colSums: Column and row sums for HDF5Matrix

In BigDataStatMeth: Title: Scalable Statistical Computing with HDF5-Backed Matrices

Column and row sums for HDF5Matrix

Description

Block-wise computation of column and row sums without loading the full matrix into RAM. Results can optionally be persisted to an HDF5 file.

Usage

colSums(x, na.rm = FALSE, dims = 1L, ...)

rowSums(x, na.rm = FALSE, dims = 1L, ...)

## S3 method for class 'HDF5Matrix'
colSums(
  x,
  na.rm = FALSE,
  dims = 1,
  paral = NULL,
  wsize = NULL,
  threads = NULL,
  save_to = NULL,
  overwrite = TRUE,
  ...
)

## S3 method for class 'HDF5Matrix'
rowSums(
  x,
  na.rm = FALSE,
  dims = 1,
  paral = NULL,
  wsize = NULL,
  threads = NULL,
  save_to = NULL,
  overwrite = TRUE,
  ...
)

Arguments

x	An HDF5Matrix object.
na.rm	Ignored (included for compatibility with the base generic).
dims	Ignored (included for compatibility with the base generic).
...	Ignored.
paral	Logical or NULL. Enable OpenMP parallelisation.
wsize	Integer or NULL. Block size for HDF5 reads (NULL = auto).
threads	Integer or NULL. Number of OpenMP threads (NULL = auto).
save_to	Where to save the result (see Details). NULL returns a plain R vector; a character string "group/dataset" saves in the same file as x; a named list list(file = "f.h5", path = "group/dataset") saves in a different file. In both non-NULL cases an HDF5Matrix handle is returned.
overwrite	Logical. Overwrite an existing dataset at save_to? (Currently always TRUE when save_to is not NULL.)

Value

A numeric vector (when save_to = NULL) or an HDF5Matrix handle to the persisted result.

Examples

tmp <- tempfile(fileext = ".h5")
X <- hdf5_create_matrix(tmp, "data/M", data = matrix(rnorm(200), 20, 10))
cs <- colSums(X)         # R vector
hdf5_close_all()
unlink(tmp)

BigDataStatMeth documentation built on May 15, 2026, 1:07 a.m.