Merge pull request #4 from focusenergy/dot

Make dot work

src/LinearFractional.jl

@@ -39,9 +39,8 @@ include("parseexpr.jl")
 getvalue(x::LinearFractionalVariable) = getvalue(x.var)/getvalue(x.model.t)
 getvalue(aff::LinearFractionalAffExpr) = getvalue(aff.afftrans)/getvalue(aff.t)
 
-# Need to define Base.one to get dot() for free, but I don't think we can
-# do this without knowing the model (to get t)
-# Base.one(::Type{LinearFractionalVariable}) = LinearFractionalAffExpr(Variable[],Float64[],1.0)
+Base.one(::Type{LinearFractionalVariable}) = 1.0
+Base.one(::LinearFractionalVariable) = one(LinearFractionalVariable)
 Base.sum(xs::Array{LinearFractionalAffExpr}) = LinearFractionalAffExpr(sum(x.afftrans for x in xs), xs[1].t)
 function Base.sum(xs::Array{LinearFractionalAffExpr}, dim)
     s = sum(collect(x.afftrans for x in xs), dim)

test/arrayvars.jl

@@ -87,3 +87,18 @@ end
     @test getvalue(xm) ≈ getvalue(x)
     @test getobjectivevalue(lfp) ≈ getobjectivevalue(m)/5.0
 end
+
+
+@testset "Misc array features" begin
+    ## dot should work
+    lfp = LinearFractionalModel(solver=ClpSolver())
+    a = [-2, 0]
+    x = @variable(lfp, [i=1:2], basename="x", lowerbound=0)
+    @constraint(lfp, x[2] <= 6)
+    @constraint(lfp, -x[1] + x[2] <= 4)
+    @constraint(lfp, 2x[1] + x[2] <= 14)
+    @numerator(lfp,  :Min, dot(a, x) + 2)
+    @denominator(lfp,  x[1] + 3x[2] + 4)
+    @test solve(lfp) == :Optimal
+    @test all(getvalue(x) .≈ [7.0, 0.0])
+end

test/runtests.jl

@@ -1,5 +1,6 @@
 using Test
 using LinearFractional
+using LinearAlgebra
 using Clp
 using JuMP