resample(smcUtils)
resample()所属R语言包：smcUtils
Resample
重新取样
译者：生物统计家园网 机器人LoveR
描述----------Description----------
A set of resampling functions with unbiased number of replicates.
重采样功能不带偏见的一些重复的一组。
用法----------Usage----------
resample(weights, num.samples=length(weights),
method = c("multinomial","residual","stratified","systematic","branching"),
nonuniformity = c("none","ess","cov","entropy"), threshold = NULL,
resample.function = multinomial.resample)
参数----------Arguments----------
参数：weights
a vector of normalized weights
一个归一化的权重向量
参数：num.samples
a scalar indicating the number of samples to return (for "branching.resample", "num.samples" is the expected number of samples as the actual number is random)
表示的样本数返回（branching.resample，num.samples是一个标量，预期数目的样本的实际数目是随机的）
参数：method
a character string indicating the resampling method to use. One of ""multinomial"" (default), ""residual"", ""stratified"", ""systematic"", or ""branching"", can be abbreviated.
重采样方法使用一个字符串，指示。之一的“多元”（默认）“，”残“”，“分层”，“系统”，或“分支”，可以缩写。
参数：nonuniformity
a character string indicating which hueristic to use for measuring weight nonuniformity. One of ""none"" (default), ""ess"", ""cov"", or ""entropy"", can be abbreviated. If "none", resampling will always be performed.
一个字符串，表示这hueristic用于测量重量不均匀。的一个“没有”（默认值），“”ESS的“，”覆盖“，或”“熵”，可以缩写。如果“无”，将始终进行重采样。
参数：threshold
a scalar indicating when to resample. If "nonuniformity="none"", resampling is always performed regardless of threshold. If "nonuniformity="ess"" or "entropy", resampling is performed when "nonuniformity<threshold". If "nonuniformity="cov"", resampling is performed when 'nonuniformity> threshold'. The default threshold is "0.5*num.samples" for "ess" and "cov", and "0.5*log2(num.samples)" for "entropy".
一个标量时重新取样。如果不均匀=“无”，重采样总是不管阈值。如果不均匀=“ESS的”或“熵”，重采样时，进行不均<阈值。如果不均匀=“覆盖”，重采样时的不均匀性“>”阈值“。预设的阈值是0 .5 * num.samples“，”ESS“和”覆盖“，0 .5 *为log2（num.samples）”，“熵”。
参数：resample.function
the resampling function to use on the remainder (only used when method = "residual")
重采样功能使用上的余数（仅使用方法=“剩余”）
值----------Value----------
<table summary="R valueblock"> <tr valign="top"><td>weights</td> <td> the component weights</td></tr> <tr valign="top"><td>indices</td> <td> an integer vector containing the indices of resampled components. If no resampling was performed, then "indices=1:length(weights)".</td></tr> </table>
<table summary="R valueblock"> <tr valign="top"> <TD> weights</ TD> <TD>成分的权重</ TD> </ TR> <TR VALIGN =“顶” > <TD> indices</ TD> <TD>指数重采样组件包含一个整数向量。如果不进行重采样，然后指数= 1：长度（权重）。</ TD> </ TR> </ TABLE>
（作者）----------Author(s)----------
Jarad Niemi
参见----------See Also----------
resampling
resampling
实例----------Examples----------
ws = renormalize.weights(runif(10))
resample(ws)
## A more involved example, comparing particle filtering with the Kalman filter[＃A更复杂的例子，比较粒子滤波与卡尔曼滤波]
# Generate data from a local level linear model[从地方一级线性模型生成数据]
N = 1000; W = 1^2; V = 1; m0 = 0; C0 = 1
true.x = rep(NA,N); true.x[1] = rnorm(1,m0,sqrt(C0))
for (i in 2:N) true.x[i] = rnorm(1,true.x[i-1],sqrt(W)) # Evolve x[进化x]
y = rnorm(N,true.x,sqrt(V)) # Noisy data[噪声数据]
# Run a particle filter[运行颗粒过滤器]
J = 1e2
x = matrix(NA,N,J); x[1,] = rnorm(J,m0,C0) # Sample from the prior for x[从现有为x的样品]
ws = matrix(NA,N,J); ws[1,] = renormalize.weights(dnorm(y[1],x[1,],sqrt(V),log=TRUE))
# Run a Kalman filter[运行一个卡尔曼滤波器]
m = rep(NA,N); m[1] = m0 # Kalman filter expectation[卡尔曼滤波器的期望]
M = rep(NA,N); M[1] = C0 # Kalman filter variance[卡尔曼滤波方差]
for (i in 2:N) {
# Particle filter[粒子过滤器]
component = resample(ws[i-1,],J,"stratified","ess",0.8*J)
x[i,] = rnorm(J,x[i-1,component$indices],sqrt(W))
log.weights = log(component$weights)+dnorm(y[i],x[i,],sqrt(V),log=TRUE)
ws[i,] = renormalize.weights(log.weights,log=TRUE)
# Kalman filter[卡尔曼滤波器]
K = (M[i-1]+W)/(M[i-1]+W+V) # Adaptive coefficient[自适应系数]
m[i] = K*y[i]+(1-K)*m[i-1]
M[i] = K*M[i-1]
}
pf.m = apply(x*ws,1,sum)
plot(m,type='l',ylim=range(pf.m,m),xlab='t',ylab='x')
lines(pf.m,col='red')
legend("bottomleft",inset=0.01,c("Kalman filter mean","Particle filter mean"),
col=c("black","red"),lty=rep(1,2),bg="white")
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。
|
发表于 2012-9-30 10:19:10
