sparsetable {frab}

Generalized sparse tables: sparsetable objects

Description

Package idiom for creating and manipulating sparsetable objects

Usage

sparsetable(i,v=1)
rspar(n=15,l=3,d=3)
rspar2(n=15,l=6)
rsparr(n=20,d=6,l=5,s=4)
sparsetable_to_array(x)
array_to_sparsetable(x)
sparsetable_to_frab(x)
## S4 method for signature 'sparsetable'
index(x)
## S4 method for signature 'sparsetable'
values(x)
## S4 method for signature 'sparsetable'
dimnames(x)
## S4 method for signature 'sparsetable'
dim(x)

Arguments

x

In functions like index(), an object of class sparsetable

i, v

In standard constructor function sparsetable(), argument i is the index matrix of strings, and v a numeric vector of values

n, l, d, s

In functions rspar(), rspar2(), and rsparr(), n is the number of terms, l the number of letters, d the dimensionality and s the number of distinct marginal values to return

Details

Most functions here mirror their equivalent in the spray package [which the C code is largely copied from] or the frab functionality. So, for example, num_eq_sparsetable() is the equivalent of num_eq_spray().

The print method treats arity-2 sparsetable objects differently from other arities. By default, arity-2 sparsetable objects are displayed as two-dimensional tables. Control this behaviour with option print_2dsparsetables_as_matrices:

  options("print_2dsparsetables_as_matrices" = FALSE)

The default value for this option, non-FALSE (including its out-of-the-box status of “unset”), directs the print method to coerce arity-2 sparsetable objects to two-dimensional tables before printing. If this option is FALSE, arity-2 sparsetables are printed using matrix index form, just the same as any other arity.

Functions rspar(), rspar2(), and rsparr() create random sparsetable objects of increasing complexity. The defaults are chosen so that the returned frabs are of sensible sizes.

Function drop() takes a sparsetable object of arity one and coerces to a frab object.

Function dim() returns a named vector, with names being the dimnames of its argument.

Extraction and replacement methods are a subset of spray methods, but most should work. There is special dispensation so that standard idiom for arrays [e.g. x['a','b','a'] and x['a','b','a'] <- 55] should work as expected, although the general expectation is that access and replacement use (character) matrices and an index object. However, indexing by disord and disindex objects should also work [e.g. x[x>7]].

The spray source code and the sparsetable functionality have about 90% overlap; there were enough small differences between the codes to make it worth maintaining two sets of source code, IMO.

There is a discussion of package idiom in the vignette, vignette("frab").

Note

The pronunciation of “sparsetable” has the emphasis on the first syllable, so it rhymes with [British river-port town] “Barnstaple”.

Author(s)

Robin K. S. Hankin

See Also

frab-class

Examples

sparsetable(matrix(sample(letters[1:4],36,replace=TRUE),ncol=2),1:18)
sparsetable(matrix(sample(letters[1:4],39,replace=TRUE),ncol=3),1:13)

(x <- rspar2(9))
(y <- rspar2(9))
x + y

x["KT","FF"] <- 100
x

rsparr()

a <- rspar(d=4)
asum(a,"Feb")