Cover methods

The module mapper.cover contains classes for various covers of the filter range. Each such class provides an iterator over the levels and a method data_index which returns the indices of the data points belonging to a certain level.

class mapper.cover.cube_cover_primitive(intervals=10, overlap=50, info={})

Primitive multidimensional cover. The patches are cubes, aligned to the coordinate axes and distributed on a rectangular grid.

Parameters :

intervals : integer or list of integers

Number of intervals for each filter component

overlap : float or list of floats

Percentage of overlap between the intervals

info : dict

(Ask Aravind)

class mapper.cover.balanced_cover_1d(intervals=10, overlap=50, info={})

One-dimensional balanced cover. The interval boundaries are distributed so that each patch covers the same fraction of the data set.

Parameters :

intervals : int

Number of intervals for each filter component

overlap : float

Percentage of overlap between the intervals

info : dict

(Ask Aravind)

class mapper.cover.subrange_decomposition_cover_1d(intervals_by_subrange, overlaps_by_subrange, subrange_boundaries, info={})