BinaryFiles: Save/read binary files

In SFSI: Sparse Family and Selection Index

Save/read binary files

Description

Save/read a numeric data as a fortran-formatted binary file at a defined precision (single or double).

Usage

saveBinary(X, file = paste0(tempdir(), "/file.bin"), 
           precision.format = c("double","single"), 
           verbose = TRUE)
  
readBinary(file = paste0(tempdir(), "/file.bin"), 
           rows = NULL, cols = NULL, 
           drop = TRUE, verbose = TRUE)

Arguments

X	(numeric matrix) Data to save
file	(character) Name of the binary file to save/read
precision.format	(character) Either 'single' or 'double' for numeric precision and memory occupancy (4 bytes/32-bit or 8 bytes/64-bit, respectively) of the matrix to save
rows	(integer vector) Which rows are to be read from the file. Default rows=NULL will read all the rows
cols	(integer vector) Which columns are to be read from the file. Default cols=NULL will read all the columns
drop	Either TRUE or FALSE to whether return a uni-dimensional vector when data is a matrix with either 1 row or 1 column
verbose	TRUE or FALSE to whether printing file information

Value

Function 'saveBinary' does not return any value but print a description of the file saved.

Function 'readBinary' returns the data that was read.

Examples

  require(SFSI)
  data(wheatHTP)

  # Read a numeric matrix
  X = scale(X_E1)
  X[1:10,1:5]
  
  # Save matrix as double-precision
  filename1 = paste0(tempdir(),"/Matrix1.bin")
  saveBinary(X, filename1, precision.format="double")
  
  # Save matrix as single-precision
  filename2 = paste0(tempdir(),"/Matrix2.bin")
  saveBinary(X, filename2, precision.format="single")

  # Read the double-precision matrix
  X2 = readBinary(filename1)
  X2[1:10,1:5]
  max(abs(X-X2))             # No loss of precision
  file.info(filename1)$size  # Size of the file

  # Read the single-precision matrix
  X3 = readBinary(filename2)
  X3[1:10,1:5]
  max(abs(X-X3))             # Loss of precision
  file.info(filename2)$size  # But smaller-size file

  # Read specific rows and columns
  rows = c(2,4,5,8,10)
  cols = c(1,2,5)
  X2 = readBinary(filename1,rows=rows,cols=cols)
  X2
  # Equal to: 
  X[rows,cols]

SFSI documentation built on Nov. 18, 2023, 9:06 a.m.