Goodwin–Staton integral

From testwiki
Jump to navigation Jump to search

Template:Short description

In mathematics, the Goodwin–Staton integral is the special function [1]

G(z)=∫0∞e−t2t+zdt,|arg⁡z|<π.

The restriction on the argument excludes the non-positive real axis, where the denominator can vanish along the path of integration. The integral is named after E. T. Goodwin and J. Staton, who published a numerical table of the corresponding real integral in 1948.[2]

Relation to other special functions

For positive real x, the Goodwin–Staton integral is related to Dawson's integral F and the exponential integral by[3]

G(x)=πF(x)−12e−x2Ei(x2),x>0.

Differential equations

Differentiating the defining integral and integrating by parts gives the first-order inhomogeneous differential equation

G′(z)+2zG(z)=π−1z.

Differentiating this relation twice and eliminating the inhomogeneous term gives the third-order linear homogeneous equation

zG‴(z)+(2+2z2)G″(z)+8zG′(z)+4G(z)=0.

Properties on the positive real axis

For x>0, differentiation under the integral sign gives

G(n)(x)=(−1)nn!∫0∞e−t2(t+x)n+1dt.

Hence

(−1)nG(n)(x)>0,n=0,1,2,…,

so G is a completely monotonic function on (0,∞). In particular, G is positive, strictly decreasing, and convex there.

Series near the origin

D. S. Jones studied the generalized Goodwin–Staton integral and obtained convergent series representations.[4] Specializing Jones's equation (2.11) gives, for |arg⁡z|<π,

G(z)=e−z2(−log⁡z−γ2−12∑m=1∞z2mm!m+π2∑m=0∞z2m+1m!(m+12)),

where log⁡z is the principal logarithm and γ is the Euler–Mascheroni constant.

Consequently, as z→0 within the cut plane,

G(z)=−log⁡z−γ2+πz+(log⁡z+γ2−12)z2−2π3z3+(38−γ4−12log⁡z)z4+O(z5).

In particular,

G(x)∼−log⁡x−γ2(x→0+).

Asymptotic expansion

For large positive x, the integral has the asymptotic expansion[4]

G(x)∼12∑n=0∞(−1)nΓ(n+12)x−n−1(x→+∞).

Thus its first terms are

G(x)∼π2x−12x2+π4x3−12x4+3π8x5−1x6+⋯.

The NIST Digital Library of Mathematical Functions refers to Olver for corresponding expansions with bounds on the remainder for both real and complex values of the argument.[5]

Generalized Goodwin–Staton integral

Jones considered the generalized integral[4]

I(μ,z)=∫0∞tμe−t2t−zdt,

where μ is real with μ>−1 and z lies off the positive real axis. The ordinary Goodwin–Staton integral is the special case

G(z)=I(0,−z).

The generalized integral is a Stieltjes transform and satisfies the recurrence

I(μ+1,z)=12Γ(μ+12)+zI(μ,z),

as well as

∂I(μ,z)∂z=μI(μ−1,z)−2I(μ+1,z),μ>0.

Jones derived convergent series, asymptotic expansions, and explicit error bounds, and also treated oscillatory and logarithmic variants.[4] Mamedov later gave another evaluation of generalized Goodwin–Staton integrals using the binomial expansion theorem.[6]

Numerical evaluation

For positive real arguments, numerical evaluation can be reduced to Dawson's integral and the exponential integral through the relation above.[3] For small and moderate arguments, Jones's convergent series can be used, while for large arguments the asymptotic expansion is more suitable.[4] The DLMF section on computation of the Goodwin–Staton integral cites the original table of Goodwin and Staton.[7]

See also

References

Template:Reflist

Further reading