Rational Exponential Integrals

• Rational exponential integrals are defined as integrals in which the kernel consists of an exponential function composed with a rational power multiplied by a rational function.
• Their closed-form representations using generalized hypergeometric functions allow for precise asymptotic analysis and reductions to elementary cases under specific parameter conditions.
• Recent advances employ these integrals in rational approximation frameworks, aiding developments in analytic number theory, special functions, and mathematical physics.

A rational exponential integral, in the most general sense, is any integral in which the kernel involves the exponential function composed with a rational power of the variable, multiplied by a rational function or, more specifically, a power function with rational exponent. This class encompasses proper generalizations of the classical exponential integral function and includes nearly all “non-elementary integrals” involving $e^{\lambda x^\beta}$ and $x^{-\alpha}$ as special or degenerate cases. Their closed forms, analytic structure, reduction to elementary or special functions, and role in explicit rational approximation are central in recent advances in both special function theory and transcendental number theory.

1. Definition and General Structure

A prototypical rational exponential integral is given by

$$\mathrm{Ei}_{\beta,\alpha}(x) = \int e^{\lambda x^{\beta}} x^{-\alpha}\,dx,$$

where $\beta \in \mathbb{Q}_{>0}$, $\alpha \in \mathbb{Q}$, and $\lambda$ is a complex—or frequently real—parameter. The integration may be indefinite (antiderivative) or definite over an interval abutting $0$ or $\infty$, with analytic and convergence constraints as dictated by the exponents (Nijimbere, 2017).

These integrals unify numerous classical cases, for instance:

• The “ordinary” exponential integral: $$\mathrm{Ei}(x) = \int_{-\infty}^x \frac{e^t}{t}\,dt$$, as $\mathrm{Ei}_{1,1}(x)$;
• Variants with algebraic powers, e.g., $\int e^{\lambda x^\beta}\,dx$;
• Integrals with generalized denominators.

Their significance lies in their closed-form expression in terms of generalized hypergeometric functions, the emergence of exact reductions for special rational $(\beta,\alpha)$, their explicit asymptotics, and their capacity for algorithmic rational approximation and recurrence analysis (Nijimbere, 2018, Wolfs et al., 2024).

2. Explicit Closed-Form Representations

A uniform closed form for all $\beta\ge1$, $\alpha \in \mathbb{R}$, and $\lambda\in \mathbb{C}$ is given by (Nijimbere, 2017, Nijimbere, 2018): $\mathrm{Ei}_{\beta,\alpha}(x) = \frac{x^{1-\alpha}}{1-\alpha} + \frac{\lambda x^{\beta-\alpha+1}}{\beta-\alpha+1}\,_2F_2\left( \begin{array}{cc} 1, & -\frac{\alpha}{\beta} + \frac{1}{\beta} + 1; \ 2, & -\frac{\alpha}{\beta} + \frac{1}{\beta} + 2; \end{array} \lambda x^{\beta} \right) + C$ where $_2F_2$ is the generalized hypergeometric function. This form is valid for $\beta\ge1$ and $\alpha < \beta + 1$, ensuring absolute convergence of the expansion.

A more general series expansion, comprising both “finite sum” (for negative integer exponent terms) and a hypergeometric tail, is available when $\alpha>\beta+1$: $\mathrm{Ei}_{\beta,\alpha}(x) = \frac{x^{1-\alpha}}{1-\alpha} + \sum_{j=0}^{m-1}\frac{\lambda^j}{j!}\frac{x^{\beta j-\alpha+1}}{\beta j-\alpha+1} + \frac{\lambda^m x}{(m+1)!(\beta+1)}\,_2F_2\left( 1, 1+\frac{1}{\beta}; m+2, 2+\frac{1}{\beta}; \lambda x^\beta \right) + C,$ where $m = (\alpha-1) / \beta$ is an integer and $C$ is the constant of integration (Nijimbere, 2018).

Series forms, power sums, and elementary functions arise as special or degenerate cases when the upper parameter $-\frac{\alpha}{\beta} + \frac{1}{\beta} + 1$ is a nonnegative integer, effecting series termination (Nijimbere, 2017).

3. Special and Degenerate Cases

The general closed form specializes as follows:

• Classical $\mathrm{Ei}(x)$: Setting $\beta = 1$, $\alpha = 1$, $\lambda = 1$, the general closed form regularizes at $\alpha=1$, giving

$\mathrm{Ei}(x) = \ln x + x\,_2F_2(1,1;2,2;x) + C.$

• Elementary Reductions: When $\beta$ and $\alpha$ are rational and $-\frac{\alpha}{\beta} + \frac{1}{\beta} + 1$ is a nonnegative integer, $_2F_2$ reduces to elementary functions, exponentials, or incomplete gamma functions. For example,
  • $$\int e^{\lambda\sqrt{x}} x^{-1/2}\,dx = (2/\lambda) \, e^{\lambda\sqrt{x}} + C$$,
  • $$\int e^{\lambda x^2} x^{-1} dx = \frac{1}{2}\,\mathrm{Ei}(\lambda x^2) + C$$ (Nijimbere, 2017).
• Asymptotics: For $|x|\to\infty$ and $\Re(\lambda x^{\beta})>0$,

$$\begin{split} \mathrm{Ei}_{\beta,\alpha}(x) \sim &\,\, \frac{x^{1-\alpha}}{1-\alpha} + (\text{const}) - \frac{x^{1-\alpha}}{\beta} + \frac{e^{\lambda x^\beta} x^{1-\beta-\alpha}}{\beta\lambda} + O\left(e^{\lambda x^\beta}x^{-\beta-\alpha}\right) \end{split}$$

with the constant involving $\lambda$ and $\Gamma$-functions. The exponential term dominates for large $|\lambda x^{\beta}|$ (Nijimbere, 2017).

4. Rational Parameter Integrals and Reductions

For definite integrals involving the exponential integral of the second kind, $E_\nu(z)$, an explicit closed form arises: $$I(\lambda,\mu,\nu;z) = \int_0^z x^\lambda E_\nu(x^\mu)\,dx = \frac{ \gamma( \frac{1+\lambda}{\mu}, z^\mu ) + z^{1+\lambda} E_\nu(z^\mu) }{ 1+\lambda+\mu(\nu-1) }$$ with parameter and domain constraints ensuring convergence (Ciotti, 2020).

If $(\lambda,\mu)$ are rational, then $(1+\lambda)/\mu$ is rational; accordingly, for half-integer or integer values, $\gamma(a,x)$ and $E_\nu(w)$ reduce to expressions with error functions, exponentials, and radicals. This gives rise to “rational” exponential integrals in the classical tables.

Two explicit examples:

• $I(0,1,\nu;z) = \frac{1-e^{-z} + z E_\nu(z)}{\nu}$,
• $$I(0,2,\nu;z) = \frac{\sqrt{\pi}\,\mathrm{Erf}(z) + z E_\nu(z^2)}{2\nu-1}$$.

This generalizes classical integrals and is constructive: all rational-exponential combinations with power-law kernels reduce to this paradigm (Ciotti, 2020).

5. Contiguous Relations, Transformations, and Analytic Domains

Hypergeometric representations of rational exponential integrals admit contiguous relations. Euler-type identities relate $_2F_2$ with arguments $\pm i\lambda x^{\beta}$ to $_1F_2$ and $_2F_3$ parametric families (see Theorem 9, (Nijimbere, 2017)). Sine and cosine analogues thus collapse to exponential integrals under complexification.

The domain of validity for $_2F_2$ forms covers all complex $x$, up to the branch cut induced by $x^{1-\alpha}$. The principal branch is fixed by convention, and analytic continuation across cuts is standard via generalized hypergeometric functional equations (Nijimbere, 2017, Nijimbere, 2018).

For definite integrals over a closed interval, convergence at the lower limit is dictated by: $\lambda > \max\{ -1, -1-\mu(\nu-1) \}$ for $I(\lambda,\mu,\nu;z)$, ensuring integrability as $x\to0^+$ (Ciotti, 2020).

6. Rational Approximation and Orthogonal Polynomial Frameworks

Recent advances construct simultaneous rational approximants to Euler’s constant, the (generalized) exponential integral, and related special values using families of mixed-type multiple orthogonal polynomials associated to the exponential integral (Wolfs et al., 2024). These polynomials are explicitly constructed as sums involving binomial squares and harmonic numbers, enjoying five-term recurrences: $$F_{n;2}(x) = \sum_{k=0}^n \binom{n}{k}^2 x^{k}, \quad F_{n;1}(x) = \sum_{k=0}^n (-3H_k+2H_{n-k})\binom{n}{k}^2 x^{k},$$ producing rational approximants $r_n(x) = F_{n;1}(x)/F_{n;2}(x)$ to $\gamma + \ln x$ with explicit sub-exponential error bounds.

In this setting, the error terms and denominators possess sub-exponential growth, as characterized by Birkhoff–Trjitzinsky theory: $$F_{n;2}(x) \sim n^{-9/8}\exp\left(4x^{1/4}n^{3/4} - x^{1/2}n^{1/2} + x^{3/4}n^{1/4}\right).$$ The error decay, $|Q_n\gamma - P_n|$, is similarly sub-exponential, achieving rates surpassing prior Aptekarev or Rivoal constructions (Wolfs et al., 2024).

Nonetheless, the sub-exponential growth of denominators precludes current irrationality proofs via linear-independence criteria, though these methods unify a wide range of exponential special functions and their rational approximations (Wolfs et al., 2024).

7. Applications and Extensions

Rational exponential integrals appear in:

• Special function theory: Treatment of non-elementary integrals encountered in analysis and mathematical physics, exact evaluation of parametric integrals, and explicit representations in terms of hypergeometric and gamma functions.
• Stellar dynamics: Multi-dimensional integrals featuring exponential and error function kernels, e.g., those arising in generalized Chandrasekhar dynamical friction models (Ciotti, 2020).
• Diophantine approximation: Construction of rational approximants to constants such as $\gamma$ or $G = e E_1(1)$ with recursions and error bounds informed by multiple orthogonal polynomial structures (Wolfs et al., 2024).
• Analytic number theory: Rational approximants extend to digamma functions $\psi(\alpha+1)$ and higher exponential integrals via mixed-type Nikishin systems and generalized hypergeometric expansions, with potential implications for transcendence and irrationality investigations (Wolfs et al., 2024).

The methodical connection between incomplete gamma functions, exponential integrals of the second kind, and analytic continuation schemes unifies these rational exponential integrals across foundational domains of mathematics and mathematical physics.