fixed conditional test that was using ifelse instead of if_else

added 2 new methods for ifelse so it can take inputs that are Real instead of just Node

updated documentation

docs/src/index.md

@@ -14,33 +14,7 @@ For f:ℝⁿ->ℝᵐ with n,m large **FD** may have better performance than conv
 
  **FD** may take much less time to compute symbolic derivatives than Symbolics.jl even in the ℝ¹->ℝ¹ case. The executables generated by **FD** may also be much faster (see [Symbolic Processing](@ref)). 
 
-As of version 0.4.1 **FD** allows you to create expressions with conditionals:
-```julia
-
-julia> @variables x y
-y
 
-julia> f = ifelse(x<y,x,y)
-(ifelse  (x < y) x y)
-
-julia> a = make_function([f],[x,y])
-
-julia> a(1.0,2.0)
-1-element Vector{Float64}:
- 1.0
-
-julia> a(2.0,1.0)
-1-element Vector{Float64}:
- 2.0
-```
-Howver, you cannot yet compute derivatives of expressions that contain conditionals:
-```julia
-julia> jacobian([f],[x,y])
-ERROR: Your expression contained ifelse. FastDifferentiation does not yet support differentiation through ifelse or any of these conditionals (max, min, copysign, &, |, xor, <, >, <=, >=, !=, ==, signbit, isreal, iszero, isfinite, isnan, isinf, isinteger, !)
-```
-This may be a breaking change for some users. In previous versions this threw an the expression `x==y` returned a `Bool`. Some data structures, such as `Dict` use `==` by default to determine if two entries are the same. This will no longer work since `x==y` will now return an expression graph. Use an `IDict` instead since this uses `===`.
-
-A future PR will add support for differentiating through conditionals.
 
 
 
@@ -66,6 +40,47 @@ This is **beta** software being modified on a daily basis. Expect bugs and frequ
 ## Notes about special derivatives
 The derivative of `|u|` is `u/|u|` which is NaN when `u==0`. This is not a bug. The derivative of the absolute value function is undefined at 0 and the way **FD** signals this is by returning NaN.
 
+## Conditionals
+
+As of version 0.4.1 **FD** allows you to create expressions with conditionals using either the builtin `ifelse` function or a new function `if_else`. `ifelse` will evaluate both inputs. By contrast `if_else` has the semantics of `if...else...end`; only the true or false branch will be executed. This is useful when your conditional is used to prevent exceptions because of illegal input values:
+```julia
+julia> f = if_else(x<0,NaN,sqrt(x))
+(if_else  (x < 0) NaN sqrt(x))
+
+julia> g = make_function([f],[x])
+
+
+julia> g([-1])
+1-element Vector{Float64}:
+ NaN
+
+julia> g([2.0])
+1-element Vector{Float64}:
+ 1.4142135623730951
+end
+```
+In this case you wouldn't want to use `ifelse` because it evaluates both the true and false branches and causes a runtime exception:
+```julia
+julia> f = ifelse(x<0,NaN,sqrt(x))
+(ifelse  (x < 0) NaN sqrt(x))
+
+julia> g = make_function([f],[x])
+...
+
+julia> g([-1])
+ERROR: DomainError with -1.0:
+sqrt was called with a negative real argument but will only return a complex result if called with a complex argument. Try sqrt(Complex(x)).
+```
+
+Howver, you cannot yet compute derivatives of expressions that contain conditionals:
+```julia
+julia> jacobian([f],[x,y])
+ERROR: Your expression contained ifelse. FastDifferentiation does not yet support differentiation through ifelse or any of these conditionals (max, min, copysign, &, |, xor, <, >, <=, >=, !=, ==, signbit, isreal, iszero, isfinite, isnan, isinf, isinteger, !)
+```
+This may be a breaking change for some users. In previous versions the expression `x==y` returned a `Bool`. Some data structures, such as `Dict` use `==` by default to determine if two entries are the same. This will no longer work since `x==y` will now return an expression graph. Use an `IdDict` instead since this uses `===`.
+
+A future PR will add support for differentiating through conditionals.
+
 
 
 

src/Methods.jl

@@ -85,6 +85,8 @@ end
 macro number_methods(T, rhs1, rhs2, options=nothing)
     #special case for ifelse because it takes three arguments
     eval(:(Base.ifelse(a::$T, b::$T, c::$T) = simplify_check_cache(Base.ifelse, a, b, c)))
+    eval(:(Base.ifelse(a::$T, b::$Real, c::$T) = simplify_check_cache(Base.ifelse, a, Node(b), c)))
+    eval(:(Base.ifelse(a::$T, b::$T, c::$Real) = simplify_check_cache(Base.ifelse, a, b, Node(c))))
 
     number_methods(T, rhs1, rhs2, options) |> esc
 end

test/FDTests.jl

@@ -2119,8 +2119,8 @@ end
 
     #conditional
     expr = x < y
-    f = ifelse(expr, x, y)
-    @test ==(FastDifferentiation.value(f), ifelse)
+    f = if_else(expr, x, y)
+    @test ==(FastDifferentiation.value(f), if_else)
     @test ===(FastDifferentiation.children(f)[1], expr)
     @test ===(FastDifferentiation.children(f)[2], x)
     @test ===(FastDifferentiation.children(f)[3], y)