R Language Parallel Computing Snow package document (Beta)

1.snow-clustercluster-level on a snow cluster

Snow-clustercluster-level on a snow cluster
Clustersplit (cl, seq)
Clustercall (cl, fun, ...)
Clusterapply (cl, x, Fun, ...)
CLUSTERAPPLYLB (cl, x, Fun, ...)
Clusterevalq (cl, expr)
Clusterexport (CL, list, EnviR =. GLOBALENV)
Clustermap (cl, fun, ..., Moreargs = NULL, RECYCLE = TRUE)

Arguments	Description
Cl	Cluster objects
Fun	Function
Expr	The expression to be executed for each node
Seq	Vectors to be split into each node
List	List of global variables passed to each node
EnviR	Variable mode (default global)
X	Matrix
...	Additional parameters to pass to the fun
Moreargs	Additional parameters of fun
RECYCLE	logical value; If true, short parameters are recycled

Detailed description
These are the basic functions for cluster computing. All programs running on child nodes run under the Trycatch function. That is, if any child nodes have errors, they will be marked on the master node. Much more detail will be considered later.

Clustercall (cl, fun, ...)
Use the same function fun for each node to output the results of each node as a list.

Clusterevalq (cl, expr)
Using fun for each node, unlike Clustercall, there are no additional parameter inputs.

Clusterapply (cl, x, Fun, ...)
Each parameter of the matrix X is assigned to each node separately, and the x parameter is calculated by looping over the number of nodes.

CLUSTERAPPLYLB (cl, x, Fun, ...)
Balanced load mode for clusterapply. If the x parameter exceeds the number of nodes, the remaining parameters are assigned to the node that has finished executing the task. Using this function makes better use of the cluster, but this approach reduces computational performance.

Clustermap (cl, fun, ..., Moreargs = NULL, RECYCLE = TRUE)
Similar to mapply, examples

Clustermap (cl,function (x, y) Seq_len (x) + y,c (a =  1, b = 2, c = 3), C (A = ten, B = 0, c = -10))
[[1]]
[1] 11
  
   [[2]
[1] 1 2

[[3]]
[1]-9-8-7
  

1 2 3 4 5 6 7 8 9

Clusterexport (CL, list, EnviR =. GLOBALENV)
Pass the list parameter as a global variable to each node.

Clustersplit (cl, seq)
The SEQ sequence is assigned to each node, and the current version requires the sequence length to be the same as the number of nodes.

> Clustersplit (cl,c)
[[1]]
[1] 1

[[2]]
[1] 2

1 2 3 4 5 6

Comprehensive Examples:

CL <-Makesockcluster (C ("localhost", "localhost")) #建立集群
clusterapply (cl, 1:2, get ("+"), 3) #第一个节点算1 +3 The second node counts 2+3
Clusterevalq (CL, library (boot)) #每个节点加载lib
x<-1
clusterexport (cl, "x") #每个节点加载x作为环境变量
Clustercall (cl, function (y) x + y, 2) #每个节点计算一次1 +3

1 2 3 4 5 6 2.snow-parallel ———-Higher level snow functions

snow-parallel ———-higher level snow functions
parlapply (cl, x, Fun, ...)
parsapply (cl, X, fun, ..., simplify = TRUE, use.) NAMES = TRUE)
parapply (cl, X, MARGIN, fun, ...)
parrapply (cl, x, Fun, ...)
parcapply (cl, x, Fun, ...)
parmm (cl, A, B)

Arguments	Description
Cl	Cluster objects
Fun,fun	Function
X,x,a,b	Matrix
...	Additional parameters to pass to the fun
Moreargs	Additional parameters of fun
MARGIN	Specify the vector dimension to use
Simplify	logical value; see sapply
Use. NAMES	logical value; see sapply

Detailed description
1.parlapply,parsapply and parapply are parallel computational versions of lapply,sapply,apply.
2.parRapply and parcapply are the corresponding versions of the row and column calculations for apply, and are more efficient than parapply.
3.parmm:is a very simple (minded) parallel matrix multiply; It's intended as an illustration (not yet functional)
4. For more details see: Simple Network of workstations for R
Example

# # NOT run:
cl <-makesockcluster (C ("localhost", "localhost"))
parsapply (cl, 1:20, get ("+"), 3)
# # END ( Not run)

1 2 3 4

[1]  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

1

Snow-rand generating random numbers in a cluster
CLUSTERSETUPRNG (cl, type = "Rngstream", ...)
Clustersetuprngstream (cl, Seed=rep (12345,6), ...)
CLUSTERSETUPSPRNG (cl, seed = round (2^32 * runif (1)), Prngkind = "Default", para = 0, ...)

Describe
Initializes the random number in the cluster, using the L ' Ecuyer or SPRNG package to generate a random number (Rlecuyer and spring packages are required respectively).

Parameters	Description
Cl	Cluster
Type	Random number generation mode type= "Rngstream" (default) type= "Sprng"
Seed	Random seeds
Prngkind	The generator form of sprng
Para	Additional parameters of the generator

CLUSTERSETUPRNG (cl, type = "Rngstream", ...)
1. If the SPRNG parameter is set, the CLUSTERSETUPSPRNG function is called. 2. If you set the Rngstream parameter, the CLUSTERSETUPSPRNG function is called.
Clustersetuprngstream (cl, Seed=rep (12345,6), ...)
Load the RSPRNG package and initialize each node separately. See the INIT.SPRNG documentation for further details.
CLUSTERSETUPSPRNG (cl, seed = round (2^32 * runif (1)), Prngkind = "Default", para = 0, ...)
Load the Rlecuyer package, create a data stream at each node and assign it to each node.

# # Not run:
clustersetupsprng (CL)
clustersetupsprng (cl, seed=1234)
clustersetuprng (cl, Seed=rep (1,6))
# # END (not run)

1 2 3) 4 5 3.snow-startstop

snow-startstop starting and stopping snow Clusters
makecluster (spec,type = Getclusterop