 |
 |
 |
numarray
User's Manual |
 |
 |
 |
| sequence=None, shape=None, typecode='O') |
The first argument, sequence, can be an arbitrary sequence of Python
objects, such as a list, tuple, or another object array.
>>> import numarray.objects as obj >>> class C: ... pass >>> c = C() >>> a = obj.array([c, c, c]) >>> a ObjectArray([c, c, c])
Like objects in Python lists, objects in object arrays are referred to, not copied, so changes to the objects are reflected in the originals because they are one and the same.
>>> a[0].attribute = 'this'
>>> c.attribute
'this'
The second argument, shape, optionally specifies the shape of the
array. If no shape is specified, the shape is implied by the
sequence.
>>> import numarray.objects as obj >>> class C: ... pass >>> c = C() >>> a = obj.fromlist([c, c, c]) >>> a ObjectArray([c, c, c])
The last argument, typecode, is there for backward compatibility with
Numeric; it must be specified as 'O'.
| obj) |
asarray converts sequences which are not object arrays into object
arrays. If obj is already an ObjectArray, it is returned
unaltered.
>>> import numarray.objects as obj >>> a = obj.asarray([1,''this'',''that'']) >>> a ObjectArray([1 'this' 'that']) >>> b = obj.asarray(a) >>> b is a True
| selector, population, output=None) |
choose selects elements from population based on the values in
selector, either returning the selected array or storing it in the
optional ObjectArray specified by output. selector should
be an integer sequence where each element is within the range 0 to
lenpopulation. population should be a sequence of
ObjectArrays. The shapes of selector and each element of
population must be mutually broadcastable.
>>> import numarray.objects as obj
>>> s = num.arange(25, shape=(5,5)) % 3
>>> p = obj.fromlist(["foo", 1, {"this":"that"}])
>>> obj.choose(s, p)
ObjectArray([['foo', 1, {'this': 'that'}, 'foo', 1],
[{'this': 'that'}, 'foo', 1, {'this': 'that'}, 'foo'],
[1, {'this': 'that'}, 'foo', 1, {'this': 'that'}],
['foo', 1, {'this': 'that'}, 'foo', 1],
[{'this': 'that'}, 'foo', 1, {'this': 'that'}, 'foo']])
| objects, axis=-1, output=None) |
sort sorts the elements from objects along the specified
axis. If an output array is specified, the result is stored there
and the return value is None, otherwise the sort is returned.
>>> import numarray.objects as obj
>>> a = obj.ObjectArray(shape=(5,5))
>>> a[:] = range(5,0,-1)
>>> obj.sort(a)
ObjectArray([[1, 2, 3, 4, 5],
[1, 2, 3, 4, 5],
[1, 2, 3, 4, 5],
[1, 2, 3, 4, 5],
[1, 2, 3, 4, 5]])
>>> a[:] = range(5,0,-1)
>>> a.transpose()
>>> obj.sort(a, axis=0)
ObjectArray([[1, 1, 1, 1, 1],
[2, 2, 2, 2, 2],
[3, 3, 3, 3, 3],
[4, 4, 4, 4, 4],
[5, 5, 5, 5, 5]])
| objects, axis=-1, output=None) |
argsort returns the sort order for the elements from objects
along the specified axis. If an output array is specified, the result
is stored there and the return value is None, otherwise the sort order is
returned.
>>> import numarray.objects as obj
>>> a = obj.ObjectArray(shape=(5,5))
>>> a[:] = ['e','d','c','b','a']
>>> obj.argsort(a)
array([[4, 3, 2, 1, 0],
[4, 3, 2, 1, 0],
[4, 3, 2, 1, 0],
[4, 3, 2, 1, 0],
[4, 3, 2, 1, 0]])
| objects, indices, axis=0) |
take returns elements of objects specified by tuple of index
arrays indices along the specified axis.
>>> import numarray.objects as obj >>> o = obj.fromlist(range(10)) >>> a = obj.arange(5)*2 >>> obj.take(o, a) ObjectArray([0, 2, 4, 6, 8])
| objects, indices, values, axis=-1) |
>>> import numarray.objects as obj >>> o = obj.fromlist(range(10)) >>> a = obj.arange(5)*2 >>> obj.put(o, a, 0); o ObjectArray([0, 1, 0, 3, 0, 5, 0, 7, 0, 9])
| objects1, objects2, out=None) |
numarray.objects defines universal functions which are named after and
use the operators defined in the standard library module operator.py. In
addition, the operator hooks of the ObjectArray class are defined to
call the operators. add applies the add operator to
corresponding elements of objects1 and objects2. Like the ufuncs
in the numerical side of numarray, the object ufuncs support reduction and
accumulation. In addition to add, there are ufuncs defined for every unary
and binary operator function in the standard library module operator.py.
Some of these are given additional synonyms so that they use numarray naming
conventions, e.g. sub has an alias named subtract.
>>> import numarray.objects as obj
>>> a = obj.fromlist(["t","u","w"])
>>> a
ObjectArray(['t', 'u', 'w'])
>>> a+a
ObjectArray(['tt', 'uu', 'ww'])
>>> obj.add(a,a)
ObjectArray(['tt', 'uu', 'ww'])
>>> obj.add.reduce(a)
'tuw' # not, as in the docs, an ObjectArray
>>> obj.add.accumulate(a)
ObjectArray(['t', 'tu', 'tuw']) # w, not v
>>> a = obj.fromlist(["t","u","w"])
>>> a
ObjectArray(['t', 'u', 'w'])
>>> a+a
ObjectArray(['tt', 'uu', 'ww'])
>>> obj.add(a,a)
ObjectArray(['tt', 'uu', 'ww'])
>>> obj.add.reduce(a)
ObjectArray('tuv')
>>> obj.add.accumulate(a)
ObjectArray(['t', 'tu', 'tuv'])
|
|
|
numarray
User's Manual |
|
|
|