symIndices: Vector of Indices Useful for Symmetric or Anti-Symmetric...
In kergp: Gaussian Process Laboratory

Description Usage Arguments Details Value Examples

Vector of indices useful for symmetric or anti-symmetric matrices

1	symIndices(n, diag = FALSE)

`n`	Size of a square matrix.
`diag`	Logical. When `FALSE` the diagonal is omitted in the lower and upper triangles.

This function is intended to provide computations which are faster than lower.tri and upper.tri.

A list containing the following integer vectors, each with length (n - 1) n / 2.

`i, j`	Row and column indices for the lower triangle to be used in a two-index style.
`kL`	Indices for the lower triangle, to be used in single-index style. The elements are picked in column order. So if `X` is a square matrix with size `n`, then `X[kL]` is the vector containing the elements of the lower triangle of `X` taken in column order.
`kU`	Indices for the upper triangle, to be used in a single-index style. The elements are picked in row order. So if `X` is a square matrix with size `n`, then `X[kU]` is the vector containing the elements of the upper triangle of `X` taken in row order.

n <- rpois(1, lambda = 10)
L <- symIndices(n)
X <- matrix(1L:(n * n), nrow = n)
max(abs(X[lower.tri(X, diag = FALSE)] - L$kL))
max(abs(t(X)[lower.tri(X, diag = FALSE)] - L$kU))
cbind(row = L$i, col = L$j)