Re: Fast calculation of block-block covariance

Christoph,

package spatialCovariance (on CRAN) seems to do this, by reading
it's description. Please share with us whether it did what you wanted
or not; I acknowledge that this is a problem that other packages could
benefit from a fast solution to this problem.

Bests,
--
Edzer

Christoph Hofer wrote:
> Dear all
>
> I'm a Ph.D. student in Environmental Sciences at the ETH Zurich and   
> new to this list, but I have searched for an answer to my problem in  
> the list archives and couldn't find anything.
>
> I have the following problem:
> I want to calculate the block-block covariance C(B1,B2) among two  
> blocks  of arbitrary  shaped polygons (non-rectangular). In the gstat  
> package the block-block covariance is calculated by discretizing the  
> blocks by points (via the command  "nblockdiscr" one can choose the  
> number of points in the polygon).
> I wrote R code with the function integrate, but to calculate the  
> block-block covariance a quad integral of the covariance function has  
> to be calculated and  therefore even for polygons with a few vertices  
> my code is  very slow.
>
> Is there a fast solution (algorithm, package or function in R, aside  
> from the gstat solution) to calculate the block-block covariance of  
> arbitrary shaped polygons?
>
> Best regards
>
> Christoph Hofer
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo@xxxxxxxxxxxxxxxxx
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>