Digamma function

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
The digamma function ,
visualized in discontinuous domain coloring
Real part plots of the digamma and the next three polygamma functions along the real line

In mathematics, the digamma function is defined as the logarithmic derivative of the gamma function:[1][2]

It is the first of the polygamma functions.

The digamma function is often denoted as ψ0(x), ψ(0)(x) or Ϝ[citation needed] (the uppercase form of the archaic Greek consonant digamma meaning double-gamma).

Relation to harmonic numbers[edit]

The gamma function obeys the equation

Taking the derivative with respect to z gives:

Dividing by Γ(z + 1) or the equivalent zΓ(z) gives:

or:

Since the harmonic numbers are defined for positive integers n as

the digamma function is related to them by

where H0 = 0, and γ is the Euler–Mascheroni constant. For half-integer arguments the digamma function takes the values

Integral representations[edit]

If the real part of z is positive then the digamma function has the following integral representation due to Gauss:[3]

Combining this expression with an integral identity for the Euler–Mascheroni constant gives:

The integral is Euler's harmonic number , so the previous formula may also be written

A consequence is the following generalization of the recurrence relation:

An integral representation due to Dirichlet is:[3]

Gauss's integral representation can be manipulated to give the start of the asymptotic expansion of .[4]

This formula is also a consequence of Binet's first integral for the gamma function. The integral may be recognized as a Laplace transform.

Binet's second integral for the gamma function gives a different formula for which also gives the first few terms of the asymptotic expansion:[5]

Infinite product representation[edit]

The function is an entire function,[6] and it can be represented by the infinite product

Here is the kth zero of (see below), and is the Euler–Mascheroni constant.

Series formula[edit]

Euler's product formula for the gamma function, combined with the functional equation and an identity for the Euler–Mascheroni constant, yields the following expression for the digamma function, valid in the complex plane outside the negative integers (Abramowitz and Stegun 6.3.16):[1]

Equivalently,

Evaluation of sums of rational functions[edit]

The above identity can be used to evaluate sums of the form

where p(n) and q(n) are polynomials of n.

Performing partial fraction on un in the complex field, in the case when all roots of q(n) are simple roots,

For the series to converge,

otherwise the series will be greater than the harmonic series and thus diverge. Hence

and

With the series expansion of higher rank polygamma function a generalized formula can be given as

provided the series on the left converges.

Taylor series[edit]

The digamma has a rational zeta series, given by the Taylor series at z = 1. This is

which converges for |z| < 1. Here, ζ(n) is the Riemann zeta function. This series is easily derived from the corresponding Taylor's series for the Hurwitz zeta function.

Newton series[edit]

The Newton series for the digamma, sometimes referred to as Stern series,[7][8] reads

where (s
k
)
is the binomial coefficient. It may also be generalized to

where m = 2,3,4,...[8]

Series with Gregory's coefficients, Cauchy numbers and Bernoulli polynomials of the second kind[edit]

There exist various series for the digamma containing rational coefficients only for the rational arguments. In particular, the series with Gregory's coefficients Gn is

where (v)n is the rising factorial (v)n = v(v+1)(v+2) ... (v+n-1), Gn(k) are the Gregory coefficients of higher order with Gn(1) = Gn, Γ is the gamma function and ζ is the Hurwitz zeta function.[9][8] Similar series with the Cauchy numbers of the second kind Cn reads[9][8]

A series with the Bernoulli polynomials of the second kind has the following form[8]

where ψn(a) are the Bernoulli polynomials of the second kind defined by the generating equation

It may be generalized to

where the polynomials Nn,r(a) are given by the following generating equation

so that Nn,1(a) = ψn(a).[8] Similar expressions with the logarithm of the gamma function involve these formulas[8]

and

Reflection formula[edit]

The digamma function satisfies a reflection formula similar to that of the gamma function:

Recurrence formula and characterization[edit]

The digamma function satisfies the recurrence relation

Thus, it can be said to "telescope" 1 / x, for one has

where Δ is the forward difference operator. This satisfies the recurrence relation of a partial sum of the harmonic series, thus implying the formula

where γ is the Euler–Mascheroni constant.

More generally, one has

for . Another series expansion is:

,

where are the Bernoulli numbers. This series diverges for all z and is known as the Stirling series.

Actually, ψ is the only solution of the functional equation

that is monotonic on + and satisfies F(1) = −γ. This fact follows immediately from the uniqueness of the Γ function given its recurrence equation and convexity restriction. This implies the useful difference equation:

Some finite sums involving the digamma function[edit]

There are numerous finite summation formulas for the digamma function. Basic summation formulas, such as

are due to Gauss.[10][11] More complicated formulas, such as

are due to works of certain modern authors (see e.g. Appendix B in Blagouchine (2014)[12]).

Gauss's digamma theorem[edit]

For positive integers r and m (r < m), the digamma function may be expressed in terms of Euler's constant and a finite number of elementary functions

which holds, because of its recurrence equation, for all rational arguments.

Asymptotic expansion[edit]

The digamma function has the asymptotic expansion

where Bk is the kth Bernoulli number and ζ is the Riemann zeta function. The first few terms of this expansion are:

Although the infinite sum does not converge for any z, any finite partial sum becomes increasingly accurate as z increases.

The expansion can be found by applying the Euler–Maclaurin formula to the sum[13]

The expansion can also be derived from the integral representation coming from Binet's second integral formula for the gamma function. Expanding as a geometric series and substituting an integral representation of the Bernoulli numbers leads to the same asymptotic series as above. Furthermore, expanding only finitely many terms of the series gives a formula with an explicit error term:

Inequalities[edit]

When x > 0, the function

is completely monotonic and in particular positive. This is a consequence of Bernstein's theorem on monotone functions applied to the integral representation coming from Binet's first integral for the gamma function. Additionally, by the convexity inequality , the integrand in this representation is bounded above by . Consequently

is also completely monotonic. It follows that, for all x > 0,

This recovers a theorem of Horst Alzer.[14] Alzer also proved that, for s ∈ (0, 1),[15]

Related bounds were obtained by Elezovic, Giordano, and Pecaric, who proved that, for x > 0 ,

where is the Euler–Mascheroni constant.[16] The constants appearing in these bounds are the best possible.[17]

The mean value theorem implies the following analog of Gautschi's inequality: If x > c, where c ≈ 1.461 is the unique positive real root of the digamma function, and if s > 0, then

Moreover, equality holds if and only if s = 1.[18]

Inspired by the harmonic mean value inequality for the classical gamma function, Horzt Alzer and Graham Jameson proved, among other things, a harmonic mean-value inequality for the digamma function:

for

Equality holds if and only if .[19]

Computation and approximation[edit]

The asymptotic expansion gives an easy way to compute ψ(x) when the real part of x is large. To compute ψ(x) for small x, the recurrence relation

can be used to shift the value of x to a higher value. Beal[20] suggests using the above recurrence to shift x to a value greater than 6 and then applying the above expansion with terms above x14 cut off, which yields "more than enough precision" (at least 12 digits except near the zeroes).

As x goes to infinity, ψ(x) gets arbitrarily close to both ln(x − 1/2) and ln x. Going down from x + 1 to x, ψ decreases by 1 / x, ln(x − 1/2) decreases by ln (x + 1/2) / (x − 1/2), which is more than 1 / x, and ln x decreases by ln (1 + 1 / x), which is less than 1 / x. From this we see that for any positive x greater than 1/2,

or, for any positive x,

The exponential exp ψ(x) is approximately x − 1/2 for large x, but gets closer to x at small x, approaching 0 at x = 0.

For x < 1, we can calculate limits based on the fact that between 1 and 2, ψ(x) ∈ [−γ, 1 − γ], so

or

From the above asymptotic series for ψ, one can derive an asymptotic series for exp(−ψ(x)). The series matches the overall behaviour well, that is, it behaves asymptotically as it should for large arguments, and has a zero of unbounded multiplicity at the origin too.

This is similar to a Taylor expansion of exp(−ψ(1 / y)) at y = 0, but it does not converge.[21] (The function is not analytic at infinity.) A similar series exists for exp(ψ(x)) which starts with

If one calculates the asymptotic series for ψ(x+1/2) it turns out that there are no odd powers of x (there is no x−1 term). This leads to the following asymptotic expansion, which saves computing terms of even order.

Special values[edit]

The digamma function has values in closed form for rational numbers, as a result of Gauss's digamma theorem. Some are listed below:

Moreover, by the series representation, it can easily be deduced that at the imaginary unit,

Roots of the digamma function[edit]

The roots of the digamma function are the saddle points of the complex-valued gamma function. Thus they lie all on the real axis. The only one on the positive real axis is the unique minimum of the real-valued gamma function on + at x0 = 1.461632144968.... All others occur single between the poles on the negative axis:

Already in 1881, Charles Hermite observed[22] that

holds asymptotically. A better approximation of the location of the roots is given by

and using a further term it becomes still better

which both spring off the reflection formula via

and substituting ψ(xn) by its not convergent asymptotic expansion. The correct second term of this expansion is 1 / 2n, where the given one works good to approximate roots with small n.

Another improvement of Hermite's formula can be given:[6]

Regarding the zeros, the following infinite sum identities were recently proved by István Mező and Michael Hoffman[6]

In general, the function

can be determined and it is studied in detail by the cited authors.

The following results[6]

also hold true.

Here γ is the Euler–Mascheroni constant.

Regularization[edit]

The digamma function appears in the regularization of divergent integrals

this integral can be approximated by a divergent general Harmonic series, but the following value can be attached to the series

See also[edit]

  • Polygamma function
  • Trigamma function
  • Chebyshev expansions of the digamma function in Wimp, Jet (1961). "Polynomial approximations to integral transforms". Math. Comp. 15 (74): 174–178. doi:10.1090/S0025-5718-61-99221-3.

References[edit]

  1. ^ a b Abramowitz, M.; Stegun, I. A., eds. (1972). "6.3 psi (Digamma) Function.". Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables (10th ed.). New York: Dover. pp. 258–259.
  2. ^ Weisstein, Eric W. "Digamma function". MathWorld.
  3. ^ a b Whittaker and Watson, 12.3.
  4. ^ Whittaker and Watson, 12.31.
  5. ^ Whittaker and Watson, 12.32, example.
  6. ^ a b c d Mező, István; Hoffman, Michael E. (2017). "Zeros of the digamma function and its Barnes G-function analogue". Integral Transforms and Special Functions. 28 (28): 846–858. doi:10.1080/10652469.2017.1376193.
  7. ^ Nörlund, N. E. (1924). Vorlesungen über Differenzenrechnung. Berlin: Springer.
  8. ^ a b c d e f g Blagouchine, Ia. V. (2018). "Three Notes on Ser's and Hasse's Representations for the Zeta-functions" (PDF). Integers (Electronic Journal of Combinatorial Number Theory). 18A: 1–45. arXiv:1606.02044. Bibcode:2016arXiv160602044B.
  9. ^ a b Blagouchine, Ia. V. (2016). "Two series expansions for the logarithm of the gamma function involving Stirling numbers and containing only rational coefficients for certain arguments related to π−1". Journal of Mathematical Analysis and Applications. 442: 404–434. arXiv:1408.3902. Bibcode:2014arXiv1408.3902B.
  10. ^ R. Campbell. Les intégrales eulériennes et leurs applications, Dunod, Paris, 1966.
  11. ^ H.M. Srivastava and J. Choi. Series Associated with the Zeta and Related Functions, Kluwer Academic Publishers, the Netherlands, 2001.
  12. ^ Blagouchine, Iaroslav V. (2014). "A theorem for the closed-form evaluation of the first generalized Stieltjes constant at rational arguments and some related summations". Journal of Number Theory. 148: 537–592. arXiv:1401.3724. doi:10.1016/j.jnt.2014.08.009.
  13. ^ Bernardo, José M. (1976). "Algorithm AS 103 psi(digamma function) computation" (PDF). Applied Statistics. 25: 315–317.
  14. ^ H. Alzer, On some inequalities for the gamma and psi functions, Math. Comp. 66 (217) (1997) 373–389.
  15. ^ Ibid.
  16. ^ N. Elezovic, C. Giordano and J. Pecaric, The best bounds in Gautschi’s inequality, Math. Inequal. Appl. 3 (2000), 239–252.
  17. ^ F. Qi and B.-N. Guo, Sharp inequalities for the psi function and harmonic numbers, arXiv:0902.2524.
  18. ^ A. Laforgia, P. Natalini, Exponential, gamma and polygamma functions: Simple proofs of classical and new inequalities, J. Math. Anal. Appl. 407 (2013) 495–504.
  19. ^ Alzer, Horst; Jameson, Graham (2017). "A harmonic mean inequality for the digamma function and related results" (PDF). Rendiconti del Seminario Matematico della Università di Padova. 70 (201): 203–209. doi:10.4171/RSMUP/137-10. ISSN 0041-8994. LCCN 50046633. OCLC 01761704.
  20. ^ Beal, Matthew J. (2003). Variational Algorithms for Approximate Bayesian Inference (PDF) (PhD thesis). The Gatsby Computational Neuroscience Unit, University College London. pp. 265–266.
  21. ^ If it converged to a function f(y) then ln(f(y) / y) would have the same Maclaurin series as ln(1 / y) − φ(1 / y). But this does not converge because the series given earlier for φ(x) does not converge.
  22. ^ Hermite, Charles (1881). "Sur l'intégrale Eulérienne de seconde espéce,". Journal für die reine und angewandte Mathematik (90): 332–338.

External links[edit]

OEISA047787 psi(1/3), OEISA200064 psi(2/3), OEISA020777 psi(1/4), OEISA200134 psi(3/4), OEISA200135 to OEISA200138 psi(1/5) to psi(4/5).