add project.toml and tests

Author: oscardssmith

Project.toml

@@ -0,0 +1,20 @@
+name = "FunctionAccuracyTests"
+uuid = "c7c8a6d5-64ef-4165-b05f-11d5e183af77"
+authors = ["Oscar Smith <oscardssmith@gmail.com>"]
+version = "0.1.0"
+
+[deps]
+Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[compat]
+Printf = "1"
+Test = "1"
+julia = "1"
+
+[extras]
+Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["Printf", "Test"]

src/FunctionAccuracyTests.jl

@@ -1,12 +1,17 @@
+module FunctionAccuracyTests
+
 using Base.Math: significand_bits, exponent_bias
 using Test, Printf
 
+export FloatIterator, test_acc
+
 struct FloatIterator{T}<:AbstractVector{T}
     min::T
     max::T
     FloatIterator{T}(min, max) where T = min > max ? error("max less than min") : new{T}(min,max)
 end
 
+FloatIterator(min::T,max::T) where T = FloatIterator{T}(min,max)
 FloatIterator{T}() where T = FloatIterator{T}(T(-Inf),T(Inf))
 Base.iterate(it::FloatIterator) = (it.min, nextfloat(it.min))
 function Base.iterate(it::FloatIterator{T}, el) where T
@@ -96,6 +101,8 @@ function test_acc(fun_table::Dict, xx; tol=1.5, debug = true, tol_debug = 5)
         t = @test rmax <= tol
     end
 end
-test_acc(fun_table::Dict, xx; tol = 1.5, debug = true, tol_debug = 5) = test_acc(eltype(xx), fun_table, xx; tol = 1.5 debug = true, tol_debug = 5)
-test_acc(f::Function, xx; tol = 1.5 debug = true, tol_debug = 5) = test_acc(Dict(f=>f), xx; tol = 1.5 debug = true, tol_debug = 5)
-test_acc(f::Function, min, max; tol = 1.5 debug = true, tol_debug = 5) = test_acc(Dict(f=>f), FloatIterator(min,max); tol = 1.5 debug = true, tol_debug = 5)
+
+test_acc(f::Function, xx; tol = 1.5, debug = true, tol_debug = 5) = test_acc(Dict(f=>f), xx; tol = 1.5, debug = true, tol_debug = 5)
+test_acc(f::Function, min, max; tol = 1.5, debug = true, tol_debug = 5) = test_acc(Dict(f=>f), FloatIterator(min,max); tol = 1.5, debug = true, tol_debug = 5)
+
+end # module

test/runtests.jl

@@ -0,0 +1,47 @@
+using Printf, Test
+using FunctionAccuracyTests
+
+MAX_EXP(n::Val{2}, ::Type{Float32}) = 128.0f0
+MAX_EXP(n::Val{2}, ::Type{Float64}) = 1024.0
+MAX_EXP(n::Val{:ℯ}, ::Type{Float32}) = 88.72284f0
+MAX_EXP(n::Val{:ℯ}, ::Type{Float64}) = 709.7827128933841
+MAX_EXP(n::Val{10}, ::Type{Float32}) = 38.53184f0
+MAX_EXP(n::Val{10}, ::Type{Float64}) = 308.25471555991675
+
+# min_exp = T(-(exponent_bias(T)+significand_bits(T)) * log(base, big(2)))
+MIN_EXP(n::Val{2}, ::Type{Float32}) = -150.0f0
+MIN_EXP(n::Val{2}, ::Type{Float64}) = -1075.0
+MIN_EXP(n::Val{:ℯ}, ::Type{Float32}) = -103.97208f0
+MIN_EXP(n::Val{:ℯ}, ::Type{Float64}) = -745.1332191019412
+MIN_EXP(n::Val{10}, ::Type{Float32}) = -45.1545f0
+MIN_EXP(n::Val{10}, ::Type{Float64}) = -323.60724533877976
+
+for (func, base) in (exp2=>Val(2), exp=>Val(:ℯ), exp10=>Val(10))
+    for T in (Float32, Float64)
+        xx = range(MIN_EXP(base,T),  MAX_EXP(base,T), length = 10^6);
+        test_acc(func, xx)
+    end
+end
+
+
+exp10(Float16(-3.764))
+exp10(Float16(-1.76))
+asinh(Float16(258.5))
+asinh(Float16(-258.5))
+@testset "Float16" begin
+    @testset "$func" for func in (atan,sinh,cosh,tanh,asinh,
+                                  exp,exp2,exp10,expm1,cbrt)
+        test_acc(func, -Inf16, Inf16, tol=.501)
+    end
+    @testset "$func" for func in (sin, cos, tan)
+        test_acc(func, -floatmax(Float16),floatmax(Float16), tol=.501)
+    end
+    @testset "$func" for func in (asin, acos, atanh)
+        test_acc(func, -one(Float16),one(Float16), tol=.501)
+    end
+    test_acc(acosh, one(Float16), Inf16, tol=.501)
+    @testset "$func" for func in (log, log2, log10, sqrt)
+        test_acc(func, zero(Float16), Inf16, tol=.501)
+    end
+    test_acc(log1p, -one(Float16), Inf16, tol=.501)
+end