Hello, I couldn't find information on whether the logarithmic integrals Li_m(x) = integral_0^x log(t)^(-m) dt for x >= 0 are available in R? Best wishes Oliver
logarithmic integrals in R?
4 messages · Hans W Borchers, Oliver Kullmann
9 days later
Oliver Kullmann <O.Kullmann <at> swansea.ac.uk> writes:
Hello, I couldn't find information on whether the logarithmic integrals Li_m(x) = integral_0^x log(t)^(-m) dt for x >= 0 are available in R?
I saw your request only this weekend.
The first logarithmic integral can be computed using the exponential
integral Ei(x) per
li(x) = Ei(log(x))
and elliptic integrals are part of the 'gsl' package, so
library('gsl')
x <- seq(2, 10, by=0.5)
y <- expint_Ei(log(x))
y
See e.g. the Handbook of Mathematical Functions for how to reduce higher
logarithmic integrals.
Another possibility is to use the Web API of 'keisan', the calculation
library of Casio.
Regards
Hans Werner
Best wishes Oliver
Thanks for the information.
On Sat, May 29, 2010 at 01:15:29PM +0000, Hans W. Borchers wrote:
Oliver Kullmann <O.Kullmann <at> swansea.ac.uk> writes:
Hello, I couldn't find information on whether the logarithmic integrals Li_m(x) = integral_0^x log(t)^(-m) dt for x >= 0 are available in R?
I saw your request only this weekend.
The first logarithmic integral can be computed using the exponential
integral Ei(x) per
li(x) = Ei(log(x))
I found gsl at http://cran.r-project.org/web/packages/gsl/index.html.
and elliptic integrals are part of the 'gsl' package, so
library('gsl')
x <- seq(2, 10, by=0.5)
y <- expint_Ei(log(x))
y
See e.g. the Handbook of Mathematical Functions for how to reduce higher
logarithmic integrals.
However here I wasn't succesful: Going through the chapter http://www.math.ucla.edu/~cbm/aands/page_228.htm I didn't find any mentioning of the higher logarithmic integrals.
Another possibility is to use the Web API of 'keisan', the calculation library of Casio.
Interesting; but again only li(x). Also a google search on "higher logarithmic integrals", "logarithmic integrals" or "li_n(x)" doesn't reveal anything, so I would be thankful for a hint. Thanks again! Oliver
Oliver Kullmann <O.Kullmann <at> swansea.ac.uk> writes:
Thanks for the information.
What I meant were formulas like
\int 1/\log(t)^2 dt = -t/\log(t) + li(t)
\int 1/\log(t)^3 dt = 1/2 * ( -t/\log(t)^2 - t/\log(t) + li(t) )
and higher forms that can be expressed through the Gamma function.
I am certain I 've seen them in AandS' handbook (where else?), but
sure cannot remember in which chapter or page.
Which logarithmic integrals do you really need, and on what range?
Regards,
Hans Werner
On Sat, May 29, 2010 at 01:15:29PM +0000, Hans W. Borchers wrote:
Oliver Kullmann <O.Kullmann <at> swansea.ac.uk> writes:
Hello, I couldn't find information on whether the logarithmic integrals Li_m(x) = integral_0^x log(t)^(-m) dt for x >= 0 are available in R?
[...]
I found gsl at http://cran.r-project.org/web/packages/gsl/index.html.
and elliptic integrals are part of the 'gsl' package, so
library('gsl')
x <- seq(2, 10, by=0.5)
y <- expint_Ei(log(x))
y
See e.g. the Handbook of Mathematical Functions for how to reduce higher
logarithmic integrals.
However here I wasn't succesful: Going through the chapter http://www.math.ucla.edu/~cbm/aands/page_228.htm I didn't find any mentioning of the higher logarithmic integrals.
[...]
Also a google search on "higher logarithmic integrals", "logarithmic integrals" or "li_n(x)" doesn't reveal anything, so I would be thankful for a hint. Thanks again! Oliver