SparseArray

class gammapy.maps.SparseArray(shape, idx=None, data=None, dtype=<class 'float'>, fill_value=0.0)[source]

Bases: object

Sparse N-dimensional array object.

This class implements a data structure for sparse n-dimensional arrays such that only non-zero data values are allocated in memory. Supports numpy conventions for indexing and slicing logic.

Parameters:
shape : tuple of ints

Shape of array.

idx : ndarray, optional

Flattened index vector that initializes the array. If none then an empty array will be created.

data : ndarray, optional

Flattened data vector that initializes the array. If none then an empty array will be created.

dtype : data-type, optional

Type of data vector.

fill_value : scalar, optional

Value assigned to array elements that are not allocated in memory.

Examples

A SparseArray is created in the same way as ndarray by passing the array shape to the constructor with an optional argument for the array type:

>>> import numpy as np
>>> from gammapy.maps import SparseArray
>>> v = SparseArray((10,20), dtype=float)

Alternatively you can create a new SparseArray from an ndarray with from_array:

>>> x = np.ones((10,20))
>>> v = SparseArray.from_array(x)

SparseArray follows numpy indexing and slicing conventions for setting/getting array elements. The primary difference with respect to the behavior of ndarray is that indexing always returns a copy rather than a view.

>>> v[0,0] = 1.0
>>> print(v[0,0])
1.0
>>> v[:,0] = 1.0
>>> print(v[:,0])
[1. 1. 1. 1. 1. 1. 1. 1. 1. 1.]

Attributes Summary

data Return the sparsified data array.
dtype Return the type of the data array member.
idx Return flattened index vector.
ndim Array number of dimensions (int).
shape Array shape.
size Return current number of elements.

Methods Summary

from_array(data[, min_value]) Create a SparseArray from a numpy array.
get(self, idx_in) Get array values at indices idx_in.
set(self, idx_in, vals[, fill]) Set array values at indices idx_in.
sum(self[, axis, dtype, out, keepdims])

Attributes Documentation

data

Return the sparsified data array.

dtype

Return the type of the data array member.

idx

Return flattened index vector.

ndim

Array number of dimensions (int).

shape

Array shape.

size

Return current number of elements.

Methods Documentation

classmethod from_array(data, min_value=0.0)[source]

Create a SparseArray from a numpy array.

Parameters:
data : numpy.ndarray

Input data array.

min_value : float

Threshold for sparsifying the data vector.

Returns:
out : SparseArray

Output sparse array.

get(self, idx_in)[source]

Get array values at indices idx_in.

set(self, idx_in, vals, fill=False)[source]

Set array values at indices idx_in.

sum(self, axis=None, dtype=None, out=None, keepdims=False, **unused_kwargs)[source]