Merge pull request #240 from JuliaDiffEq/myb/const

Add diff rule on constants

Author: ChrisRackauckas

Project.toml

@@ -1,7 +1,7 @@
 name = "ModelingToolkit"
 uuid = "961ee093-0014-501f-94e3-6117800e7a78"
 authors = ["Chris Rackauckas <[email protected]>"]
-version = "1.2.3"
+version = "1.2.4"
 
 [deps]
 DiffEqBase = "2b5f629d-d688-5b77-993f-72d75c75574e"

src/differentials.jl

@@ -95,6 +95,7 @@ sin(x())
 ```
 """
 derivative(O::Operation, idx) = derivative(O.op, (O.args...,), Val(idx))
+derivative(O::Constant, ::Any) = Constant(0)
 
 # Pre-defined derivatives
 import DiffRules, SpecialFunctions, NaNMath

src/direct.jl

@@ -1,14 +1,14 @@
-function gradient(O::Operation, vars::AbstractVector{Operation}; simplify = true)
+function gradient(O::Expression, vars::AbstractVector{<:Expression}; simplify = true)
     out = [expand_derivatives(Differential(v)(O)) for v in vars]
     simplify ? simplify_constants.(out) : out
 end
 
-function jacobian(ops::AbstractVector{Operation}, vars::AbstractVector{Operation}; simplify = true)
+function jacobian(ops::AbstractVector{<:Expression}, vars::AbstractVector{<:Expression}; simplify = true)
     out = [expand_derivatives(Differential(v)(O)) for O in ops, v in vars]
     simplify ? simplify_constants.(out) : out
 end
 
-function hessian(O::Operation, vars::AbstractVector{Operation}; simplify = true)
+function hessian(O::Expression, vars::AbstractVector{<:Expression}; simplify = true)
     out = [expand_derivatives(Differential(v2)(Differential(v1)(O))) for v1 in vars, v2 in vars]
     simplify ? simplify_constants.(out) : out
 end
@@ -25,9 +25,7 @@ function simplified_expr(O::Operation)
   return Expr(:call, Symbol(O.op), simplified_expr.(O.args)...)
 end
 
-function simplified_expr(c::Constant)
-    c.value
-end
+simplified_expr(c::Constant) = c.value
 
 function simplified_expr(eq::Equation)
     Expr(:(=), simplified_expr(eq.lhs), simplified_expr(eq.rhs))

test/derivatives.jl

@@ -64,3 +64,11 @@ jac = calculate_jacobian(sys)
 # n-ary * and +
 isequal(ModelingToolkit.derivative(Operation(*, [x, y, z*ρ]), 1), y*(z*ρ))
 isequal(ModelingToolkit.derivative(Operation(+, [x*y, y, z]), 1), 1)
+
+@test iszero(ModelingToolkit.derivative(ModelingToolkit.Constant(42), x))
+@test all(iszero, ModelingToolkit.gradient(ModelingToolkit.Constant(42), [t, x, y, z]))
+@test all(iszero, ModelingToolkit.hessian(ModelingToolkit.Constant(42), [t, x, y, z]))
+@test isequal(ModelingToolkit.jacobian([t, x, ModelingToolkit.Constant(42)], [t, x]),
+              Expression[ModelingToolkit.Constant(1)  ModelingToolkit.Constant(0)
+                         Differential(t)(x)           ModelingToolkit.Constant(1)
+                         ModelingToolkit.Constant(0)  ModelingToolkit.Constant(0)])