dispatch int gamma

Author: heltonmc

src/besselj.jl

@@ -348,10 +348,10 @@ In general, this is most accurate for small arguments and when nu > x.
 function besselj_power_series(v, x::T) where T
     MaxIter = 3000
     S = promote_type(T, Float64)
-    v, x = S(v), S(x)
+    x = S(x)
 
     out = zero(S)
-    a = (x/2)^v / gamma(v + one(S))
+    a = (x/2)^v / gamma(v + 1)
     t2 = (x/2)^2
     for i in 0:MaxIter
         out += a

src/gamma.jl

@@ -1,5 +1,5 @@
 # Adapted from Cephes Mathematical Library (MIT license https://en.smath.com/view/CephesMathLibrary/license) by Stephen L. Moshier
-gamma(z::Number) = _gamma(float(z))
+gamma(x::Float64) = _gamma(x)
 _gamma(x::Float32) = Float32(_gamma(Float64(x)))
 
 function _gamma(x::Float64)
@@ -50,7 +50,6 @@ function _gamma(x::Float64)
     return z * p / q
 end
 
-
 function gamma(n::Integer)
     n < 0 && throw(DomainError(n, "`n` must not be negative."))
     n == 0 && return Inf*float(n)

test/gamma_test.jl

@@ -3,3 +3,6 @@ x = rand(10000)*170
 @test SpecialFunctions.gamma.(BigFloat.(-x)) ≈ Bessels.gamma.(-x)
 @test isnan(Bessels.gamma(NaN))
 @test isinf(Bessels.gamma(Inf))
+
+x = [0, 1, 2, 3, 8, 15, 20, 30]
+@test SpecialFunctions.gamma.(x) ≈ Bessels.gamma.(x)