MvInv: Invert jump heights function

In konkam/BNPdensity: Ferguson-Klass Type Algorithm for Posterior Normalized Random Measures

Invert jump heights function

Description

Determines the jump heights of an increasing additive process by inverting the M(v) function.

Usage

MvInv(eps, u = 0.5, alpha = 1, kappa = 1, gama = 1/2, N = 3001)

Arguments

eps	Dummy argument kept for consistency with past versions of the functions
u	Real number. The value of the latent variable at the current step.
alpha	Numeric constant. Total mass of the centering measure.
kappa	Numeric positive constant.
gama	Numeric constant. Discount parameter of the NRMI process.
N	Number of steps in the discretization scheme for the grid inversion. ## The function has been optimised but it is morally defined as: function(eps, u = 0.5, alpha = 1, kappa = 1, gama = 1 / 2, N = 3001) n <- length(w) v <- rep(NA, n) x <- -log(seq(from = exp(-1e-05), to = exp(-10), length = N)) f <- alpha / gamma(1 - gama) * x^(-(1 + gama)) * exp(-(u + kappa) * x) dx <- diff(x) h <- (f[-1] + f[-N]) / 2 Mv <- rep(0, N) for (i in seq(N - 1, 1)) Mv[i] <- Mv[i + 1] + dx[i] * h[i] for (j in seq(n)) v[j] <- x[which.min(Mv > w[j])] return(v)

Details

For internal use.

konkam/BNPdensity documentation built on March 14, 2024, 7:15 a.m.