Refactor SpecialFunctionsExt

ext/TaylorDiffSFExt.jl

@@ -1,34 +1,8 @@
 module TaylorDiffSFExt
 using TaylorDiff, SpecialFunctions
-using Symbolics: @variables
-using SymbolicUtils, SymbolicUtils.Code
-using SymbolicUtils: Pow
-using TaylorDiff: value, raise
-using ChainRules, ChainRulesCore
 
-dummy = (NoTangent(), 1)
-@variables z
-# logerfc, logerfcx, erfinv, gamma, digamma, trigamma
 for func in (erf, erfc, erfcinv, erfcx, erfi)
-    F = typeof(func)
-    # base case
-    @eval function (op::$F)(t::TaylorScalar{T, 2}) where {T}
-        t0, t1 = value(t)
-        TaylorScalar{T, 2}(frule((NoTangent(), t1), op, t0))
-    end
-    der = frule(dummy, func, z)[2]
-    term, raiser = der isa Pow && der.exp == -1 ? (der.base, raiseinv) : (der, raise)
-    # recursion by raising
-    @eval @generated function (op::$F)(t::TaylorScalar{T, N}) where {T, N}
-        der_expr = $(QuoteNode(toexpr(term)))
-        f = $func
-        quote
-            $(Expr(:meta, :inline))
-            z = TaylorScalar{T, N - 1}(t)
-            df = $der_expr
-            $$raiser($f(value(t)[1]), df, t)
-        end
-    end
+    TaylorDiff.define_unary_function(func, TaylorDiffSFExt)
 end
 
 end

src/TaylorDiff.jl

@@ -2,6 +2,7 @@ module TaylorDiff
 
 include("scalar.jl")
 include("primitive.jl")
+include("utils.jl")
 include("codegen.jl")
 include("derivative.jl")
 include("chainrules.jl")

src/codegen.jl

@@ -1,40 +1,10 @@
-using ChainRules
-using ChainRulesCore
-using Symbolics: @variables
-using SymbolicUtils, SymbolicUtils.Code
-using SymbolicUtils: Pow
-
-func_list = (
+for unary_func in (
     +, -, deg2rad, rad2deg,
     sinh, cosh, tanh,
     asin, acos, atan, asec, acsc, acot,
     log, log10, log1p, log2,
     asinh, acosh, atanh, asech, acsch,
     acoth,
     abs, sign)
-
-dummy = (NoTangent(), 1)
-@variables z
-for func in func_list
-    F = typeof(func)
-    # base case
-    @eval function (op::$F)(t::TaylorScalar{T, 2}) where {T}
-        t0, t1 = value(t)
-        f0, f1 = frule((NoTangent(), t1), op, t0)
-        TaylorScalar{T, 2}(f0, zero_tangent(f0) + f1)
-    end
-    der = frule(dummy, func, z)[2]
-    term, raiser = der isa Pow && der.exp == -1 ? (der.base, raiseinv) : (der, raise)
-    # recursion by raising
-    @eval @generated function (op::$F)(t::TaylorScalar{T, N}) where {T, N}
-        der_expr = $(QuoteNode(toexpr(term)))
-        f = $func
-        quote
-            $(Expr(:meta, :inline))
-            z = TaylorScalar{T, N - 1}(t)
-            f0 = $f(value(t)[1])
-            df = zero_tangent(z) + $der_expr
-            $$raiser(f0, df, t)
-        end
-    end
+    define_unary_function(unary_func, TaylorDiff)
 end

src/utils.jl

@@ -0,0 +1,32 @@
+using ChainRules
+using ChainRulesCore
+using Symbolics: @variables
+using SymbolicUtils, SymbolicUtils.Code
+using SymbolicUtils: Pow
+
+dummy = (NoTangent(), 1)
+@variables z
+
+function define_unary_function(func, m)
+    F = typeof(func)
+    # base case
+    @eval m function (op::$F)(t::TaylorScalar{T, 2}) where {T}
+        t0, t1 = value(t)
+        f0, f1 = frule((NoTangent(), t1), op, t0)
+        TaylorScalar{T, 2}(f0, zero_tangent(f0) + f1)
+    end
+    der = frule(dummy, func, z)[2]
+    term, raiser = der isa Pow && der.exp == -1 ? (der.base, raiseinv) : (der, raise)
+    # recursion by raising
+    @eval m @generated function (op::$F)(t::TaylorScalar{T, N}) where {T, N}
+        der_expr = $(QuoteNode(toexpr(term)))
+        f = $func
+        quote
+            $(Expr(:meta, :inline))
+            z = TaylorScalar{T, N - 1}(t)
+            f0 = $f(value(t)[1])
+            df = zero_tangent(z) + $der_expr
+            $$raiser(f0, df, t)
+        end
+    end
+end