Add more tests for linear_terms

Author: bowenszhu

src/utils.jl

@@ -76,11 +76,11 @@ function linear_terms(exprs::AbstractVector, vars)
         if expr isa Number # just a number, e.g. Int, Float64
             const_terms[i] = expr
         elseif expr in vars # expr is a variables in vars
-            push_sparse_coeff(i, term, 1)
+            push_sparse_coeff(i, expr, 1)
         elseif SymbolicUtils.ismul(expr) && length(expr.dict) == 1
             base, exp = first(expr.dict)
             if base in vars && exp == 1 # a var with a coeff
-                push_sparse_coeff(i, term, expr.coeff)
+                push_sparse_coeff(i, base, expr.coeff)
             else
                 const_nonlinear(i, expr)
             end

test/utils.jl

@@ -1,21 +1,61 @@
 using Test, ModelOrderReduction
 using Symbolics
 
+@variables t w(t) x(t) y(t) z(t)
+
 @testset "linear_terms" begin
-    @variables t w(t) x(t) y(t) z(t)
-    vars = [x, y, z]
-    exprs = [3.0x + 4.5y + 6.0
-             2.0z + 3.4w + 7.0 + sin(x)
-             9.8 + x * (1.0 - y)
-             5.6y + 1.3z^2]
-    A, c, n = ModelOrderReduction.linear_terms(exprs, vars)
-    @test size(A) == (length(exprs), length(vars))
-    @test A == [3.0 4.5 0.0
-                0.0 0.0 2.0
-                0.0 0.0 0.0
-                0.0 5.6 0.0]
-    @test length(c) == length(exprs)
-    @test isequal(c, [6.0, 3.4w + 7.0, 9.8, 0.0])
-    @test length(n) == length(exprs)
-    @test isequal(n, [0.0, sin(x), x * (1.0 - y), 1.3z^2])
+    @testset "linear_terms full" begin
+        vars = [x, y, z]
+        exprs = [3.0x + 4.5y + 6.0
+                 2.0z + 3.4w + 7.0 + sin(x)
+                 9.8 + x * (1.0 - y)
+                 5.6y + 1.3z^2]
+        A, c, n = ModelOrderReduction.linear_terms(exprs, vars)
+        @test size(A) == (length(exprs), length(vars))
+        @test A == [3.0 4.5 0.0
+                    0.0 0.0 2.0
+                    0.0 0.0 0.0
+                    0.0 5.6 0.0]
+        @test length(c) == length(exprs)
+        @test isequal(c, [6.0, 3.4w + 7.0, 9.8, 0.0])
+        @test length(n) == length(exprs)
+        @test isequal(n, [0.0, sin(x), x * (1.0 - y), 1.3z^2])
+    end
+
+    @testset "linear_terms empty exprs" begin
+        vars = [x, y, z]
+        exprs = Vector{Num}(undef, 4)
+        fill!(exprs, false)
+        A, c, n = ModelOrderReduction.linear_terms(exprs, vars)
+        @test size(A) == (length(exprs), length(vars))
+        @test iszero(A)
+        @test length(c) == length(exprs)
+        @test iszero(c)
+        @test length(n) == length(exprs)
+        @test iszero(n)
+    end
+
+    @testset "linear_terms diagonal" begin
+        vars = [x, y, z]
+        exprs = [x, 2y, 3z, 4w]
+        A, c, n = ModelOrderReduction.linear_terms(exprs, vars)
+        @test size(A) == (length(exprs), length(vars))
+        @test A == [1.0 0.0 0.0
+                    0.0 2.0 0.0
+                    0.0 0.0 3.0
+                    0.0 0.0 0.0]
+        @test length(c) == length(exprs)
+        @test isequal(c, [0.0, 0.0, 0.0, 4w])
+        @test length(n) == length(exprs)
+        @test iszero(n)
+    end
+
+    @testset "linear_terms nonunique vars" begin
+        vars = [x, y, y]
+        exprs = [3.0x + 4.5y + 6.0
+                 2.0z + 3.4w + 7.0 + sin(x)
+                 9.8 + x * (1.0 - y)
+                 5.6y + 1.3z^2]
+        @test_throws ArgumentError ModelOrderReduction.linear_terms(exprs, vars)
+    end
 end