Learn R Programming

SimDesign (version 2.17.1)

genSeeds: Generate random seeds

Description

Generate seeds to be passed to runSimulation's seed input. Values are sampled from 1 to 2147483647, or are generated using L'Ecuyer-CMRG's (2002) method (returning either a list if arrayID is omitted, or the specific row value from this list if arrayID is included).

Usage

genSeeds(design = 1L, iseed = NULL, arrayID = NULL, old.seeds = NULL)

gen_seeds(...)

Arguments

design

design matrix that requires a unique seed per condition, or a number indicating the number of seeds to generate. Default generates one number

iseed

the initial set.seed number used to generate a sequence of independent seeds according to the L'Ecuyer-CMRG (2002) method. This is recommended whenever quality random number generation is required across similar (if not identical) simulation jobs (e.g., see runArraySimulation). If arrayID is not specified then this will return a list of the associated seed for the full design

arrayID

(optional) single integer input corresponding to the specific row in the design object when using the iseed input. This is used in functions such as runArraySimulation to pull out the specific seed rather than manage a complete list, and is therefore more memory efficient

old.seeds

(optional) vector or matrix of last seeds used in previous simulations to avoid repeating the same seed on a subsequent run. Note that this approach should be used sparingly as seeds set more frequently are more likely to correlate, and therefore provide less optimal random number behaviour (e.g., if performing a simulation on two runs to achieve 5000 * 2 = 10,000 replications this is likely reasonable, but for simulations with 100 * 2 = 200 replications this is more likely to be sub-optimal). Length must be equal to the number of rows in design

...

does nothing

Author

Phil Chalmers rphilip.chalmers@gmail.com

Examples

Run this code

# generate 1 seed (default)
genSeeds()

# generate 5 unique seeds
genSeeds(5)

# generate from nrow(design)
design <- createDesign(factorA=c(1,2,3),
                       factorB=letters[1:3])
seeds <- genSeeds(design)
seeds

# construct new seeds that are independent from original (use this sparingly)
newseeds <- genSeeds(design, old.seeds=seeds)
newseeds

# can be done in batches too
newseeds2 <- genSeeds(design, old.seeds=cbind(seeds, newseeds))
cbind(seeds, newseeds, newseeds2) # all unique

############
# generate seeds for runArraySimulation()
(iseed <- genSeeds())  # initial seed
seed_list <- genSeeds(design, iseed=iseed)
seed_list

# expand number of unique seeds given iseed (e.g., in case more replications
# are required at a later date)
seed_list_tmp <- genSeeds(nrow(design)*2, iseed=iseed)
str(seed_list_tmp) # first 9 seeds identical to seed_list

# more usefully for HPC, extract only the seed associated with an arrayID
arraySeed.15 <- genSeeds(nrow(design)*2, iseed=iseed, arrayID=15)
arraySeed.15

Run the code above in your browser using DataLab