5
$\begingroup$

Let's consider the affinely extended real line. The functions that have a limit on positive or negative infinity $\lim_{x\to+\infty} f(x)$ or $\lim_{x\to-\infty} f(x)$ can be generalized to the values at infinity $f(+\infty)$, $f(-\infty)$ based on these limits.

But what about functions that do not have such limits?

For instance, if we take these integrals following Borel (or Abel) generalization, we get:

$\int_0^\infty \cos x=0$

and

$\int_0^\infty \sin x=1$

Since integral of $\cos x$ is $\sin x$ and integral of $\sin x$ is $-\cos x$, we can probably generalize these functions to infinity (assuming the fundamental theorem of calculus should still hold): $\sin (\infty)=\cos(\infty)=0$ (notice, this would apparently break the main trigonometric identity $\sin^2 x+\cos^2 x =1$ at this point but one of the books linked in the comments resolves the paradox by claiming that $(\sin \infty)^2\ne \sin^2 \infty$ and the same for cosine).

Still, I wonder whether there were other attempts at such generalization? Possibly, something along Cesaro lines (limit of mean value of the function) or sometyhing else?

$\endgroup$
  • 1
    $\begingroup$ See the notion of "Banach limit" en.wikipedia.org/wiki/Banach_limit $\endgroup$ – Gerald Edgar Jul 5 '17 at 12:57
  • $\begingroup$ Of possible historical interest are the 1800s items I cite in this 14 January 2009 sci.math post. $\endgroup$ – Dave L Renfro Jul 5 '17 at 14:36
  • $\begingroup$ @Dave L Renfro the links from your post do not work :( $\endgroup$ – Anixx Jul 5 '17 at 16:04
  • 1
    $\begingroup$ For the JFM reviews (whose links in my post don't work), here's the JFM look-up site itself. Enter author last name, part of title, etc. as needed. Alternatively, you can also obtain JFM reviews at the Zbl reviews look-up site (select "Structured Search"). $\endgroup$ – Dave L Renfro Jul 5 '17 at 16:35
  • 1
    $\begingroup$ @Dave L Renfro interestingly, they claim $\sin \infty=\cos \infty=0$ but $\sin^2\infty=\cos^2\infty=1/2$ And, indeed the value at infinity is the mean value (so the Cesaro method applies). $\endgroup$ – Anixx Jul 5 '17 at 17:34
4
$\begingroup$

In physics we use the Sokhotski–Plemelj theorem to evaluate integrals from zero to infinity of cosine and sine in the form:

$$F(k)=\int_0^\infty (\cos kx +i\sin kx)\,dx=\pi\delta(k)+i\,{\cal P}\frac{1}{k}$$

where $\delta(k)$ is the Dirac delta function and ${\cal P}$ is a reminder that when the right-hand-side is integrated over $k$ one should take the Cauchy principal value of the integral: $$\int_{-\infty}^\infty F(k)g(k)\,dk=\pi g(0)+i\,{\cal P}\,\int_{-\infty}^\infty \frac{g(k)}{k}\,dk.$$

Incidentally, from $F(1)=i$ I would indeed associate $\int_0^\infty \cos x\,dx$ and $\int_0^\infty \sin x\,dx$ with $0$ and $1$, respectively, as in the OP.

$\endgroup$
  • 1
    $\begingroup$ Still not clear, principal value of what integral?... $\endgroup$ – Anixx Jul 5 '17 at 12:03
  • $\begingroup$ If you get -1, you definitely made some mistake, I think. It is still unclear though how the first and second formulas are related (what is g(x) for instance). $\endgroup$ – Anixx Jul 5 '17 at 13:11
  • $\begingroup$ indeed, my mistake, corrected, thanks; $g(x)$ is an arbitrary function. $\endgroup$ – Carlo Beenakker Jul 5 '17 at 13:25
  • $\begingroup$ Purely logically the integral should be positive because sine function is initially positive. It is analogious to the series 1-1+1-1+... which sums up to 1/2, while if the first term is -1, the sum would be -1/2 $\endgroup$ – Anixx Jul 5 '17 at 15:36
  • 1
    $\begingroup$ the Wikipedia article has several applications of the Sokhotski–Plemelj integral relation, and a Google search gives many more pointers; a Fourier-transform formulation is here (appendix). $\endgroup$ – Carlo Beenakker Jul 5 '17 at 16:04
2
$\begingroup$

Many people developed a theory of limits of distributions at finite or infinite values in the 50's and 60's and for these all of your formulae hold. A particularly elementary version was developed by J. Sebastião e Silva and a complete description can be found online by googling has name and "limits of distributions at infinity". There is also a text book version by J. Campos Ferreira et al.---"Introduction to the theory of distributions".

$\endgroup$
  • $\begingroup$ I could not find anything using your search suggestion. $\endgroup$ – Anixx Jul 5 '17 at 17:19
  • $\begingroup$ The article I am referring to appeared in "Theory of Distributions", the proceedings of a summer institute published by the Gulbenkian Foundation, Lisbon, 1964. There is a site in Portugal which contains the complete works of Silva online. I will try to dig out the details for you. $\endgroup$ – cart Jul 5 '17 at 17:28
  • $\begingroup$ Unfortunately, I cannot read in Partugal... $\endgroup$ – Anixx Jul 5 '17 at 17:33
  • $\begingroup$ the article is not in portuguese. The site is "sebastiãoesilva100anos.org. $\endgroup$ – cart Jul 5 '17 at 17:34
  • $\begingroup$ then click on "publicações". There is a french and an english version, as I recall. Anyway, the text of Campos Ferreira is in english. $\endgroup$ – cart Jul 5 '17 at 17:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.