I'm implementing a function to compute the moore-penrose inverse, using a code
from the article: Fast Computation of Moore-Penrose Inverse Matrices. Neural
Information Processing - Letters and Reviews. Vol.8, No.2, August 2005
However, the R presents an error message when I use the geninv.
The odd thing is that the error occurs for some arrays, however they have the
same size. And the R indicates the lack of compatibility between the matrix!
Below is an example:
Creating the function geninv
geninv <- function(x)
{
m <- dim(x)[1]
n <- dim(x)[2]
tr <- 0
if(m < n) {
a <- tcrossprod(x)
n <- m
tr <- 1
}
else a <- crossprod(x)
dA <- diag(a)
tol=min(dA[dA>0])*1e-9
L = a*0
r = 0
for(k in 1:n){
r = r+1
L[k:n,r] = a[k:n,k]-(L[k:n,1:(r-1)]%*%(t(L[k,1:(r-1)])))
if(L[k,r] > tol){
L[k,r] <- sqrt(L[k,r])
if (k < n) L[(k+1):n,r] <- L[(k+1):n,r]/L[k,r]
}
else r <- r-1
}
L <- L[,1:r]
M <- solve(crossprod(L))
if (tr == 1) Y <- t(x)%*%L%*%M%*%M%*%t(L)
else Y <- L%*%M%*%M%*%t(L)%*%t(x)
return(Y)
}
# Perfect result! This result is identical of the function ginv!
library(MASS)
mp <- 10
np <- 5
a <- matrix(c(1:mp*np),mp,np)
dim(a) # 10,5
geninv(a)
ginv(a)
# Problem
a <- matrix(c(1:50),mp,np) # The difference is the vector (1:50)
dim(a) # 10,5
geninv(a)
Error in L[k:n, 1:(r - 1)] %*% (t(L[k, 1:(r - 1)])) :
arguments are not compatible
The problem this in matrix definition?
Thanks very much!
Fábio Mathias Corrêa
Estatística e Experimentação Agropecuária/UFLA
____________________________________________________________________________________
Veja quais são os assuntos do momento no Yahoo! +Buscados
http://br.maisbuscados.yahoo.com
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
