bayesSurv (version 2.4)

vecr2matr: Transform single component indeces to double component indeces

Description

Components of a bivariate G-spline can be indexed in several ways. Suppose that the knots in the first dimension are $\mu_{1,-K_1},\dots,\mu_{1,K_1}$ and the knots in the second dimension $\mu_{2,-K_2},\dots,\mu_{2,K_2}.$ I.e. we have $2K_1+1$ knots in the first dimension and $2K_2+1$ knots in the second dimension. Each G-spline component can have a double index $(k_1,k_2)$ assigned which means that it corresponds to the knot $(\mu_{1,k_1},\mu_{2,k_2})$ or alternatively the same G-spline component can have a~single index $$r=(k_2 + K_2)\times(2K_1+1) + k_1 + K_1 + 1$$ assigned where $r$ takes values from $1,\dots,K_1\times K_2$. Single indexing is used for example by files r.sim and r_2.sim generated by functions bayesHistogram, bayesBisurvreg, bayessurvreg2 to save some space.

This function serves to translate single indeces to double indeces using the relationship $$k_1 = (r - 1) \mbox{ mod } (2K_1+1) - K_1$$ $$k_2 = (r - 1) \mbox{ div } (2K_1+1) - K_2$$

The function can be used also in one dimensional case when a~simple relationship holds $$r = k + K + 1$$ $$k = r - 1 - K$$

Usage

vecr2matr(vec.r, KK)

Arguments

vec.r

a~vector of single indeces

KK

a~vector with numbers of knots on each side of the central knot for each dimension of the G-spline. The length of KK determines dimension of the G-spline

Value

• In bivariate case: a~matrix with two columns and as many rows as the length of vec.r.

  In univariate case: a~vector with as ,amy components as the length of vec.r.

Examples

### Bivariate G-spline
### with 31 knots in each dimension
KK <- c(15, 15)

### First observation in component (-15, -15),
### second observation in component (15, 15),
### third observation in component (0, 0)
vec.r <- c(1, 961, 481)
vecr2matr(vec.r, KK)