R stuff

Abajo he colocado algunas cosas simples de R que me han ayudado a resolver problemas más elaborados, junto a algunos links que me han aparecido interesantes. Por otro lado, tienes un Snippet Box de R por si quieres probar algunos códigos!

Snippet Box de R:

Efficient computations in R:

1. Matrix crossproducts: Use crossprod(A, B) instead of t(A) %*% B (or tcrossprod(A, B) if you want to compute A %*% t(B)).

2. Trace of the product of two matrices A and B: Use sum(A*t(B)) instead of sum(diag(A %*% B)). Another option is crossprod(c(A), c(t(B))).

3. Inverse of a symmetric matrix A: Use chol2inv(chol(A)) instead of solve(A).

4. Logarithm of the determinant of a symmetric matrix A: Use sum(log(eigen(A, symmetric = TRUE, only.values = TRUE)$values)) instead of log(diag(A)).

5. Reconstruction of a two-by-two partitioned matrix A:

   1. 1. 1. 1. 1. 1. 1. bblock2 <- function(A11, A12, A21, A22){

                     2.   block <- rbind(cbind(A11, A12), cbind(A21, A22))

                     3.   unname(block, force = FALSE)

                     4. }

6. Inverse of a two-by-two partitioned matrix A:  

   1. 1. 1. 1. 1. 1. 1. inv_bblock2 <- function(A11, A12, A21, A22){

                     2.   # Assumptions:

                     3.   # A11 and A22 are symmetric matrices

                     4.   # A21 = t(A12)

                     5.   M1 <- chol2inv(chol(A11))

                     6.   M2 <- crossprod(A12, M1)

                     7.   M3 <- tcrossprod(M2, A21)

                     8.   S22 <- chol2inv(chol(A22 - M3))

                     9.   S21 <- - S22 %*% M2

                     10.   S12 <- t(S21)

                     11.   S11 <- M1 - crossprod(M2, S21)

                     12.   S <- bblock2(S11, S21, S21, S22)

                     13.   output <- list(S11 = S11, S12 = S12, S21 = S21, S22 = S22, S = S)

                     14.   return(output)

                     15. }

Building R packages:

1. Including datasets