pboot: Parallel Bootstrapping

Description Arguments Details Author(s) See Also

View source: R/pboot.R


pboot() generates R bootstrap replicates of a statistic applied to data. It implements a parallel version of the bootstrapping method boot() from the boot R package.



array of data, if a 2D array then each row is considered as one multivariate observation


function, when sim is set to parametric, the first argument to statistic must be the data. For each replicate a simulated dataset returned by ran.gen will be passed. In all other cases, statistic must take at least two arguments. The first argument passed will always be the original data. The second will be a vector of indices, frequencies or weights which define the bootstrap sample.


number of bootstrap replicates


string, indicates the type of simulation. The default value is "ordinary". Other possible values are parametric, balanced, permutation, and antithetic. Importance resampling is specified by including importance weights; the type of importance resampling must still be specified but may only be ordinary or balanced in this case.


string, indicates what the second argument of statistic represents. The default value is i for indices. Other possible values are f for frequencies and w for weights. It is not used when sim is set to parametric.


vector of integer, specifies the strata for multi-sample problems. This may be specified for any simulation, but is ignored when sim is set to parametric. When strata is supplied for a nonparametric bootstrap, the simulations are done within the specified strata.


vector of influence values evaluated at the observations. This is used only when sim is set to antithetic. If not supplied, they are calculated through a call to empinf. This will use the infinitesimal jackknife provided that stype is set to w otherwise the usual jackknife is used.


the number of predictions which are to be made at each bootstrap replicate. This is most useful for (generalized) linear models. This can only be used when sim is ordinary. m will usually be a single integer but, if there are strata, it may be a vector with length equal to the number of strata, specifying how many of the errors for prediction should come from each strata. The actual predictions should be returned as the final part of the output of statistic, which should also take an argument giving the vector of indices of the errors to be used for the predictions.


array of importance weights. If a vector then it should have as many elements as there are observations in the input data. When simulation from more than one set of weights is required, weights should be a matrix where each row of the matrix is one set of importance weights. If weights is a matrix then the number of bootstrap replicates R must be a vector of length nrow(weights). This parameter is ignored if sim is not set to ordinary or balanced.


function, used only when sim is set to parametric. It describes how random values are to be generated. It should be a function of two arguments. The first argument should be the observed data and the second argument consists of any other information needed (e.g. parameter estimates). The second argument may be a list, allowing any number of items to be passed to ran.gen. The returned value should be a simulated data set of the same form as the observed data which will be passed to statistic to get a bootstrap replicate. It is important that the returned value be of the same shape and type as the original dataset. If ran.gen is not specified, the default is a function which returns the original input data in which case all simulation should be included as part of statistic. Setting sim to parametric and using a suitable ran.gen allows the user to implement any types of nonparametric resampling which are not supported directly.


secong argument to ran.gen, typically these will be maximum likelihood estimates of the parameters. For efficiency mle is often a list containing all of the objects needed by ran.gen which can be calculated using the original data set only.


boolean, can only be set to TRUE if sim is set to ordinary, stype is set to I and n is set to 0. Otherwise it is ignored and generates a warning. By default a n by R index array is created which can be large. If simple is set to TRUE, this is avoided by sampling separately for each replication, which is slower but uses less memory.


other named arguments for statistic which are passed unchanged each time.


This version is an early but fully working prototype. However, it is not compatible with other SPRINT functions, i.e. you cannot bootstrap other parallel functions from the SPRINT library. It is therefore recommended to use it only as a standalone function.

N.B. Please see the SPRINT User Guide for how to run the code in parallel using the mpiexec command.


University of Edinburgh SPRINT Team [email protected] www.r-sprint.org

See Also


sprint documentation built on May 30, 2017, 8:25 a.m.