compute_post_fun_iwp {BayesGP} | R Documentation |
Computing the posterior samples of the function or its derivative using the posterior samples of the basis coefficients for iwp
Description
Computing the posterior samples of the function or its derivative using the posterior samples of the basis coefficients for iwp
Usage
compute_post_fun_iwp(
samps,
global_samps = NULL,
knots,
refined_x,
p,
degree = 0,
intercept_samps = NULL
)
Arguments
samps |
A matrix that consists of posterior samples for the O-spline basis coefficients. Each column represents a particular sample of coefficients, and each row is associated with one basis function. This can be extracted using 'sample_marginal' function from 'aghq' package. |
global_samps |
A matrix that consists of posterior samples for the global basis coefficients. If NULL, assume there will be no global polynomials and the boundary conditions are exactly zero. |
knots |
A vector of knots used to construct the O-spline basis, first knot should be viewed as "0", the reference starting location. These k knots will define (k-1) basis function in total. |
refined_x |
A vector of locations to evaluate the O-spline basis |
p |
An integer value indicates the order of smoothness |
degree |
The order of the derivative to take, if zero, implies to consider the function itself. |
intercept_samps |
A matrix that consists of posterior samples for the intercept parameter. If NULL, assume the function evaluated at zero is zero. |
Value
A data.frame that contains different samples of the function or its derivative, with the first column being the locations of evaluations x = refined_x.
Examples
knots <- c(0, 0.2, 0.4, 0.6)
samps <- matrix(rnorm(n = (3 * 10)), ncol = 10)
result <- compute_post_fun_iwp(samps = samps, knots = knots, refined_x = seq(0, 1, by = 0.1), p = 2)
plot(result[, 2] ~ result$x, type = "l", ylim = c(-0.3, 0.3))
for (i in 1:9) {
lines(result[, (i + 1)] ~ result$x, lty = "dashed", ylim = c(-0.1, 0.1))
}
global_samps <- matrix(rnorm(n = (2 * 10), sd = 0.1), ncol = 10)