Asymptotic expansion

Template:Short description

In mathematics, an asymptotic expansion, asymptotic series, or Poincaré expansion is a formal series used to approximate a given function near a specific point or at infinity. The term asymptotic series is commonly reserved for a divergent series whose terms initially decrease to a minimum magnitude and then increase without bound.Template:Sfn Template:Rp Although divergent, the truncated series provides an approximation with increasing accuracy as the function's argument approaches the asymptotic limit. These expansions are useful when a standard Taylor series converges too slowly, as an asymptotic series can often give high accuracy with only a few terms. An asymptotic expansion depends on the domain and limit point. A function of a real variable may require distinct series near the origin versus at infinity.Template:Sfn Template:Rp An entire function of a complex variable may require distinct asymptotic series for different sectors of the complex plane, a behavior called the Stokes phenomenon.Template:Sfn Template:Rp

The most common type of asymptotic expansion is a power series in either positive or negative powers. Asymptotic series commonly occur when using the Euler–Maclaurin summation formula and integral transforms such as the Laplace and Mellin transforms. Repeated integration by parts will often generate an asymptotic expansion.Template:Sfn Template:Rp

Asymptotic analysis plays an important role in singular perturbation theory and for solving nonlinear equations in fluid mechanics.Template:Sfn Template:Sfn Today, asymptotic series are used in the analysis of algorithms in computer science, the evaluation of complex integrals and partial sums, and the study of differential equations and difference equations.Template:Sfn Template:Sfn Template:Sfn

History

Early mathematicians considered asymptotic series unreliable due to their divergence. In a 1747 letter to John Canton, the Reverend Thomas Bayes observed that the Stirling series for the logarithm of the factorial function eventually diverges. Bayes maintained that asymptotic series were “not to be depended upon," arguing that any method generating an inexact result is flawed.Template:Sfn Leonhard Euler held a more insightful view, stating that "the sum of any infinite series is the finite expression, by the expansion of which the series is generated."Template:Sfn This view anticipated the mathematical concept of regularization, which assigns a definite value to a divergent asymptotic series.Template:Sfn Template:Rp

A major breakthrough occurred when George Stokes studied the Airy function. He discovered that although the Airy function is an entire function, it requires different asymptotic series in distinct sectors of the complex plane, an effect called the Stokes phenomenon. He observed that truncating a divergent series at its minimum term led to accurate approximation, a method that would later lead to superasymptotic expansions.Template:Sfn Template:Rp

In 1886, Henri Poincaré and Thomas Joannes Stieltjes independently established the mathematical foundations for asymptotic expansions.Template:Sfn Template:Sfn While Poincaré focused on their formal analytical properties for use in celestial mechanics, Stieltjes sought practical numerical approximations for specific functions and integrals.Template:Sfn Template:Rp Template:Sfn Stieltjes proved that the truncation error is bounded by the magnitude of the first omitted term. He introduced the half the least term rule, showing that adding half the least term to the sum of all preceding terms often gives an accurate approximation for alternating series. Stieltjes was able to estimate the remainder terms for several integral-defined functions.Template:Sfn Template:Rp

Modern efforts focus on how to further improve the accuracy of asymptotic approximations. Robert Balson Dingle introduced terminants, integrals used to assign a value to the series divergent tail.Template:Sfn Building on this foundation, Michael Berry and Christopher Howls developed hyperasymptotic expansions which give high accuracy approximations. Template:Sfn

Definitions

If $φ_{n}$ is a sequence of functions of asymptotic variable $x$ on some domain, and if $L$ is a limit point of the domain, then the sequence is an asymptotic scale if for every $n$ Template:Sfn Template:Rp

φ_{n + 1} (x) = o (φ_{n} (x)) (x \to L)

.

Here, $o$ is the little-o notation. Each function within the scale is called a gauge function.Template:Sfn Template:Rp Equivalently, each successive gauge function in the scale vanishes or grows strictly slower than its predecessor as $x \to L$ . Commonly, $L$ is taken to be zero or infinity. If $L$ is a finite number, it is called a distinguished point of the scale.Template:Sfn Template:Rp

If $f$ is a function on the domain of the asymptotic scale, then $f$ has a standard asymptotic expansion with respect to the scale as a formal series:Template:Sfn Template:Rp

\sum_{m = 0}^{n} a_{m} φ_{m} (x)

if

f (x) - \sum_{m = 0}^{n - 1} a_{m} φ_{m} (x) = O (φ_{n} (x)) (x \to L)

or the weaker conditionTemplate:Sfn Template:Rp

f (x) - \sum_{m = 0}^{n - 1} a_{m} φ_{m} (x) = o (φ_{n - 1} (x)) (x \to L)

is satisfied. Here, $O$ is the Big-O notation. If one or the other holds for all $n$ , this is indicated as: Template:Sfn Template:Rp

f (x) \sim \sum_{m = 0}^{\infty} a_{m} φ_{m} (x) (x \to L) .

A convergent series for any fixed $x$ approximates the function as the number of terms approaches infinity. In contrast, the asymptotic series for a fixed number of terms approximates the function as $x$ approaches the limit $L$ .

The function $f$ is a sum of its truncated asymptotic series, identified by its truncation index $n$ , and a remainder, $r_{n} (x)$ :

f (x) = \sum_{m = 0}^{n - 1} a_{m} φ_{m} (x) + r_{n} (x) .

This finite series approximates the function with a truncation error equal to the remainder. The series tail, $t_{n} (x)$ , is the remaining part of the series:

t_{n} (x) = \sum_{m = n}^{\infty} a_{m} φ_{m} (x) .

Regularization of an asymptotic series identifies the underlying function, $f (x)$ , that generates the asymptotic series expansion, assigning the series a unique value for each point in its domain.Template:Sfn Template:Rp

Properties

Unique expansion

The asymptotic expansion of a function $f (x)$ is unique for a specific region, scale and limit. The function has uniquely determined coefficients ${a_{n}}$ in the following way:Template:Sfn Template:Rp $\begin{matrix} a_{0} & = lim_{x \to L} \frac{f (x)}{φ_{0} (x)} \\ a_{1} & = lim_{x \to L} \frac{f (x) - a_{0} φ_{0} (x)}{φ_{1} (x)} \\ ⋮ \\ a_{n} & = lim_{x \to L} \frac{f (x) - \sum_{m = 0}^{n - 1} a_{m} φ_{m} (x)}{φ_{n} (x)} \end{matrix}$ where $L$ is the limit point of this asymptotic expansion (may be $\pm \infty$ ).

However, a function may have distinct expansions using distinct asymptotic scales. An example is:Template:Sfn Template:Rp $\begin{matrix} \frac{1}{(1 + x)} & \sim \sum_{m = 1}^{\infty} (- 1)^{m} x^{- m} (x \to \infty) \\ \frac{1}{(1 + x)} & \sim \sum_{m = 1}^{\infty} (x - 1) x^{- 2 m} (x \to \infty) \end{matrix}$

Sector of validity

In complex analysis, the limit of a function as it approaches a point in the complex plane may depend on the path's direction. Therefore, it is necessary to specify the sector of validity, a wedge-shaped region where the function has a specified limit. The sector of validity's angular width is called its opening angle and its boundaries are rays whose angles are given by the $\arg (x)$ function.Template:Sfn Template:Rp For complex-variable functions, a complete description of a divergent asymptotic series must specify its sector of validity, as this region varies by series.Template:Sfn Template:Rp

Nonunique function

If the function $f (x)$ has this power series expansion:

f (x) \sim \sum_{m = 0}^{\infty} a_{m} x^{m} (x \to 0),

adding an exponential term, $e^{- \frac{1}{x^{2}}}$ , leads to the same expansion as $x \to 0$ along the real axis:

f (x) + e^{- \frac{1}{x^{2}}} \sim \sum_{m = 0}^{\infty} a_{m} x^{m} (x \to 0) .

This occurs because the exponential function's own Taylor coefficients at the origin all vanish. Thus, two distinct functions can share the exact same asymptotic expansion.Template:Sfn Template:Rp

Subdominance

If two functions are equivalent $f (x) \sim g (x)$ as $x \to x_{0}$ within a sector of validity, their difference $δ (x) = f (x) - g (x)$ is subdominant to both functions. A function subdominant in one sector may become dominant in a neighboring sector, a process associated with the Stokes phenomenon.Template:Sfn Template:Rp.

Examples

Gamma function (Stirling's approximation): Template:Sfn

\frac{e^{x}}{x^{x} \sqrt{2 π x}} Γ (x + 1) \sim 1 + \frac{1}{12 x} + \frac{1}{288 x^{2}} - \frac{139}{51840 x^{3}} - \dots, (x \to \infty), for | \arg (x) | < π .

Exponential integral:Template:Sfn

x e^{x} E_{1} (x) \sim \sum_{n = 0}^{\infty} \frac{(- 1)^{n} n!}{x^{n}}, (x \to \infty), for | \arg (x) | < π .

Logarithmic integral for real variable $x$ :Template:Sfn

li (x) \sim \frac{x}{\ln x} \sum_{k = 0}^{\infty} \frac{k!}{(\ln x)^{k}}, (x \to \infty) .

Riemann zeta function: Template:Sfn

ζ (s) \sim \sum_{n = 1}^{N} n^{- s} + \frac{N^{1 - s}}{s - 1} - \frac{N^{- s}}{2} + N^{- s} \sum_{m = 1}^{\infty} \frac{B_{2 m} s^{\overline{2 m - 1}}}{(2 m)! N^{2 m - 1}}, for | \arg (s) | < π, Re (s) > - 2 n, n = 1, 2, 3, \dots

where

B_{2 m}

are Bernoulli numbers and

s^{\overline{2 m - 1}}

is a rising factorial. This expansion is valid for all complex

s

and is often used to compute the zeta function by using a large enough value of

N

, for instance

N > | s | .

Complementary error function:Template:Sfn

\sqrt{π} x e^{x^{2}} e r f c (x) \sim 1 + \sum_{n = 1}^{\infty} (- 1)^{n} \frac{(2 n - 1)!!}{(2 x^{2})^{n}} (x \to \infty), for | \arg (x) | < \frac{3 π}{4}

where

(2 n - 1)!!

is the double factorial.

Worked example

This example generates an asymptotic expansion for a function related to the exponential integral defined using the Cauchy principal value (PV):

e^{- 1 / t} Ei (\frac{1}{t}) = - e^{- 1 / t} PV \int_{- \frac{1}{t}}^{\infty} \frac{e^{- u}}{u} d u .

The substitution $w = u t + 1$ transforms this expression into:

e^{- 1 / t} Ei (\frac{1}{t}) = PV \int_{0}^{\infty} \frac{e^{- w / t}}{1 - w} d w .

Applying the finite geometric series formula to the integrand's denominator expresses the function as a sum of integrals:

e^{- 1 / t} Ei (\frac{1}{t}) = \sum_{m = 0}^{n - 1} \int_{0}^{\infty} w^{m} e^{- w / t} d w + PV \int_{0}^{\infty} \frac{w^{n} e^{- w / t}}{1 - w} d w .

Evaluating these integrals generates a finite series with factorial coefficients and an integral remainder:

e^{- 1 / t} Ei (\frac{1}{t}) = \sum_{m = 0}^{n - 1} m! t^{m + 1} + PV \int_{0}^{\infty} \frac{w^{n} e^{- w / t}}{1 - w} d w .

Although the series is divergent, truncating the series may provide a good approximation of the exponential integral for sufficiently small $t$ .

The full asymptotic expansion includes all series terms:Template:Sfn Template:Rp

e^{- 1 / t} Ei (\frac{1}{t}) \sim \sum_{m = 0}^{\infty} m! t^{m + 1} (t \to 0^{+}) .

Repeated integration by parts is an alternative method to derive this identical expansion.Template:Sfn Template:Rp Substituting $x = - \frac{1}{t}$ and noting that $Ei (x) = - E_{1} (- x)$ recovers the expansion given earlier in the article:

Ei (x) = \frac{e^{x}}{x} (\sum_{m = 0}^{n} \frac{m!}{x^{m}} + e_{n} (x)), e_{n} (x) \equiv (n + 1)! x e^{- x} \int_{- \infty}^{x} \frac{e^{t}}{t^{n + 2}} d t .

For any fixed $x$ , the absolute value of the error term $| e_{n} (x) |$ decreases, then increases as $n$ increases. The minimum occurs at $n \sim | x |$ , at which point $| e_{n} (x) | \leq \sqrt{\frac{2 π}{| x |}} e^{- | x |}$ . Exponentially small truncation errors are common for optimally truncated series.Template:Sfn Template:Rp

Asymptotic expansions may arise when a series is integrated beyond its boundary of convergence.Template:Sfn Template:Rp

Regularization methods

Optimal truncation

These regularization methods assign a value to an asymptotic series using truncated series.

The superasymptotic expansion is the truncated series which minimizes the magnitude of the remainder term. A remainder formula or remainder upper bound formula may determine the optimal truncation index $n$ . If an exact formula is unavailable, heuristic methods may determine the optimum truncation index such as truncating the series immediately before or at its minimum term (summation to the least term), or using formulas that determine the optimal index $n$ from the value of the asymptotic variable $x$ .Template:Sfn Template:Sfn Template:Sfn Template:Sfn

A term $δ (x)$ is said to lie beyond-all-orders of an asymptotic expansion if the term is asymptotically smaller than every gauge function $φ_{n} (x)$ of the expansion:

δ (x) = o (φ_{n} (x)) (x \to L) .

Commonly, an optimally truncated power series expansion at the origin has a beyond-all-orders truncation error of order $O (e^{- q / x})$ with positive constant $q$ as $x \to 0$ in the sector $| \arg (x) | < π / 2$ . At the origin, this truncation error is non-analytic and cannot be expressed as a Taylor series.Template:Sfn Template:Rp The term asymptotics beyond-all-orders refers to the collection of methods used to measure the effects of these beyond all order terms.Template:Sfn Template:Rp

For an alternating series, the remainder is often approximately half of the series' least term. Adding half of the least term to the sum of the preceding terms often generates a highly accurate approximation. Template:Sfn Template:Rp

The hyperasymptotic expansion replaces the remainder of a superasymptotic expansion with an optimally truncated second series and second smaller remainder, often using a different scale. Repeating this process generates exponentially smaller remainder terms.Template:Sfn Template:Rp

Summation of the series tail

These regularization methods compute a series remainder by applying Borel summation or Mellin-Barnes regularization to the series tail. The expansion's underlying function is the sum of this computed remainder and the truncated series.

Borel summation

When a truncated series has a factorially divergent tail, Borel summation evaluates the remainder as the product of the tail's first term and an integral called a terminant.

For an alternating-sign tail:

\sum_{m = n}^{\infty} Γ (m + α + 1) (- x)^{m}, 0 \leq α < 1, | \arg (x) | < π,

Borel summation generates a type I terminant $Λ_{s} (x)$ , defined by a Cauchy principal value integral:

Λ_{s} (x) = \frac{1}{Γ (s + 1)} PV \int_{0}^{\infty} \frac{t^{s} e^{- t}}{1 + x t} d t .

Multiplying this terminant by the tail's first term generates the exact remainder:Template:Sfn Template:Rp

Γ (n + α + 1) (- x)^{n} Λ_{n + α} (x) .

For a same-sign tail:

\sum_{m = n}^{\infty} Γ (m + α + 1) x^{m}, 0 \leq α < 1,

the positive real axis forms a Stokes line. Borel summation generates a type II terminant ${\bar{Λ}}_{s} (- x)$ with the Stokes constant $S$ :Template:Sfn Template:Rp

\begin{matrix} {\bar{Λ}}_{s} (- x) = Λ_{s} (- x) + 2 S \frac{i π x^{- s - 1} e^{- 1 / x}}{Γ (s + 1)}, \\ S = {\begin{matrix} + \frac{1}{2}, & if 0 < \arg (x) < 2 π, \\ 0, & if \arg (x) = 0, \\ - \frac{1}{2}, & if - 2 π < \arg (x) < 0. \end{matrix} \end{matrix}

The resulting remainder is the product:Template:Sfn Template:Rp

Γ (n + α + 1) x^{n} {\bar{Λ}}_{n + α} (- x) .

For more general divergent tails, a linear combination of late-series terms and terminants, called a distribution over basic terminants, represents the remainder.Template:Sfn Template:Rp

Mellin–Barnes tail regularization

Mellin–Barnes regularization evaluates the series tail separately for alternating-sign coefficients and same-sign coefficients.

For an alternating-sign series tail:

\sum_{m = n}^{\infty} f (m) (- x^{β})^{m},

Mellin–Barnes regularization computes this remainder:Template:Sfn Template:Rp

\int_{c - i \infty}^{c + i \infty} \frac{x^{β s} f (s)}{e^{- i π s} - e^{i π s}} d s, \frac{- π - ε_{1}}{β} < \arg (x) < \frac{π + ε_{2}}{β},

where $n - 1 < c = Re (s) < n$ .Template:Sfn Template:Rp The asymptotic behavior of the integrand as $x$ approaches the limits of integration determines the constants $ε_{1}$ and $ε_{2} .$ Template:Sfn Template:Rp

For a same-sign series tail:

\sum_{m = n}^{\infty} f (m) (x^{β})^{m},

the positive real axis forms a Stokes line. For the sector above the Stokes line, the method generates this remainder:

\int_{c - i \infty}^{c + i \infty} \frac{x^{β s} e^{i π s} f (s)}{e^{- i π s} - e^{i π s}} d s + \frac{1}{2} \int_{c - i \infty}^{c + i \infty} x^{β s} f (s) d s, 0 < \arg (x) < \frac{2 π - B}{β},

where $n - 1 < c = Re (s) < n .$

For the sector below the Stokes line, the remainder becomes:

\int_{c - i \infty}^{c + i \infty} \frac{x^{β s} e^{- i π s} f (s)}{e^{- i π s} - e^{i π s}} d s - \frac{1}{2} \int_{c - i \infty}^{c + i \infty} x^{β s} f (s) d s, \frac{- 2 π + A}{β} < \arg (x) < 0,

where $n - 1 < c = Re (s) < n .$ Template:Sfn Template:Rp The asymptotic behavior of the integrand as $x$ approaches the limits of integration determines the constants $A$ and $B .$ Template:Sfn Template:Rp Additional integrals, convergent in overlapping sectors, allow analytic continuation in the complex plane for both types of series.Template:Sfn Template:Rp

Summation of the full series

This regularization method applies Borel summation or Mellin-Barnes regularization to the entire series. Computing the Borel sum of a typical Gevrey order 1 asymptotic series requires a Borel transform, analytic continuation, and a Laplace transform.

The Borel transform converts this divergent power series with zero radius of convergence into a convergent series with a nonzero radius of convergence called the Borel transformed series.Template:Sfn A series acceleration method can accurately sum a slow converging series.Template:Sfn

The analytic continuation of the Borel transformed series is the Borel transformed function. The Borel transformed function's domain now extends across the complex plane, excluding the function's singularities. Comparing the transformed series coefficients to established tables of function expansions and coefficient sequences may identify the analytic continuation as a special function, an integral representation, or a generating function.Template:Sfn Template:Rp Template:Sfn Template:Sfn Template:Sfn An alternative approach uses rational function approximation of the transformed series, guided by singularity analysis and conformal mapping.Template:Sfn Template:Sfn A bijective conformal map (uniform map) is especially effective for accurate approximation.Template:Sfn In comparison to the Padé approximant, the Adaptive Antoulas–Anderson (AAA) algorithm is numerically more stable and less prone to generating spurious poles.Template:Sfn

Finally, the Laplace transform converts the Borel transformed function to the Borel sum, a finite value assigned to the original divergent series.Template:Sfn

Convergent alternative expansions

Series such as inverse factorial series (or Nörlund series), binary rational inverse factorial series, or modified Hadamard series are alternatives to asymptotic expansions because they may converge for functions with divergent asymptotic series.Template:Sfn

The inverse factorial series (or Nörlund series) with coefficients $c_{m}$ has the following form:Template:Sfn Template:Sfn $f (x) = c_{0} + \sum_{m = 1}^{\infty} \frac{c_{m}}{x (x + 1) (x + 2) \dots (x + m - 1)} .$

A binary rational inverse factorial series uses a double summation over the coefficients $c_{k, m}$ ; it has the following form:Template:Sfn $f (z) = \sum_{k = 1}^{\infty} \sum_{m = 0}^{\infty} \frac{c_{k, m}}{2^{k m}} \frac{m!}{(2^{k} z + 1)^{m + 1}} .$ This binary rational variant typically converges more rapidly than the standard inverse factorial series.Template:Sfn

A modified Hadamard series is a convergent series designed to overcome the divergence or slow convergence of traditional asymptotic expansions. Each term in the series has a normalized incomplete gamma function that can gradually decrease the magnitude of later terms. Selecting optimal integration intervals and rearranging how later terms are summed leads to high accuracy.Template:Sfn Template:Rp

Notes

Template:Reflist

References

Template:Refbegin

Template:Refend

External links

Template:Authority control

Asymptotic expansion

Contents

History

Definitions

Properties

Unique expansion

Sector of validity

Nonunique function

Subdominance

Examples

Worked example

Regularization methods

Optimal truncation

Summation of the series tail

Borel summation

Mellin–Barnes tail regularization

Summation of the full series

Convergent alternative expansions

See also

Related fields

Asymptotic methods

Notes

References

External links

Navigation menu

Asymptotic expansion

History

Definitions

Properties

Unique expansion

Sector of validity

Nonunique function

Subdominance

Examples

Worked example

Regularization methods

Optimal truncation

Summation of the series tail

Borel summation

Mellin–Barnes tail regularization

Summation of the full series

Convergent alternative expansions

See also

Related fields

Asymptotic methods

Notes

References

External links

Navigation menu

Search