Implement the T-distribution via the F-distribution (#149)

src/distrs/tdist.jl

@@ -1,19 +1,5 @@
 # functions related to student's T distribution
 
-# R implementations
-using .RFunctions:
-    # tdistpdf,
-    # tdistlogpdf,
-    tdistcdf,
-    tdistccdf,
-    tdistlogcdf,
-    tdistlogccdf,
-    tdistinvcdf,
-    tdistinvccdf,
-    tdistinvlogcdf,
-    tdistinvlogccdf
-
-# Julia implementations
 tdistpdf(ν::Real, x::Real) = exp(tdistlogpdf(ν, x))
 
 tdistlogpdf(ν::Real, x::Real) = tdistlogpdf(promote(ν, x)...)
@@ -22,3 +8,67 @@ function tdistlogpdf(ν::T, x::T) where {T<:Real}
     νp12 = (ν + 1) / 2
     return loggamma(νp12) - (logπ + log(ν)) / 2 - loggamma(ν / 2) - νp12 * log1p(x^2 / ν)
 end
+
+function tdistcdf(ν::T, x::T) where T<:Real
+    if isinf(ν)
+        return normcdf(x)
+    elseif x < 0
+        return fdistccdf(one(ν), ν, x^2)/2
+    else
+        return 1 - fdistccdf(one(ν), ν, x^2)/2
+    end
+end
+tdistcdf(ν::Real, x::Real) = tdistcdf(map(float, promote(ν, x))...)
+
+tdistccdf(ν::Real, x::Real) = tdistcdf(ν, -x)
+
+function tdistlogcdf(ν::T, x::T) where T<:Real
+    if isinf(ν)
+        return normlogcdf(x)
+    elseif x < 0
+        ret = fdistlogccdf(one(ν), ν, x^2)
+        return ret - logtwo
+    else
+        return log1p(-fdistccdf(one(ν), ν, x^2)/2)
+    end
+end
+tdistlogcdf(ν::Real, x::Real) = tdistlogcdf(map(float, promote(ν, x))...)
+
+tdistlogccdf(ν::Real, x::Real) = tdistlogcdf(ν, -x)
+
+function tdistinvcdf(ν::T, p::T) where T<:Real
+    if isinf(ν)
+        return norminvcdf(p)
+    elseif p < 0.5
+        return -sqrt(fdistinvccdf(one(ν), ν, 2*p))
+    else
+        return sqrt(fdistinvccdf(one(ν), ν, 2*(1 - p)))
+    end
+end
+tdistinvcdf(ν::Real, p::Real) = tdistinvcdf(map(float, promote(ν, p))...)
+
+tdistinvccdf(ν::Real, p::Real) = -tdistinvcdf(ν, p)
+
+if VERSION < v"1.7.0-DEV.1172"
+    function _expm1(x::Float16)
+        if -0.2 < x < 0.1
+            return @evalpoly(x, Float16(0), Float16(1), Float16(1/2), Float16(1/6), Float16(1/24), Float16(1/120))
+        else
+            return exp(x) - 1
+        end
+    end
+end
+_expm1(x::Number) = expm1(x)
+
+function tdistinvlogcdf(ν::T, logp::T) where T<:Real
+    if isinf(ν)
+        return norminvlogcdf(logp)
+    elseif logp < -log(2)
+        return -sqrt(fdistinvlogccdf(one(ν), ν, logp + logtwo))
+    else
+        return sqrt(fdistinvccdf(one(ν), ν, -2*_expm1(logp)))
+    end
+end
+tdistinvlogcdf(ν::Real, logp::Real) = tdistinvlogcdf(map(float, promote(ν, logp))...)
+
+tdistinvlogccdf(ν::Real, logp::Real) = -tdistinvlogcdf(ν, logp)