As far as I can tell, R doesn’t have a function for building block diagonal matrices so as I needed one, I’ve coded it myself. It might save someone some time.

Example:

Let `m1` and `m2` two square matrices.

By passing any number of matrices as argument:

Or a list of matrices:

It produces:

Select All Code:

1 2 | colourScale<-seq(0,1,length.out=100) image(blockMatrix[30:1,],asp=1,col=rgb(colourScale,colourScale,colourScale),ann=FALSE,xaxt="n",yaxt="n",axes=FALSE) |

That is, a matrix which is zero everywhere but with `m1`,`m2`,`m2` and `m1` on the diagonal.

Code:

Select All Code:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 | # builds a block matrix whose diagonals are the square matrices provided. # m1=matrix(runif(10*10),nrow=10,ncol=10) # m2=matrix(runif(5*5),nrow=5,ncol=5) # blockMatrix<-blockMatrixDiagonal(m1,m2,m2,m1) # or # blockMatrix<-blockMatrixDiagonal(list(m1,m2,m2,m1)) # C.Ladroue blockMatrixDiagonal<-function(...){ matrixList<-list(...) if(is.list(matrixList[[1]])) matrixList<-matrixList[[1]] dimensions<-sapply(matrixList,FUN=function(x) dim(x)[1]) finalDimension<-sum(dimensions) finalMatrix<-matrix(0,nrow=finalDimension,ncol=finalDimension) index<-1 for(k in 1:length(dimensions)){ finalMatrix[index:(index+dimensions[k]-1),index:(index+dimensions[k]-1)]<-matrixList[[k]] index<-index+dimensions[k] } finalMatrix } |

Hmmm. Did you try library(sos); findFn(“block diagonal”) ?

library(Matrix)

m1 <- matrix(runif(4),ncol=2)

m2 <- matrix(1:4,ncol=2)

as.matrix(bdiag(m1,m2))

No I didn’t, although I suspected it had been done before (I started using R 2 months ago, I’m still climbing the learning curve).

I didn’t know this ‘sos’ package, I’ve just installed it now. Thanks, it’ll probably save me time in the future.