r语言 - parLapply - 如何解决错误"Could not find function "绑定ToEnv " "？

Question

我想使用parLapply，我正在设置我的代码，就像这里介绍的那样：http://www.win-vector.com/blog/2016/01/parallel-computing-in-r/

最近几次效果很好。但是，使用我当前的parLapply呼叫，我收到错误Error in checkForRemoteErrors(val) : 3 nodes produced errors; first error: could not find function "bindToEnv".

这里有一个简短的例子：

#' Copy arguments into env and re-bind any function's lexical scope to bindTargetEnv .
#' 
#' See http://winvector.github.io/Parallel/PExample.html for example use.
#' 
#' 
#' Used to send data along with a function in situations such as parallel execution 
#' (when the global environment would not be available).  Typically called within 
#' a function that constructs the worker function to pass to the parallel processes
#' (so we have a nice lexical closure to work with).
#' 
#' @param bindTargetEnv environment to bind to
#' @param objNames additional names to lookup in parent environment and bind
#' @param names of functions to NOT rebind the lexical environments of
bindToEnv <- function(bindTargetEnv=parent.frame(),objNames,doNotRebind=c()) {
# Bind the values into environment
# and switch any functions to this environment!
for(var in objNames) {
val <- get(var,envir=parent.frame())
if(is.function(val) && (!(var %in% doNotRebind))) {
# replace function's lexical environment with our target (DANGEROUS)
environment(val) <- bindTargetEnv
}
# assign object to target environment, only after any possible alteration
assign(var,val,envir=bindTargetEnv)
}
}
ccc <- 1
# Parallel
cl <- parallel::makeCluster(getOption("cl.cores", 3))
junk <- parallel::clusterEvalQ(cl, c(library(data.table)))
f <- function(x) {
bindToEnv(objNames = 'ccc')
return(x+x)  
}
b <- do.call(rbind, parallel::parLapply(cl, 1:10,  f))

如果我不添加bindToEnv一切正常。我做错了什么？谢谢！

Answer 1

在创建集群之前，您需要使用clusterExport()导出您定义的已用函数和对象。

library(parallel)
cl <- makeCluster(getOption("cl.cores", 3))
clusterEvalQ(cl, c(library(data.table)))
clusterExport(cl, c("bindToEnv", "ccc"), 
envir=environment())
f <- function(x) {
bindToEnv(objNames='ccc')
return(x+x)  
}
b <- do.call(rbind, parallel::parLapply(cl, 1:10,  f))
b
#        ,1]
#  [1,]    2
#  [2,]    4
#  [3,]    6
#  [4,]    8
#  [5,]   10
#  [6,]   12
#  [7,]   14
#  [8,]   16
#  [9,]   18
# [10,]   20
stopCluster(cl)