| SparseArraySeed-class {DelayedArray} | R Documentation |
SparseArraySeed objects are used internally to support block processing of array-like objects.
## Constructor function: SparseArraySeed(dim, nzindex=NULL, nzdata=NULL, check=TRUE) ## Getters (in addition to dim() and length()): nzindex(x) nzdata(x) sparsity(x) ## Two low-level utilities: dense2sparse(x) sparse2dense(sas)
dim |
The dimensions (specified as an integer vector) of the SparseArraySeed object to create. |
nzindex |
A matrix containing the array indices of the nonzero data. This must be an integer matrix like one returned by
|
nzdata |
A vector of length |
check |
Should the object be validated upon construction? |
x |
A SparseArraySeed object for the An array-like object for |
sas |
A SparseArraySeed object. |
For SparseArraySeed(): A SparseArraySeed instance.
For nzindex(): The matrix containing the array indices of the
nonzero data.
For nzdata(): The vector of nonzero data.
For sparsity(): The number of zero-valued elements
in the implicit array divided by the total number of array
elements (a.k.a. the length of the array).
For dense2sparse(): A SparseArraySeed instance.
For sparse2dense(): An ordinary array.
The read_sparse_block function.
blockApply for more information about block processing
of array-like objects.
DelayedArray objects.
arrayInd in the base package.
array objects in base R.
## ---------------------------------------------------------------------
## EXAMPLE 1
## ---------------------------------------------------------------------
nzindex1 <- rbind(c(2,4,3), c(2,1,3), c(5,4,3), c(5,3,3),
c(5,4,1), c(5,1,1), c(5,4,2), c(5,4,1))
nzdata1 <- 11.11 * seq_len(nrow(nzindex1))
sas1 <- SparseArraySeed(5:3, nzindex1, nzdata1)
dim(sas1) # the dimensions of the implicit array
length(sas1) # the length of the implicit array
nzindex(sas1)
nzdata(sas1)
sparsity(sas1)
sparse2dense(sas1)
as.array(sas1) # same as sparse2dense(sas1)
## Not run:
as.matrix(sas1) # error!
## End(Not run)
## ---------------------------------------------------------------------
## EXAMPLE 2
## ---------------------------------------------------------------------
m2 <- matrix(c(5:-2, rep.int(c(0L, 99L), 11)), ncol=6)
sas2 <- dense2sparse(m2)
dim(sas2)
length(sas2)
nzindex(sas2)
nzdata(sas2)
sparsity(sas2)
stopifnot(identical(sparse2dense(sas2), m2))
as.matrix(sas2) # same as sparse2dense(sas2)
t(sas2)
stopifnot(identical(as.matrix(t(sas2)), t(as.matrix(sas2))))
## Go back and forth between SparseArraySeed and dgCMatrix objects:
M2 <- as(sas2, "dgCMatrix")
stopifnot(identical(M2, as(m2, "dgCMatrix")))
sas2b <- as(M2, "SparseArraySeed")
## 'sas2b' is the same as 'sas2' except that
## 'nzdata(sas2b)' is of type numeric instead of integer:
all.equal(sas2b, sas2)
typeof(nzdata(sas2b)) # numeric
typeof(nzdata(sas2)) # integer
## ---------------------------------------------------------------------
## SEED CONTRACT
## ---------------------------------------------------------------------
## SparseArraySeed objects comply with the "seed contract".
## In particular they support extract_array():
extract_array(sas1, list(c(5, 3:2, 5), NULL, 3))
## See '?extract_array' for more information about the "seed contract".
## This means that they can be wrapped in a DelayedArray object:
A1 <- DelayedArray(sas1)
A1
## A big very sparse DelayedMatrix object:
nzindex3 <- cbind(sample(25000, 600000, replace=TRUE),
sample(195000, 600000, replace=TRUE))
nzdata3 <- runif(600000)
sas3 <- SparseArraySeed(c(25000, 195000), nzindex3, nzdata3)
sparsity(sas3)
M3 <- DelayedArray(sas3)
M3
colSums(M3[ , 1:20])