expansion
   @ ~/.julia/dev/JuMP/src/macros.jl:816 [inlined]
 [5] top-level scope
   @ REPL[5]:1

julia> foo(a, b) = a * b
foo (generic function with 1 method)

julia> foo.(u, c)
JuMP.Containers.SparseAxisArray{AffExpr, 2, Tuple{Int64, Int64}} with 5 entries:
  [1, 1]  =  0.527184280519112 u[1,1]
  [1, 2]  =  0.5577508885495441 u[1,2]
  [1, 3]  =  0.11878560090114121 u[1,3]
  [2, 2]  =  0.6836319966831192 u[2,2]
  [2, 3]  =  0.8282711576813722 u[2,3]