Exponential integral and incomplete gamma function for R

The exponential integral `E_1(x) = int_x^Inf exp(-t)/t dt`

and the incomplete gamma function `G(a, x) = int_x^Inf t^(a-1) exp(-t) dt`

are closely related functions that arise in various fields of mathematics and statistics.

**expint** is a small R package that provides functions to compute the exponential integral and the incomplete gamma function.

Most conveniently for R package developers, the package also gives access to the underlying C workhorses through an API.

- R interface consisting of one main and four auxiliary functions to compute the exponential integral, and one function to compute the incomplete gamma function.
- Underlying C routines derived from the GNU Scientific Library mature and stable code.
- Included test package implementing the access to the C routine through the API. This test package uses the
`.External`

R to C interface and, as a bonus, shows how to vectorize an R function on the C side. - Exhaustive package vignette providing all the details about the formulas that are implemented in the package as well as a complete presentation of the API and its usage.

We tabulate the values of the exponential integral of order `n`

for `x = 1.275, 10, 12.3`

and `n = 1, 2, ..., 10`

as found in examples 4 through 6 of Abramowitz and Stegun (1972), section 5.3.

```
R> x <- c(1.275, 10, 12.3)
R > n <- 1:10
R> structure(t(outer(x, n, expint)),
+ dimnames = list(n, paste("x =", x)))
x = 1.275 x = 10 x = 12.3
1 0.14080993 4.156969e-06 3.439534e-07
2 0.09989831 3.830240e-06 3.211177e-07
3 0.07603031 3.548763e-06 3.009983e-07
4 0.06083077 3.304101e-06 2.831550e-07
5 0.05046793 3.089729e-06 2.672346e-07
6 0.04301687 2.900528e-06 2.529517e-07
7 0.03743074 2.732441e-06 2.400730e-07
8 0.03310097 2.582217e-06 2.284066e-07
9 0.02965340 2.447221e-06 2.177930e-07
10 0.02684699 2.325303e-06 2.080990e-07
```

We also tabulate the values of the incomplete gamma function for `a = -1.5, -1, -0.5, 1`

and `x = 1, 2, ..., 10`

.

```
R> a <- c(-1.5, -1, -0.5, 1)
R> x <- 1:10
R> structure(t(outer(a, x, gammainc)),
+ dimnames = list(x, paste("a =", a)))
a=-1.5 a=-1 a=-0.5 a=1
1 1.264878e-01 1.484955e-01 1.781477e-01 3.678794e-01
2 1.183299e-02 1.876713e-02 3.009876e-02 1.353353e-01
3 1.870260e-03 3.547308e-03 6.776136e-03 4.978707e-02
4 3.706365e-04 7.995573e-04 1.733500e-03 1.831564e-02
5 8.350921e-05 1.992938e-04 4.773965e-04 6.737947e-03
6 2.045031e-05 5.304291e-05 1.379823e-04 2.478752e-03
7 5.310564e-06 1.478712e-05 4.127115e-05 9.118820e-04
8 1.440569e-06 4.267206e-06 1.266464e-05 3.354626e-04
9 4.042025e-07 1.264846e-06 3.964430e-06 1.234098e-04
10 1.165117e-07 3.830240e-07 1.260904e-06 4.539993e-05
```

Unless important bugs are discovered either upstream in the GSL functions or in our adaptation, the code base of the package is not expected to change much in the future. In other words: stable.

The package should be installed from CRAN using

`install.packages("expint")`

Vincent Goulet is the author and maintainer of the package. Many other copyright holders (for the original GSL code and for parts of the base R code) are credited in the `DESCRIPTION`

file.

GPL (>= 2)