Convergence criteria for Frullani-type integrals involving differences of cosines
2026
Read Convergence criteria for Frullani-type integrals involving differences of cosines (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-26153.txt.
Introduction
Atiratch Laoharenoo and Chanatip Sujsuntinukul settle exactly when a family of integrals built from powers of cosine differences converge or diverge. For whole-number powers and real-valued shifts, they map every parameter range and give exact values in all cases where the integrals converge. The same method yields a family of counting identities from sine-and-cosine series. It extends to a matching class of integrals built from sine differences.
Catalog record
| YEAR | 2026 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47467 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.26153
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Convergence criteria for Frullani-type integrals involving differences of cosines Abstract For p , q ∈ ℕ p,q\in\mathbb{N} and α , β ∈ ℝ \alpha,\beta\in\mathbb{R} , we investigate the family of improper integrals ∫ 0 ∞ ( cos α x − cos β x ) p x q 𝑑 x . \int_{0}^{\infty}\frac{(\cos\alpha x-\cos\beta x)^{p}}{x^{q}}dx. We establish a complete classification of the parameter ranges ( p , q , α , β ) (p,q;\alpha,\beta) for which the integrals converge or diverge, and we derive explicit closed-form evaluations in all convergent cases. The analysis also reveals a family of combinatorial identities arising naturally from coefficients in the trigonometric power expansions. As a further application of the same method, we study an analogous class of integrals involving powers of sine differences. This extends the work of Laoharenoo and Boonklurb in 2022. Atiratch Laoharenoo and Chanatip Sujsuntinukul Keywords: Frullani integral, Improper integral, Cosine function, Sine function, Abel’s summation, Combinatorial identity Mathematics subject classification (2020): 05A19, 26A06, 26A09. Introduction In the early 19th century, an Italian mathematician Frullani introduced an interesting integral which relates a “difference under scaling” to a logarithmic factor [ 8 ] . In its standard form, if f\mathrel{\mathop{\ordinarycolon}}[0,\infty)\to\mathbb{R} is continuous, has a finite limit at infinity f(\infty)\mathrel{\mathop{\ordinarycolon}}=\lim_{x\to\infty}f(x) , and \alpha,\beta>0 , then (assuming convergence) one has 1 1 1 Throughout this paper, \log refers to the natural logarithm (some prefer to use \ln instead). \int_{0}^{\infty}\frac{f(\alpha x)-f(\beta x)}{x}dx=\bigl(f(\infty)-f(0)\bigr)\log\frac{\alpha}{\beta}. If f(\infty) does not exist, but \int_{\gamma}^{\infty}(f(x)/x)dx exists for some \gamma>0 , then one has the alternative form \int_{0}^{\infty}\frac{f(\alpha x)-f(\beta x)}{x}dx=-f(0)\log\frac{\alpha}{\beta}. This identity has served as a prototype for many extensions in which one seeks either broader hypotheses guaranteeing convergence and evaluation, or explicit closed forms for particular choices of f . For instance, a notable direction replaces pointwise information about f(x) as x\to\infty by averaged information. In 1949, Ostrowski [ 2 ] introduced the mean value \displaystyle M(f)\mathrel{\mathop{\ordinarycolon}}=\lim_{x\to\infty}\frac{1}{x}\int_{0}^{x}f(t)\,dt, (1.1) and showed that, under suitable assumptions ensuring both existence and convergence, the same Frullani integral can be expressed in terms of M(f) via \int_{0}^{\infty}\frac{f(\alpha x)-f(\beta x)}{x}\,dx=M(f)\log\frac{\alpha}{\beta}. These integrals of Frullani-type have, in fact, attracted the interest of many mathematicians (including Ramanujan and Hardy, among others) during the mid 20th century and in recent years. See, for example, [ 3 ] , [ 9 ] , [ 10 ] , [ 11 ] , [ 12 ] . One recurring theme is understanding how oscillatory structure (such as taking f to be sine or cosine functions) and modified powers in the numerator or denominator affect both convergence and the possibility of closed-form evaluation. More recently, Laoharenoo and Boonklurb [ 1 ] , in 2022, introduced a family of Frullani-type integrals built from trigonometric oscillations, \displaystyle\mathcal{I}_{p,q}(\alpha,\beta)\mathrel{\mathop{\ordinarycolon}}=\int_{0}^{\infty}f_{p,q}[\alpha,\beta](x)\,dx, (1.2) where f_{p,q}[\alpha,\beta](x)\mathrel{\mathop{\ordinarycolon}}=\frac{(\cos\alpha x-\cos\beta x)^{p}}{x^{q}}. In this case, we allow p,q\in\mathbb{N} and \alpha,\beta\in\mathbb{R} , |\alpha|\neq|\beta| (Indeed, since the cosine function is even, it suffices to consider when \alpha,\beta\geq 0 .). The special case p=q=2 appears (in a different context) in Hardy’s work [ 5 ] . In addition, not so long ago, a similar class of such improper integrals has also been studied using the Mellin transform [ 6 ] . Beyond these connections, the general family seems not to have been systematically analysed, and in particular it is not recorded in the large compendium [ 7 ] . These integrals appear to be somewhat challenging to study. First, they exhibit singular behaviour at both endpoints of integration: the integrand has a potential singularity at x=0 , whilst the upper limit extends to infinity. Establishing convergence thus requires careful analysis of both singularities. Moreover, extracting explicit values is nontrivial, due to the complexity of the oscillatory terms involved. In [ 1 ] , they were able to compute the exact values of ( 1.2 ) for the cases where p\geq q and q is even. However, the remaining cases are yet to be investigated. Based on the discussion above, this paper provides a complete analysis of the family of integrals in ( 1.2 ). After collecting the necessary preliminaries in Section 2 , we first give a full classification of convergence in Section 3 . Owing to the atypical singular behaviour of f_{p,q}[\alpha,\beta] , the treatment of the case q=1 differs somewhat from that of q\geq 2 . Consequently, in Section 4 , we derive explicit closed-form values for \mathcal{I}_{p,q}(\alpha,\beta) in all convergent cases. In Section 5 , we will look at the special case when \alpha=0 . Interestingly, these values turn out to exhibit some interesting combinatorial patterns. Finally, in Section 6 , building on the same theory, we also roughly study the convergence classification of the family similar to ( 1.2 ), but the cosine is replaced by the sine. Preliminaries In this section, we collect several lemmas that will be important to our analysis later on. The first two lemmas can be found in the literature. Lemma 2.1 . [ 7 ] Let \alpha,\beta,\gamma\in\mathbb{R} such that \alpha,\beta\neq 0 . Then we have \int_{0}^{\infty}\frac{\cos\alpha x-\cos\beta x}{x}dx=\log\left|\frac{\beta}{\alpha}\right|\quad\text{and}\quad\int_{0}^{\infty}\frac{\sin\gamma x}{x}dx=\begin{cases}\displaystyle\frac{\pi}{2}\text{sgn}(\gamma)&\text{if }\gamma\neq 0,\\ 0&\text{if }\gamma=0.\end{cases} In this case, for t\in\mathbb{R} , we define the signum function by \text{sgn}(t)\mathrel{\mathop{\ordinarycolon}}=\begin{cases}1&\text{if }t>0,\\ 0&\text{if }t=0,\\ -1&\text{if }t<0.\end{cases} Lemma 2.2 ( [ 1 ] , Lemma 4.2) . Let f\mathrel{\mathop{\ordinarycolon}}\mathbb{R}\to\mathbb{R} be a smooth function for which x=0 is a zero of order k for some k\geq 2 . Suppose there exists an integer 1\leq\ell\leq k such that \int_{0}^{\infty}\frac{f^{(\ell-1)}(x)}{x}dx exists, and f^{(m)} is bounded on \mathbb{R} for 0\leq m\leq\ell-2 . Then \int_{0}^{\infty}\frac{f(x)}{x^{\ell}}dx=\frac{1}{(\ell-1)!}\int_{0}^{\infty}\frac{f^{(\ell-1)}(x)}{x}dx. The next three lemmas can be shown easily. Lemma 2.3 . Let \alpha\in[0,1] and x\in[0,\pi/2] . Then we have \displaystyle x\sin x+2\cos x\leq\alpha x\sin\alpha x+2\cos\alpha x. (2.1) Proof. First we record the elementary inequality 2 2 2 This can be proven using the mean value theorem. x\cos x\leq\sin x for x\in[0,\pi/2] . Now fix x\in[0,\pi/2] and let g_{x}(\alpha)\mathrel{\mathop{\ordinarycolon}}=2\cos\alpha x+\alpha x\sin\alpha x for \alpha\in(0,1] . Then g_{x}^{\prime}(\alpha)=x(\alpha x\cos\alpha x-\sin\alpha x)\leq 0 , since \alpha x\in[0,\pi/2] , showing g_{x} is decreasing on \alpha\in(0,1] . Hence ( 2.1 ) holds for \alpha\in(0,1] . For \alpha=0 , we just let \alpha\to 0^{+} and apply continuity. ∎ Lemma 2.4 . Let 0\leq\alpha\leq 1 and p,q\in\mathbb{N} with q>2p>0 . Then f_{p,q}[\alpha,1](x) is positive and decreasing on (0,\pi/2] . Proof. First we see that the function g(x)\mathrel{\mathop{\ordinarycolon}}=\frac{\cos\alpha x-\cos x}{x^{2}} is decreasing on (0,\pi/2] due to Lemma 2.3 and g^{\prime}(x)=\frac{1}{x^{3}}(x\sin x+2\cos x-\alpha x\sin\alpha x-2\cos\alpha x)\leq 0. Since g(x)\geq 0 for x\in(0,\pi/2] , thus implies g^{p} is decreasing on (0,\pi/2] . Next as the function 1/x^{q-2p} is positive and decreasing on (0,\pi/2] , the product \frac{g^{p}(x)}{x^{q-2p}}=\frac{(\cos\alpha x-\cos x)^{p}}{x^{q}} is positive and decreasing on (0,\pi/2] as desired. ∎ Remark 2.5 . Although we primarily care about p,q\in\mathbb{N} , the result in fact holds for p,q\in\mathbb{R} . Lemma 2.6 . Let \alpha,\beta\in\mathbb{R} such that |\alpha|\neq|\beta| . If p,q\in\mathbb{N} with q\leq 2p , then we have 3 3 3 Here and in the sequel, we use the usual Landau’s asymptotic notation X=O(Y) to denote the bound |X|\leq CY for some absolute constant C . f_{p,q}[\alpha,\beta](x)=O(1) as x\to 0 . Proof. This is obvious by considering the Taylor expansion around x=0 of \displaystyle(\cos\alpha x-\cos\beta x)^{p}=\left(\sum_{n=1}^{\infty}(-1)^{n+1}\frac{(\beta^{2n}-\alpha^{2n})}{(2n)!}x^{2n}\right)^{p}=\left(\frac{\beta^{2}-\alpha^{2}}{2}\right)^{p}x^{2p}+O(x^{2p+2}). Hence, as x\to 0 , f_{p,q}[\alpha,\beta]\to((\beta^{2}-\alpha^{2})/2)^{p} if q=2p and f_{p,q}[\alpha,\beta]\to 0 if q<2p . ∎ Convergence Classification We first classify