R/dmats.R

In quint: Qualitative Interaction Trees

makedmat<-function(nnod){
  ##creates designmatrix D with all possible assignments of the terminal nodes to the three  partition classes
  ##nnod = I = number of terminal nodes after a split
  ##dmat=K * I matrix: K=number of possible assignments;
  rmat<-3^(nnod)
  #rmat is total number of rows
  dmat<-matrix(unlist(lapply(1:nnod,function(jj,rmat){as.double(gl(3,rmat/(3^jj),rmat))},rmat=rmat)),ncol=nnod,nrow=rmat)
  return(dmat)}

makedmats<-function(dmat){
  #check of boundary condition: partition class cardinality condition:P1 and P2 may not be empty
  #creates D'(dmats): matrix D with admissible assignments(K'); dmats= K' * I matrix;
  sel1<-numeric(dim(dmat)[1])
  sel2<-numeric(dim(dmat)[1])
  #count the assignments to p1 for each row of dmat
  sel1<-apply(dmat==1,1,sum)
  #count the assignments to p2 for each row of dmat
  sel2<-apply(dmat==2,1,sum)
  #select the rows for which sel1 & sel2 not equals 0
  dmats<-dmat[sel1&sel2!=0,]
  return(dmats)}