is large, sums or differences of psi functions may be computed without unnecessary loss of precision, by analytically combining the logarithmic terms. For example, the difference

d = ψ (x + \frac{1}{2}) - ψ (x)

has an asymptotic behaviour for large

x

given by

d \sim \ln (x + \frac{1}{2}) - \ln x + O (\frac{1}{x^{2}}) \sim \ln (1 + \frac{1}{2 x}) \sim \frac{1}{2 x}

Computing

d

directly would amount to subtracting two large numbers which are close to

\ln (x + \frac{1}{2})

and

\ln x

to produce a small number close to

\frac{1}{2 x}

, resulting in a loss of significant digits. However, using this function to compute

f (x) = ψ (x) - \ln x

, we can compute

d = f (x + \frac{1}{2}) - f (x) + \ln (1 + \frac{1}{2 x})

, and the dominant logarithmic term may be computed accurately from its power series when

x

is large. Thus we avoid the unnecessary loss of precision.

The function is derived from the function PSIFN in Amos (1983).

4 References

NIST Digital Library of Mathematical Functions

Amos D E (1983) Algorithm 610: A portable FORTRAN subroutine for derivatives of the psi function ACM Trans. Math. Software 9 494–502

5 Arguments

1: $x$ – double Input: On entry: the argument $x$ of the function.

Constraint: $x > 0.0$ .
2: $fail$ – NagError * Input/Output: The NAG error argument (see Section 7 in the Introduction to the NAG Library CL Interface).

6 Error Indicators and Warnings

NE_ALLOC_FAIL: Dynamic memory allocation failed.
See Section 3.1.2 in the Introduction to the NAG Library CL Interface for further information.
NE_INTERNAL_ERROR: An internal error has occurred in this function. Check the function call and any array sizes. If the call is correct then please contact NAG for assistance.
See Section 7.5 in the Introduction to the NAG Library CL Interface for further information.
NE_NO_LICENCE: Your licence key may have expired or may not have been installed correctly.
See Section 8 in the Introduction to the NAG Library CL Interface for further information.
NE_OVERFLOW_LIKELY: Computation halted due to likelihood of overflow. x may be too small. $x = ⟨ value ⟩$ .
NE_REAL: On entry, $x = ⟨ value ⟩$ .
Constraint: $x > 0.0$ .
NE_UNDERFLOW_LIKELY: Computation halted due to likelihood of underflow. x may be too large. $x = ⟨ value ⟩$ .

7 Accuracy

All constants in s14acc are given to approximately

18

digits of precision. Calling the number of digits of precision in the floating-point arithmetic being used

t

, then clearly the maximum number of correct digits in the results obtained is limited by

p = \min (t, 18)

With the above proviso, results returned by this function should be accurate almost to full precision, except at points close to the zero of

ψ (x)

x ≃ 1.461632

, where only absolute rather than relative accuracy can be obtained.

8 Parallelism and Performance

s14acc is not threaded in any implementation.

9 Further Comments

None.

10 Example

The example program reads values of the argument

x

from a file, evaluates the function at each value of

x

and prints the results.

s14ac: FL CL CPP AD

NAG CL Interfaces14acc (polygamma)

▸▿ Contents

1 Purpose

2 Specification

3 Description

4 References

5 Arguments

6 Error Indicators and Warnings

7 Accuracy

8 Parallelism and Performance

9 Further Comments

10 Example

10.1 Program Text

10.2 Program Data

10.3 Program Results

NAG CL Interface
s14acc (polygamma)