Common operations on DelayedArray objects

Description

Common operations on DelayedArray objects.

Details

The operations currently supported on DelayedArray objects are:

Delayed operations:

• all the members of the Ops, Math, and Math2 groups

• !

• is.na, is.finite, is.infinite, is.nan

• nchar, tolower, toupper

• pmax2 and pmin2

• rbind and cbind (documented in DelayedArray)

Block-processed operations:

• anyNA, which

• all the members of the Summary group

• mean

• apply

• matrix multiplication (%*%) of an ordinary matrix by a DelayedMatrix object

• matrix row/col summarization [DelayedMatrix objects only]: rowSums, colSums, rowMeans, colMeans, rowMaxs, colMaxs, rowMins, colMins, rowRanges, and colRanges

See Also

• is.na, !, mean, apply, and %*% in the base package for the corresponding operations on ordinary arrays or matrices.

• rowSums in the base package and rowMaxs in the matrixStats package for row/col summarization of an ordinary matrix.

• setRealizationBackend for how to set a realization backend.

• writeHDF5Array in the HDF5Array package for writting an array-like object to an HDF5 file and other low-level utilities to control the location of automatically created HDF5 datasets.

• DelayedArray objects.

• HDF5Array objects in the HDF5Array package.

• S4groupGeneric in the methods package for the members of the Ops, Math, and Math2 groups.

• array objects in base R.

Examples

library(HDF5Array)
toy_h5 <- system.file("extdata", "toy.h5", package="HDF5Array")
h5ls(toy_h5)

M1 <- HDF5Array(toy_h5, "M1")
range(M1)
M1 >= 0.5 & M1 < 0.75
log(M1)

M2 <- HDF5Array(toy_h5, "M2")
pmax2(M2, 0)

M3 <- rbind(M1, t(M2))
M3

## ---------------------------------------------------------------------
## MATRIX MULTIPLICATION
## ---------------------------------------------------------------------

## Matrix multiplication is not delayed: the output matrix is realized
## block by block. The current "realization backend" controls where
## realization happens e.g. in memory if set to NULL or in an HDF5 file
## if set to "HDF5Array". See '?realize' for more information about
## "realization backends".
## The output matrix is returned as a DelayedMatrix object with no delayed
## operations on it. The exact class of the object depends on the backend
## e.g. it will be HDF5Matrix with "HDF5Array" backend.

m <- matrix(runif(50000), ncol=nrow(M1))

## Set backend to NULL for in-memory realization:
setRealizationBackend()
P1 <- m %*% M1
P1

## Set backend to HDF5Array for realization in HDF5 file:
setRealizationBackend("HDF5Array")

## With the HDF5Array backend, the output matrix will be written to an
## automatic location on disk:
getHDF5DumpFile()  # HDF5 file where the output matrix will be written
lsHDF5DumpFile()

P2 <- m %*% M1
P2

lsHDF5DumpFile()

## Use setHDF5DumpFile() and setHDF5DumpName() from the HDF5Array package
## to control the location of automatically created HDF5 datasets.

stopifnot(identical(as.array(P1), as.array(P2)))

## ---------------------------------------------------------------------
## MATRIX ROW/COL SUMMARIZATION
## ---------------------------------------------------------------------

rowSums(M1)
colSums(M1)

rowMeans(M1)
colMeans(M1)

rmaxs <- rowMaxs(M1)
cmaxs <- colMaxs(M1)

rmins <- rowMins(M1)
cmins <- colMins(M1)

rranges <- rowRanges(M1)
cranges <- colRanges(M1)

stopifnot(identical(cbind(rmins, rmaxs, deparse.level=0), rranges))
stopifnot(identical(cbind(cmins, cmaxs, deparse.level=0), cranges))