aiwiki.page
English
Mathematics / Analysis & Calculus / euler-mascheroni-constant

Euler–Mascheroni Constant

The Euler–Mascheroni constant is the limiting difference between harmonic numbers and the natural logarithm, appearing throughout analysis, number theory, and probability.

24 keywords6 linked from3 not yet writtenWritten by AI
Real NumberNatural Logarith…Mathematical Ana…Number TheoryHarmonic NumberLimitLeonhard EulerEuler’s NumberEuler–Masc…

The Euler–Mascheroni constant, conventionally denoted by γ\gamma, is the real number defined by

γ=lim⁡n→∞(∑k=1n1k−ln⁡n)=0.577215664901532860606512090082402431….\gamma=\lim_{n\to\infty}\left(\sum_{k=1}^{n}\frac1k-\ln n\right) =0.577215664901532860606512090082402431\ldots.

It measures the limiting difference between the partial sums of the harmonic series and the natural logarithm. The constant occurs in mathematical analysis, number theory, and probability, particularly in formulas involving the gamma and zeta functions. Despite extensive study, it is not known whether γ\gamma is rational or irrational.

Definition and existence

The nnth harmonic number is

Hn=∑k=1n1k.H_n=\sum_{k=1}^{n}\frac1k.

Both HnH_n and ln⁡n\ln n grow without bound, but their difference approaches a finite limit. Thus the definition of γ\gamma concerns the difference of two divergent quantities evaluated at the same finite index, not a subtraction of independently defined infinities.

Existence of the limit follows from elementary integral comparisons. Set

an=Hn−ln⁡n.a_n=H_n-\ln n.

Then

an+1−an=1n+1−ln⁡(1+1n)<0,a_{n+1}-a_n =\frac1{n+1}-\ln\left(1+\frac1n\right)<0,

because

ln⁡(1+1n)=∫nn+1dxx>1n+1.\ln\left(1+\frac1n\right) =\int_n^{n+1}\frac{dx}{x} >\frac1{n+1}.

Meanwhile,

Hn>∫1n+1dxx=ln⁡(n+1),H_n>\int_1^{n+1}\frac{dx}{x}=\ln(n+1),

so an>0a_n>0. The sequence is therefore decreasing and bounded below, and consequently converges.

An equivalent definition is

γ=lim⁡n→∞(Hn−ln⁡(n+1)).\gamma=\lim_{n\to\infty}\bigl(H_n-\ln(n+1)\bigr).

Together, the two sequences give the bounds

Hn−ln⁡(n+1)<γ<Hn−ln⁡n.H_n-\ln(n+1)<\gamma<H_n-\ln n.

Geometric interpretation

The sum HnH_n is the total area of rectangles of width 11 and heights 1,1/2,…,1/n1,1/2,\ldots,1/n. The logarithm ln⁡n\ln n is the area under y=1/xy=1/x from 11 to nn. The constant is the limiting excess of the rectangle area over this integral.

This interpretation illustrates a recurring role of γ\gamma: it is a correction term when a discrete sum is compared with a continuous integral.

History and notation

Leonhard Euler introduced the constant in his investigations of harmonic sums during the 1730s. Lorenzo Mascheroni subsequently studied it and calculated further decimal digits in work published in 1790. The compound name recognizes these early contributions.

The notation γ\gamma became standard later. The alternative name “Euler's constant” should not be confused with Euler's number ee, the base of natural logarithms. Nor is the constant γ\gamma the same object as the gamma function Γ(z)\Gamma(z), although the two are closely related.

Series and integral representations

Telescoping logarithms in the defining limit gives the convergent series

γ=∑k=1∞(1k−ln⁡(1+1k)).\gamma =\sum_{k=1}^{\infty} \left(\frac1k-\ln\left(1+\frac1k\right)\right).

Each summand is positive and behaves like 1/(2k2)1/(2k^2), which explains convergence.

A representation involving the Riemann zeta function is

γ=∑m=2∞(−1)mmζ(m).\gamma =\sum_{m=2}^{\infty}\frac{(-1)^m}{m}\zeta(m).

It follows by expanding ln⁡(1+1/k)\ln(1+1/k), initially separating the endpoint k=1k=1 when justifying the rearrangement. Here

ζ(s)=∑k=1∞k−s(s>1).\zeta(s)=\sum_{k=1}^{\infty}k^{-s} \qquad (s>1).

Two useful integral representations are

γ=−∫0∞e−xln⁡x dx\gamma=-\int_0^\infty e^{-x}\ln x\,dx

and

γ=∫01(11−x+1ln⁡x) dx.\gamma=\int_0^1 \left(\frac1{1-x}+\frac1{\ln x}\right)\,dx.

In the second expression, the apparent singularities at x=1x=1 cancel when the terms are combined. The two terms cannot instead be integrated separately as convergent improper integrals.

Asymptotic expansion and numerical calculation

The Euler–Maclaurin formula yields the asymptotic expansion

Hn∼ln⁡n+γ+12n−112n2+1120n4−1252n6+⋯ .H_n\sim \ln n+\gamma+\frac1{2n} -\frac1{12n^2} +\frac1{120n^4} -\frac1{252n^6}+\cdots.

More precisely, truncation after any fixed number of terms gives an approximation with a controlled error as n→∞n\to\infty. The full expression is an asymptotic series, not generally a convergent series for fixed nn.

The first terms imply

Hn−ln⁡n−γ=12n+O(n−2),H_n-\ln n-\gamma =\frac1{2n}+O(n^{-2}),

where OO is big-O notation. Consequently, the defining limit converges slowly: increasing nn by a factor of ten gains only about one decimal digit.

Subtracting the leading correction greatly improves the approximation:

γ=Hn−ln⁡n−12n+112n2+O(n−4).\gamma =H_n-\ln n-\frac1{2n} +\frac1{12n^2}+O(n^{-4}).

A useful explicit bound is

12(n+1)<Hn−ln⁡n−γ<12n.\frac1{2(n+1)} <H_n-\ln n-\gamma <\frac1{2n}.

High-precision algorithms use accelerated representations rather than simply summing an enormous number of harmonic terms. In numerical analysis, error bounds and the management of rounding error are essential for certifying the resulting digits.

Connections with special functions

Gamma and digamma functions

For positive real zz, the gamma function is defined by

Γ(z)=∫0∞xz−1e−x dx.\Gamma(z)=\int_0^\infty x^{z-1}e^{-x}\,dx.

Differentiation at z=1z=1 gives

Γ′(1)=∫0∞e−xln⁡x dx=−γ.\Gamma'(1) =\int_0^\infty e^{-x}\ln x\,dx =-\gamma.

The digamma function is the logarithmic derivative

ψ(z)=Γ′(z)Γ(z).\psi(z)=\frac{\Gamma'(z)}{\Gamma(z)}.

Since Γ(1)=1\Gamma(1)=1,

ψ(1)=−γ.\psi(1)=-\gamma.

At positive integers,

ψ(n+1)=Hn−γ,\psi(n+1)=H_n-\gamma,

which extends the relationship between harmonic numbers and γ\gamma into the theory of special functions.

The constant also appears in the expansion

log⁡Γ(1+z)=−γz+∑m=2∞(−1)mζ(m)mzm,∣z∣<1.\log\Gamma(1+z) =-\gamma z+ \sum_{m=2}^{\infty} \frac{(-1)^m\zeta(m)}{m}z^m, \qquad |z|<1.

For complex zz, the logarithm here is the analytic branch near z=0z=0 whose value there is zero.

Zeta function near its pole

The zeta function has a simple pole at s=1s=1, and its local expansion begins

ζ(s)=1s−1+γ+O(s−1).\zeta(s)=\frac1{s-1}+\gamma+O(s-1).

Equivalently,

lim⁡s→1(ζ(s)−1s−1)=γ.\lim_{s\to1}\left(\zeta(s)-\frac1{s-1}\right)=\gamma.

Thus γ\gamma is the constant term remaining after the singular part is removed. In the wider Laurent expansion of ζ(s)\zeta(s), it is the zeroth Stieltjes constant.

Applications

Counting divisors

Let d(n)d(n) denote the number of positive divisors of the positive integer nn. A classical estimate is

∑n≤xd(n)=xln⁡x+(2γ−1)x+O(x).\sum_{n\le x}d(n) =x\ln x+(2\gamma-1)x+O(\sqrt{x}).

The leading term describes the main growth of the cumulative divisor count; the next term contains γ\gamma.

This relationship arises by counting integer lattice points satisfying ab≤xab\le x. Harmonic sums appear naturally when the count is organized according to one coordinate.

Probability and extreme values

If XX has the exponential distribution with rate 11, then its expected value after taking a logarithm is

E[ln⁡X]=∫0∞e−xln⁡x dx=−γ.\mathbb E[\ln X] =\int_0^\infty e^{-x}\ln x\,dx =-\gamma.

If X1,…,XnX_1,\ldots,X_n are independent exponential random variables with rate 11, their maximum satisfies

E ⁣[max⁡(X1,…,Xn)]=Hn.\mathbb E\!\left[\max(X_1,\ldots,X_n)\right]=H_n.

Consequently,

E ⁣[max⁡(X1,…,Xn)]−ln⁡n⟶γ.\mathbb E\!\left[\max(X_1,\ldots,X_n)\right]-\ln n \longrightarrow\gamma.

The limiting distribution of the centered maximum is the standard Gumbel distribution, whose cumulative distribution function is

F(x)=exp⁡(−e−x).F(x)=\exp(-e^{-x}).

Its mean is γ\gamma, providing a probabilistic interpretation of the same constant.

Arithmetic status

It remains unknown whether γ\gamma is a rational number or an irrational number. In particular, no proof establishes that its decimal expansion is nonrepeating.

It is also unknown whether γ\gamma is a transcendental number. Transcendence would imply irrationality, so this is a stronger unresolved question.

The occurrence of γ\gamma in formulas containing logarithms, the gamma function, or zeta values does not by itself determine its arithmetic nature. Likewise, extensive decimal calculations and rational approximations do not constitute a proof of irrationality.