Python numpy.ufunc() Examples

def __array_prepare__(self, result, context=None):
        """
        Gets called prior to a ufunc
        """

        # nice error message for non-ufunc types
        if context is not None and not isinstance(self._values, np.ndarray):
            obj = context[1][0]
            raise TypeError("{obj} with dtype {dtype} cannot perform "
                            "the numpy op {op}".format(
                                obj=type(obj).__name__,
                                dtype=getattr(obj, 'dtype', None),
                                op=context[0].__name__))
        return result

    # complex 

def get_module_functions(modules):
  """Finds functions that do not have implemented derivatives.

  Args:
    modules: A list of Python modules. Functions contained in these modules
        will be checked for membership in 'implemented', and if not found,
        will be added to an 'unimplemented' set
    implemented: A Python object containing implemented derivatives. A function
        should be checkable for membership using the `fn in implemented` syntax.

  Returns:
    module_fns: A set of functions, builtins or ufuncs in `modules`.
  """
  module_fns = set()
  for module in modules:
    for key in dir(module):
      attr = getattr(module, key)
      if isinstance(
          attr, (types.BuiltinFunctionType, types.FunctionType, numpy.ufunc)):
        module_fns.add(attr)
  return module_fns 

def __ua_convert__(value, dispatch_type, coerce):
        if dispatch_type is ufunc and hasattr(fn, value.name):
            return getattr(fn, value.name)

        if value is None:
            return None

        if dispatch_type is ndarray:
            if not coerce and not isinstance(value, xnd.xnd):
                return NotImplemented

            return convert(value, coerce=coerce)

        if dispatch_type is dtype:
            return ndt(str(value))

        return NotImplemented 

def __ua_convert__(value, dispatch_type, coerce):
    if dispatch_type is ufunc:
        return getattr(np, value.name)

    if value is None:
        return None

    if dispatch_type is ndarray:
        if not coerce:
            if not isinstance(value, sparse.SparseArray):
                return NotImplemented

        if isinstance(value, sparse.SparseArray):
            return value

        return sparse.as_coo(np.asarray(value))

    return value 

def astype(self, dtype, copy=True):
        """
        Copy of the array, cast to a specified type.

        See also
        --------
        scipy.sparse.coo_matrix.astype : SciPy sparse equivalent function
        numpy.ndarray.astype : NumPy equivalent ufunc.
        :obj:`COO.elemwise`: Apply an arbitrary element-wise function to one or two
            arguments.
        """
        # this matches numpy's behavior
        if self.dtype == dtype and not copy:
            return self
        return self.__array_ufunc__(
            np.ndarray.astype, "__call__", self, dtype=dtype, copy=copy
        ) 

def test_coverage(self):
        """Test that we cover all ufunc's"""

        all_np_ufuncs = set([ufunc for ufunc in np.core.umath.__dict__.values()
                             if isinstance(ufunc, np.ufunc)])

        all_q_ufuncs = (qh.UNSUPPORTED_UFUNCS |
                        set(qh.UFUNC_HELPERS.keys()))
        # Check that every numpy ufunc is covered.
        assert all_np_ufuncs - all_q_ufuncs == set()
        # Check that all ufuncs we cover come from numpy or erfa.
        # (Since coverage for erfa is incomplete, we do not check
        # this the other way).
        all_erfa_ufuncs = set([ufunc for ufunc in erfa_ufunc.__dict__.values()
                               if isinstance(ufunc, np.ufunc)])
        assert (all_q_ufuncs - all_np_ufuncs - all_erfa_ufuncs == set()) 

def __init__(self, scalar_op, inplace_pattern=None, name=None,
                 nfunc_spec=None, openmp=None):
        if inplace_pattern is None:
            inplace_pattern = {}
        self.name = name
        self.scalar_op = scalar_op
        self.inplace_pattern = inplace_pattern
        self.destroy_map = dict((o, [i]) for o, i in inplace_pattern.items())

        self.ufunc = None
        self.nfunc = None
        if nfunc_spec is None:
            nfunc_spec = getattr(scalar_op, 'nfunc_spec', None)
        self.nfunc_spec = nfunc_spec
        if nfunc_spec:
            self.nfunc = getattr(numpy, nfunc_spec[0])

        # precompute the hash of this node
        self._rehash()
        super(Elemwise, self).__init__(openmp=openmp) 

def __init__(self, scalar_op, inplace_pattern=None, name=None,
                 nfunc_spec=None, openmp=None):
        if inplace_pattern is None:
            inplace_pattern = {}
        self.name = name
        self.scalar_op = scalar_op
        self.inplace_pattern = inplace_pattern
        self.destroy_map = dict((o, [i]) for o, i in inplace_pattern.items())

        self.ufunc = None
        self.nfunc = None
        if nfunc_spec is None:
            nfunc_spec = getattr(scalar_op, 'nfunc_spec', None)
        self.nfunc_spec = nfunc_spec
        if nfunc_spec:
            self.nfunc = getattr(numpy, nfunc_spec[0])

        # precompute the hash of this node
        self._rehash()
        super(Elemwise, self).__init__(openmp=openmp) 

def _transform_batch(self, func, return_schema):
        from databricks.koalas.series import Series
        from databricks import koalas as ks

        if isinstance(func, np.ufunc):
            f = func
            func = lambda *args, **kwargs: f(*args, **kwargs)

        if return_schema is None:
            # TODO: In this case, it avoids the shortcut for now (but only infers schema)
            #  because it returns a series from a different DataFrame and it has a different
            #  anchor. We should fix this to allow the shortcut or only allow to infer
            #  schema.
            limit = ks.get_option("compute.shortcut_limit")
            pser = self._kser.head(limit)._to_internal_pandas()
            transformed = pser.transform(func)
            kser = Series(transformed)
            spark_return_type = kser.spark.data_type
        else:
            spark_return_type = return_schema

        pudf = pandas_udf(func, returnType=spark_return_type, functionType=PandasUDFType.SCALAR)
        return self._kser._with_new_scol(scol=pudf(self._kser.spark.column)).rename(self._kser.name) 

def _get_ufuncs():
    ufuncs = []
    ufunc_names = []
    for name in sorted(sc.__dict__):
        obj = sc.__dict__[name]
        if not isinstance(obj, np.ufunc):
            continue
        msg = KNOWNFAILURES.get(obj)
        if msg is None:
            ufuncs.append(obj)
            ufunc_names.append(name)
        else:
            fail = pytest.mark.xfail(run=False, reason=msg)
            ufuncs.append(pytest.param(obj, marks=fail))
            ufunc_names.append(name)
    return ufuncs, ufunc_names 

def set_ufunc(self, scalar_op):
        # This is probably a speed up of the implementation
        if isinstance(scalar_op, theano.scalar.basic.Add):
            self.ufunc = numpy.add
        elif isinstance(scalar_op, theano.scalar.basic.Mul):
            self.ufunc = numpy.multiply
        elif isinstance(scalar_op, theano.scalar.basic.Maximum):
            self.ufunc = numpy.maximum
        elif isinstance(scalar_op, theano.scalar.basic.Minimum):
            self.ufunc = numpy.minimum
        elif isinstance(scalar_op, theano.scalar.basic.AND):
            self.ufunc = numpy.bitwise_and
        elif isinstance(scalar_op, theano.scalar.basic.OR):
            self.ufunc = numpy.bitwise_or
        elif isinstance(scalar_op, theano.scalar.basic.XOR):
            self.ufunc = numpy.bitwise_xor
        else:
            self.ufunc = numpy.frompyfunc(scalar_op.impl, 2, 1) 

def set_ufunc(self, scalar_op):
        # This is probably a speed up of the implementation
        if isinstance(scalar_op, theano.scalar.basic.Add):
            self.ufunc = numpy.add
        elif isinstance(scalar_op, theano.scalar.basic.Mul):
            self.ufunc = numpy.multiply
        elif isinstance(scalar_op, theano.scalar.basic.Maximum):
            self.ufunc = numpy.maximum
        elif isinstance(scalar_op, theano.scalar.basic.Minimum):
            self.ufunc = numpy.minimum
        elif isinstance(scalar_op, theano.scalar.basic.AND):
            self.ufunc = numpy.bitwise_and
        elif isinstance(scalar_op, theano.scalar.basic.OR):
            self.ufunc = numpy.bitwise_or
        elif isinstance(scalar_op, theano.scalar.basic.XOR):
            self.ufunc = numpy.bitwise_xor
        else:
            self.ufunc = numpy.frompyfunc(scalar_op.impl, 2, 1) 

def __array_prepare__(self, result, context=None):
        """
        Gets called prior to a ufunc
        """

        # nice error message for non-ufunc types
        if context is not None and not isinstance(self._values, np.ndarray):
            obj = context[1][0]
            raise TypeError("{obj} with dtype {dtype} cannot perform "
                            "the numpy op {op}".format(
                                obj=type(obj).__name__,
                                dtype=getattr(obj, 'dtype', None),
                                op=context[0].__name__))
        return result

    # complex 

def helper_frexp(f, unit):
    if not unit.is_unity():
        raise UnitTypeError("Can only apply '{}' function to "
                            "unscaled dimensionless quantities"
                            .format(f.__name__))
    return [None], (None, None)


# TWO ARGUMENT UFUNC HELPERS
#
# The functions below take a two arguments. The output of the helper function
# should be two values: a tuple of two converters to be used to scale the
# inputs before being passed to the ufunc (None if no conversion is needed),
# and the unit the output will be in. 

def apply(self, func, axis=0, broadcast=False, reduce=False):
        """
        Analogous to DataFrame.apply, for SparseDataFrame

        Parameters
        ----------
        func : function
            Function to apply to each column
        axis : {0, 1, 'index', 'columns'}
        broadcast : bool, default False
            For aggregation functions, return object of same size with values
            propagated

        Returns
        -------
        applied : Series or SparseDataFrame
        """
        if not len(self.columns):
            return self
        axis = self._get_axis_number(axis)

        if isinstance(func, np.ufunc):
            new_series = {}
            for k, v in compat.iteritems(self):
                applied = func(v)
                applied.fill_value = func(v.fill_value)
                new_series[k] = applied
            return self._constructor(
                new_series, index=self.index, columns=self.columns,
                default_fill_value=self._default_fill_value,
                default_kind=self._default_kind).__finalize__(self)
        else:
            if not broadcast:
                return self._apply_standard(func, axis, reduce=reduce)
            else:
                return self._apply_broadcast(func, axis) 

def test_invariant_twoarg_scalar(self, ufunc):

        q_i1 = 4.7 * u.m
        q_i2 = 9.4 * u.km
        q_o = ufunc(q_i1, q_i2)
        assert isinstance(q_o, u.Quantity)
        assert q_o.unit == q_i1.unit
        assert_allclose(q_o.value, ufunc(q_i1.value, q_i2.to_value(q_i1.unit))) 

def _create_value_wrapper(self, wrapper, model):
        """Wraps a getter/setter function to support optionally passing in
        a reference to the model object as the second argument.
        If a model is tied to this parameter and its getter/setter supports
        a second argument then this creates a partial function using the model
        instance as the second argument.
        """

        if isinstance(wrapper, np.ufunc):
            if wrapper.nin != 1:
                raise TypeError("A numpy.ufunc used for Parameter "
                                "getter/setter may only take one input "
                                "argument")
        elif wrapper is None:
            # Just allow non-wrappers to fall through silently, for convenience
            return None
        else:
            inputs, params = get_inputs_and_params(wrapper)
            nargs = len(inputs)

            if nargs == 1:
                pass
            elif nargs == 2:
                self._model_required = True
                if model is not None:
                    # Don't make a partial function unless we're tied to a
                    # specific model instance
                    model_arg = inputs[1].name
                    wrapper = functools.partial(wrapper, **{model_arg: model})
            else:
                raise TypeError("Parameter getter/setter must be a function "
                                "of either one or two arguments")

        return wrapper 

def __array_wrap__(self, array, context=None):
        """
        Wrapper for multiplication with numpy arrays.
        """
        if type(context[0]) == np.ufunc:
            return NotImplemented
        else:
            return array 

def test_invariant_twoarg_one_arbitrary(self, ufunc, arbitrary):
        q_i1 = np.array([-3.3, 2.1, 10.2]) * u.kg / u.s
        q_o = ufunc(q_i1, arbitrary)
        assert isinstance(q_o, u.Quantity)
        assert q_o.unit == q_i1.unit
        assert_allclose(q_o.value, ufunc(q_i1.value, arbitrary)) 

def test_invariant_array(self, ufunc):

        q_i = np.array([-3.3, 2.1, 10.2]) * u.kg / u.s
        q_o = ufunc(q_i)
        assert isinstance(q_o, u.Quantity)
        assert q_o.unit == q_i.unit
        assert np.all(q_o.value == ufunc(q_i.value)) 

def __array_wrap__(self, result, context=None):
        """
        Gets called after a ufunc
        """
        return self._constructor(result, index=self.index,
                                 copy=False).__finalize__(self) 

def aggregate(cls, dataset, dimensions, function, **kwargs):
        reindexed = dataset.reindex(dimensions)
        grouped = (cls.groupby(reindexed, dimensions, list, 'raw')
                   if len(dimensions) else [((), reindexed.data)])

        rows = []
        for k, group in grouped:
            if isinstance(function, np.ufunc):
                reduced = function.reduce(group, axis=0, **kwargs)
            else:
                reduced = function(group, axis=0, **kwargs)
            rows.append(np.concatenate([k, (reduced,) if np.isscalar(reduced) else reduced]))
        return np.atleast_2d(rows), [] 

def _resolve_op(self, op, dataset, data, flat, expanded, ranges,
                    all_values, keep_index, compute, strict):
        args = op['args']
        fn = op['fn']
        kwargs = dict(op['kwargs'])
        fn_name = self._numpy_funcs.get(fn)
        if fn_name and hasattr(data, fn_name):
            if 'axis' not in kwargs and not isinstance(fn, np.ufunc):
                kwargs['axis'] = None
            fn = fn_name

        if isinstance(fn, basestring):
            accessor = kwargs.pop('accessor', None)
            fn_args = []
        else:
            accessor = False
            fn_args = [data]

        for arg in args:
            if isinstance(arg, dim):
                arg = arg.apply(
                    dataset, flat, expanded, ranges, all_values,
                    keep_index, compute, strict
                )
            arg = resolve_dependent_value(arg)
            fn_args.append(arg)
        fn_kwargs = {}
        for k, v in kwargs.items():
            if isinstance(v, dim):
                v = v.apply(
                    dataset, flat, expanded, ranges, all_values,
                    keep_index, compute, strict
                )
            fn_kwargs[k] = resolve_dependent_value(v)
        args = tuple(fn_args[::-1] if op['reverse'] else fn_args)
        kwargs = dict(fn_kwargs)
        return fn, fn_name, args, kwargs, accessor 

def __array_ufunc__(self, *args, **kwargs):
        ufunc = args[0]
        kwargs = {k: v for k, v in kwargs.items() if v is not None}
        return type(self)(self, ufunc, **kwargs) 

def reduce(self, dimensions=None, function=None, **reduce_map):
        """
        Reduces the Raster using functions provided via the
        kwargs, where the keyword is the dimension to be reduced.
        Optionally a label_prefix can be provided to prepend to
        the result Element label.
        """
        function, dims = self._reduce_map(dimensions, function, reduce_map)
        if len(dims) == self.ndims:
            if isinstance(function, np.ufunc):
                return function.reduce(self.data, axis=None)
            else:
                return function(self.data)
        else:
            dimension = dims[0]
            other_dimension = [d for d in self.kdims if d.name != dimension]
            oidx = self.get_dimension_index(other_dimension[0])
            x_vals = self.dimension_values(other_dimension[0].name, False)
            reduced = function(self._zdata, axis=oidx)
            if oidx and hasattr(self, 'bounds'):
                reduced = reduced[::-1]
            data = zip(x_vals, reduced)
            params = dict(dict(self.param.get_param_values(onlychanged=True)),
                          kdims=other_dimension, vdims=self.vdims)
            params.pop('bounds', None)
            params.pop('extents', None)
            return Table(data, **params) 

def collapse_data(cls, data_list, function, kdims=None, **kwargs):
        param.main.param.warning(
            'Raster.collapse_data is deprecated, collapsing '
            'may now be performed through concatenation '
            'and aggregation.')
        if isinstance(function, np.ufunc):
            return function.reduce(data_list)
        else:
            return function(np.dstack(data_list), axis=-1, **kwargs) 

def test_np_ufunc_notimplemented(self, maps_release_only, ufunc):
        ha = maps_release_only.emline_gflux_ha_6564
        nii = maps_release_only.emline_gflux_nii_6585

        with pytest.raises(NotImplementedError) as ee:
            if getattr(getattr(np, ufunc), 'nargs') <= 2:
                getattr(np, ufunc)(ha)

            else:
                getattr(np, ufunc)(nii, ha)

        expected = 'np.{0} is not implemented for Map.'.format(getattr(np, ufunc).__name__)
        assert str(ee.value) == expected 

def matrix_transform(data, fun, *args, **kwargs):
    """Perform a numerical transformation to data

    Parameters
    ----------
    data : array-like, shape=[n_samples, n_features]
        Input data
    fun : callable
        Numerical transformation function, `np.ufunc` or similar.
    args, kwargs : additional arguments, optional
        arguments for `fun`. `data` is always passed as the first argument

    Returns
    -------
    data : array-like, shape=[n_samples, n_features]
        Transformed output data
    """
    if is_sparse_dataframe(data) or is_SparseDataFrame(data):
        data = data.copy()
        for col in data.columns:
            data[col] = fun(data[col], *args, **kwargs)
    elif sparse.issparse(data):
        if isinstance(data, (sparse.lil_matrix, sparse.dok_matrix)):
            data = data.tocsr()
        else:
            # avoid modifying in place
            data = data.copy()
        data.data = fun(data.data, *args, **kwargs)
    else:
        data = fun(data, *args, **kwargs)
    return data 

def _grouped_reduce(x, groups, method, **kwargs):
    """
    Performs a :code:`ufunc` grouped reduce.

    Parameters
    ----------
    x : np.ndarray
        The data to reduce.
    groups : np.ndarray
        The groups the data belongs to. The groups must be
        contiguous.
    method : np.ufunc
        The :code:`ufunc` to use to perform the reduction.
    kwargs : dict
        The kwargs to pass to the :code:`ufunc`'s :code:`reduceat`
        function.

    Returns
    -------
    result : np.ndarray
        The result of the grouped reduce operation.
    inv_idx : np.ndarray
        The index of the first element where each group is found.
    counts : np.ndarray
        The number of elements in each group.
    """
    # Partial credit to @shoyer
    # Ref: https://gist.github.com/shoyer/f538ac78ae904c936844
    inv_idx, counts = _calc_counts_invidx(groups)
    result = method.reduceat(x, inv_idx, **kwargs)
    return result, inv_idx, counts 

def round(self, decimals=0, out=None):
        """
        Evenly round to the given number of decimals.

        See also
        --------
        :obj:`numpy.round` : NumPy equivalent ufunc.
        :obj:`COO.elemwise`: Apply an arbitrary element-wise function to one or two
            arguments.
        """
        if out is not None and not isinstance(out, tuple):
            out = (out,)
        return self.__array_ufunc__(
            np.round, "__call__", self, decimals=decimals, out=out
        )