R help - Jan 2011 - Integration of two lines

Hello,

I need to integrate the absolute difference between two lines measured
on different points.

# For example :
x <- seq(0, 1, 1/100)
f_x <- runif(101) + x
y <- seq(0, 1, 1/23)
f_y <- runif(24) + (1 - y)

plot(x, f_x, type="l")
lines(y, f_y)

Then I would like to compute Integral( | f_x - f_y | )dx.
(This is not the same as | Integral(f_x)dx -  Integral(f_y)dx |.)

Computing this integral looks non trivial. I guess I should interpolate
the points of f_y over x and integrate both lines on these intervals.
Even then I would miss points where the lines cross.

There are functions to integrate below *one* line (I'm thinking about
the trapz function in caTools).
Do you know if there is a function to do this integration properly with
two lines (and especially their absolute difference)?

Regards,
Xavier

g <- function(x) abs(f1(x)-f2(x))

now you have one function and you can integrate it.

Rich

Sent from my iPhone

On Jan 25, 2011, at 7:32, Xavier Robin <Xavier.Robin at unige.ch> wrote:
> Hello,
> 
> I need to integrate the absolute difference between two lines measured
> on different points.
> 
> # For example :
> x <- seq(0, 1, 1/100)
> f_x <- runif(101) + x
> y <- seq(0, 1, 1/23)
> f_y <- runif(24) + (1 - y)
> 
> plot(x, f_x, type="l")
> lines(y, f_y)
> 
> Then I would like to compute Integral( | f_x - f_y | )dx.
> (This is not the same as | Integral(f_x)dx -  Integral(f_y)dx |.)
> 
> Computing this integral looks non trivial. I guess I should interpolate
> the points of f_y over x and integrate both lines on these intervals.
> Even then I would miss points where the lines cross.
> 
> There are functions to integrate below *one* line (I'm thinking about
> the trapz function in caTools).
> Do you know if there is a function to do this integration properly with
> two lines (and especially their absolute difference)?
> 
> Regards,
> Xavier
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

Xavier Robin <Xavier.Robin <at> unige.ch> writes:
> Hello,
> 
> I need to integrate the absolute difference between two lines measured
> on different points.
> 
> # For example :
> x <- seq(0, 1, 1/100)
> f_x <- runif(101) + x
> y <- seq(0, 1, 1/23)
> f_y <- runif(24) + (1 - y)
> 
> plot(x, f_x, type="l")
> lines(y, f_y)
> 
> Then I would like to compute Integral( | f_x - f_y | )dx.
> (This is not the same as | Integral(f_x)dx -  Integral(f_y)dx |.)
First define a function from those points:

    fx <- approxfun(x, f_x)
    fy <- approxfun(y, f_y)
    f  <- function(x) abs(fx(x)-fy(x))

and now you can apply integrate() or trapz():

    xx <- sort(c(x, y))
    yy <- f(xx)
    trapz(xx, yy)

trapz() should return the more accurate (i.e. exact) result.

--Hans Werner
> Computing this integral looks non trivial. I guess I should interpolate
> the points of f_y over x and integrate both lines on these intervals.
> Even then I would miss points where the lines cross.
> 
> There are functions to integrate below *one* line (I'm thinking about
> the trapz function in caTools).
> Do you know if there is a function to do this integration properly with
> two lines (and especially their absolute difference)?
> 
> Regards,
> Xavier
> 
>

Le 25.01.2011 15:23, Rmh a ?crit :
> g <- function(x) abs(f1(x)-f2(x))
> 
> now you have one function and you can integrate it.
Thank you Rich.

Unfortunately I have no f1 and f2 functions, only a set of observed
points on two lines - and no idea about the underlying distribution to
create a function.

Other ideas?

Xavier