CODEWITHSUNDEEP
rvectorization
dyluns
dyluns

Reputation: 155

R: how to create a vectorized function where the upper of an integral is the variable

I'm trying to apply ks.test test to a group of data and a custom CDF. The CDF has the following form:

my_cdf <- function(x) integrate(my_pdf, 0, x, subdivisions=2000)$value

where my_pdf is a function that takes a single value as the argument.

Now, in order to use ks.test, I need my_cdf to produce a vectorized output given a vector input. How should I do that?

Example my_pdf (It is not a real PDF but that should not matter):

my_pdf <- function(y) y + 1

Example objective:

my_cdf(c(1,2,3))

[1] 1.5 4 7.5

Upvotes: 0

Views: 56

Answers (2)

G. Grothendieck
G. Grothendieck

Reputation: 269870

Use Vectorize to transform my_cdf into a vectorized version:

my_cdfv <- Vectorize(my_cdf)
my_cdfv(1:3)
## [1] 1.5 4.0 7.5

Upvotes: 3

Allan Cameron
Allan Cameron

Reputation: 174293

You can use sapply. Personally, I would allow any pdf to be supplied to the function

my_cdf <- function(x, pdf)
{
  sapply(x, function(y) integrate(pdf, 0, y, subdivisions=2000)$value)
}

So we can do:

my_pdf <- function(y) y + 1

my_cdf(c(1, 2, 3), pdf = my_pdf)
#> [1] 1.5 4.0 7.5

or

my_cdf(c(0.1, 0.5, 1), pdf = dnorm)
#> [1] 0.03982784 0.19146246 0.34134475

If you're sure you'll never need to supply a different pdf, then you can just do

my_cdf <- function(x) sapply(x, function(y) integrate(my_pdf, 0, y, subdivisions=2000)$value)

Upvotes: 1

Related Questions