generate.subspaces: Random generation of consecutive subspaces

In edouardfouche/R-streamgenerator: Statistical multidimensional synthetic stream generator

Description

Random generation of partly overlapping consecutive subspaces with a minimum number of dimensions mindim and a maximum number of dimensions maxdim from a total number of dimensions dim.

Usage

generate.subspaces(dim = 20, mindim = 2, maxdim = 4,
  allowOverlap = FALSE)

Arguments

dim	Total number of dimensions from which to generate the subspaces.
mindim	Minimum number of dimensions for a subspace.
maxdim	Maximum number of dimensions for a subspace.
allowOverlap	States whether overlap between subspaces are allowed. (might be a bit more complex if TRUE.). Note that full overlap and subset relation are never allowed.

Details

Generate at most round((dim*0.75)/length(mindim:maxdim)/x) subspaces of each size x For example, for generate.subspaces(dim=100, maxdim=5, mindim=2):

• at most 4 5-dim subspaces

• at most 5 4-dim subspaces

• at most 6 3-dim subspaces

• at most 10 2-dim subspaces This is done so to make sure that roughly a least 25 This also make sure the dimensions are not dominated by high-dimensional subspaces TODO: Can this be improved?

Value

A list of partly overlapping subspaces spanning over the dim dimensions

Author(s)

Edouard Fouché, [e-mail]

Examples

subspaces <- generate.subspaces(dim=100, maxdim=5, mindim=2)

edouardfouche/R-streamgenerator documentation built on May 15, 2019, 11:02 p.m.