[R-sig-Geo] Reproducible results with overlay and clusterR

[Hugo Costa]

Thank you very much.
I'll implement this with my real data.
Hugo

Bede-Fazekas Ákos <bfalevlist using gmail.com> escreveu no dia terça, 28/05/2019
à(s) 06:15:

> Dear Hugo,
> please find below a solution. It might be too complicated and shorter
> solutions might exist, but it works.
>
> library(raster)
> r1 <- r2 <- raster()
> r1[]<-runif(ncell(r1), 1, 100)
> r2[]<-1
> r2[1:30000]<-2
> r2[30001:35000]<-NA
> plot(stack(r1,r2))
>
>
> ## Function
> prob<-data.frame(value=1:100, r1=log(1:100), r2=max(log(1:100))-log(1:100))
> f <- function(x,y,seed,...){
>    ifelse(is.na(y),NA,rbinom(n=1, size=1, prob=prob[[y]][x]))
> }
> f <- Vectorize(f)
>
> ## Wrapper function
> wrapper <- function(raster,seed,...){
>      set.seed(seed)
>      overlay(raster,fun=f)
> }
>
> ## Overlay (reproducible)
> result1<-wrapper(stack(r1,r2),seed=0)
> result2<-wrapper(stack(r1,r2),seed=0)
> compareRaster(result1,result2,values=TRUE)
>
> ## Cluster overlay (Not reproducible)
> beginCluster(2)
> result3<-clusterR(stack(r1,r2),wrapper,args=list(seed=0),
> export=c("prob", "f"))
> endCluster()
>
> beginCluster(2)
> result4<-clusterR(stack(r1,r2),wrapper,args=list(seed=0),
> export=c("prob", "f"))
> endCluster()
>
> compareRaster(result3,result4,values=TRUE)
>
> HTH,
> Ákos Bede-Fazekas
> Hungarian Academy of Sciences
>
> 2019.05.27. 22:59 keltezéssel, Hugo Costa írta:
> > Dear all
> > I need to implement a raster::overlay operation with clusterR (which
> > works), but don't figure out how to do it reproducible. How to set seeds
> in
> > the cluster? Please find a example below.
> > Thank you
> > Hugo
> >
> > ## Rasters
> > library(raster)
> > r1 <- r2 <- raster()
> > r1[]<-runif(ncell(r1), 1, 100)
> > r2[]<-1
> > r2[1:30000]<-2
> > r2[30001:35000]<-NA
> > plot(stack(r1,r2))
> >
> >
> > ## Function
> > prob<-data.frame(value=1:100, r1=log(1:100),
> r2=max(log(1:100))-log(1:100))
> > f <- function(x,y,...){
> >    ifelse(is.na(y),NA,rbinom(n=1, size=1, prob=prob[[y]][x]))
> > }
> > f <- Vectorize(f)
> >
> > ## Overlay (reproducible)
> > set.seed(0)
> > result1<-overlay(stack(r1,r2), fun=f)
> > set.seed(0)
> > result2<-overlay(stack(r1,r2), fun=f)
> > compareRaster(result1,result2,values=TRUE)
> >
> > ## Cluster overlay (Not reproducible)
> > set.seed((0))
> > beginCluster(2)
> > result3<-clusterR(stack(r1,r2),overlay,args=list(fun=f), export="prob")
> > endCluster()
> >
> > set.seed((0))
> > beginCluster(2)
> > result4<-clusterR(stack(r1,r2),overlay,args=list(fun=f), export="prob")
> > endCluster()
> >
> > compareRaster(result3,result4,values=TRUE)
> >
> >       [[alternative HTML version deleted]]
> >
> > _______________________________________________
> > R-sig-Geo mailing list
> > R-sig-Geo using r-project.org
> > https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> >
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

	[[alternative HTML version deleted]]