Module: Distribution::MathExtension::Gammastar

Defined in:
lib/distribution/math_extension/gammastar.rb

Overview

Derived from GSL-1.9.

Constant Summary collapse

C0 =
1.quo(12)
C1 =
-1.quo(360)
C2 =
1.quo(1260)
C3 =
-1.quo(1680)
C4 =
1.quo(1188)
C5 =
-691.quo(360_360)
C6 =
1.quo(156)
C7 =
-3617.quo(122_400)

Class Method Summary collapse

Class Method Details

.evaluate(x, with_error = false) ⇒ Object



28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
# File 'lib/distribution/math_extension/gammastar.rb', line 28

def evaluate(x, with_error = false)
  fail(ArgumentError, 'x must be positive') if x <= 0
  if x < 0.5
    STDERR.puts("Warning: Don't know error on lg_x, error for this function will be incorrect") if with_error
    lg = Math.lgamma(x).first
    lg_err = Float::EPSILON # Guess
    lx = Math.log(x)
    c    = 0.5 * (LN2 + LNPI)
    lnr_val = lg - (x - 0.5) * lx + x - c
    lnr_err = lg_err + 2.0 * Float::EPSILON * ((x + 0.5) * lx.abs + c)
    with_error ? exp_err(lnr_val, lnr_err) : Math.exp(lnr_val)
  elsif x < 2.0
    t = 4.0 / 3.0 * (x - 0.5) - 1.0
    ChebyshevSeries.evaluate(:gstar_a, t, with_error)
  elsif x < 10.0
    t = 0.25 * (x - 2.0) - 1.0
    c = ChebyshevSeries.evaluate(:gstar_b, t, with_error)
    c, c_err = c if with_error

    result      = c / (x * x) + 1.0 + 1.0 / (12.0 * x)
    with_error ? [result, c_err / (x * x) + 2.0 * Float::EPSILON * result.abs] : result
  elsif x < 1.0 / Math::ROOT4_FLOAT_EPSILON
    series x, with_error
  elsif x < 1.0 / Float::EPSILON # Stirling
    xi = 1.0 / x
    result = 1.0 + xi / 12.0 * (1.0 + xi / 24.0 * (1.0 - xi * (139.0 / 180.0 + 571.0 / 8640.0 * xi)))
    result_err = 2.0 * Float::EPSILON * result.abs
    with_error ? [result, result_err] : result
  else
    with_error ? [1.0, 1.0 / x] : 1.0
  end
end

.series(x, with_error = false) ⇒ Object



18
19
20
21
22
23
24
25
26
# File 'lib/distribution/math_extension/gammastar.rb', line 18

def series(x, with_error = false)
  # Use the Stirling series for the correction to Log(Gamma(x)),
  # which is better behaved and easier to compute than the
  # regular Stirling series for Gamma(x).
  y      = 1.quo(x * x)
  ser    = C0 + y * (C1 + y * (C2 + y * (C3 + y * (C4 + y * (C5 + y * (C6 + y * C7))))))
  result = Math.exp(ser / x)
  with_error ? [result, 2.0 * Float::EPSILON * result * [1, ser / x].max] : result
end