R: Setup parallel RNG on all slaves

mpi.setup.sprng {Rmpi}

R Documentation

Setup parallel RNG on all slaves

Description

mpi.setup.sprng setups SPRNG (rsprng) on all slaves. mpi.setup.rngstream setups RNGstream (rlecuyer) on all slaves.

Usage

mpi.setup.sprng(seed=runif(1, 1, 2^31-1), kindprng="default", para=0, comm=1)
mpi.setup.rngstream(seed=c(runif(3,0,2^32-210),runif(3,0,2^32-22854)), 
		comm = 1)

Arguments

`seed`	an integer or a vector of 6 integers of random number seed(s).
`kindprng`	a character string of the desired kind of parallel random number generator.
`para`	additional parameters for the parallel random number generators. If `para` is 0, default parameters for each PRNG are used. When invalid parameter is given, a warning is issued and the default parameter is used.
`comm`	a comm number.

Details

mpi.setup.sprng and mpi.setup.rngstream can be run only on master node. Please use one and only one of them on one session. According to SPRNG specification, one can only run mpi.setup.sprng once before using any RNG on slaves. Otherwise independence of streams on slaves are not guarantied. If several simulations with different seeds are desirable on one session, mpi.setup.rngstream is preferred.

Value

If setup fails, an error message will be returned. It is most likely that the rsprng or rlecuyer is not installed properly on all slaves. mpi.setup.sprng and mpi.setup.rngstream save their seeds as .Sprng.seed and .Rngstream.seed respectively.

Author(s)

Hao Yu

[Package Rmpi version 0.5-9 Index]