f_crps_wgm: Compute CRPS score for a Gaussian mixture

In robertsy/assimilr: Local Ensemble Kalman Particle Filters

Description

unequal weights case

Usage

f_crps_wgm(mu, sig2, obs, w)

Arguments

mu	mixture means
sig2	mixture variance
obs	observations
w	mixture component probability

Value

the crps score

Examples

set.seed(1)
truth <- 0
n <- 20
ens.means <- rnorm(n, truth, 1)
w <- dnorm( ens.means - truth)
w <- w/sum(w)
f_crps_wgm(mu=ens.means, sig2=.1, obs=truth, w=w)
sigvec <- rgamma(n, 1)
f_crps_wgm(mu=ens.means, sig2=sigvec, obs=truth, w=w)
f_crps_wgm(  mu=ens.means, sig2=.1, obs=truth, w=rep(1/n, n))

robertsy/assimilr documentation built on May 27, 2019, 10:33 a.m.