How to use the cupy.empty function in cupy

def test_invalid_argument_normal_double(self):
        out = cupy.empty((1,), dtype=numpy.float64)
        with self.assertRaises(ValueError):
            curand.generateNormalDouble(
                self.generator, out.data.ptr, 1, 0.0, 1.0)

def asfortranarray(a, dtype=None):
    """Return an array laid out in Fortran order in memory.

    Args:
        a (~cupy.ndarray): The input array.
        dtype (str or dtype object, optional): By default, the data-type is
            inferred from the input data.

    Returns:
        ~cupy.ndarray: The input `a` in Fortran, or column-major, order.

    .. seealso:: :func:`numpy.asfortranarray`

    """
    ret = cupy.empty(a.shape[::-1], a.dtype if dtype is None else dtype).T
    ret[...] = a
    return ret

def create_dropout_states(handle):
    state_size = cudnn.dropoutGetStatesSize(handle)
    return cupy.empty((state_size,), dtype='b')

def _generate_normal(self, func, size, dtype, *args):
        # curand functions below don't support odd size.
        # * curand.generateNormal
        # * curand.generateNormalDouble
        # * curand.generateLogNormal
        # * curand.generateLogNormalDouble
        size = core.get_size(size)
        element_size = six.moves.reduce(operator.mul, size, 1)
        if element_size % 2 == 0:
            out = cupy.empty(size, dtype=dtype)
            func(self._generator, out.data.ptr, out.size, *args)
            return out
        else:
            out = cupy.empty((element_size + 1,), dtype=dtype)
            func(self._generator, out.data.ptr, out.size, *args)
            return out[:element_size].reshape(size)

def forward(self, xs):
        N, T = xs.shape
        V, D = self.W.shape

        out = cp.empty((N, T, D), dtype='f')
        self.layers = []

        for t in range(T):
            layer = Embedding(self.W)
            out[:, t, :] = layer.forward(xs[:, t])
            self.layers.append(layer)

        return out

Otherwise, it returns only the array.
        dtype: Data type specifier. It is inferred from the start and stop
            arguments by default.

    Returns:
        cupy.ndarray: The 1-D array of ranged values.

    """
    if num < 0:
        raise ValueError('linspace with num<0 is not supported')

    if dtype is None:
        # In actual implementation, only float is used
        dtype = float

    ret = cupy.empty((num,), dtype=dtype)
    if num == 0:
        step = float('nan')
    elif num == 1:
        ret.fill(start)
        step = float('nan')
    else:
        div = (num - 1) if endpoint else num
        step = float(stop - start) / div
        stop = float(stop)

        if step == 0.0:
            # for underflow
            _linspace_ufunc_underflow(start, stop - start, div, ret,
                                      casting='unsafe')
        else:
            _linspace_ufunc(start, step, ret, casting='unsafe')

if log:
        log = {'err': []}

    # we assume that no distances are null except those of the diagonal of
    # distances
    if nbb:
        u = np.ones((Nini, nbb)) / Nini
        v = np.ones((Nfin, nbb)) / Nfin
    else:
        u = np.ones(Nini) / Nini
        v = np.ones(Nfin) / Nfin

    # print(reg)

    # Next 3 lines equivalent to K= np.exp(-M/reg), but faster to compute
    K = np.empty(M.shape, dtype=M.dtype)
    np.divide(M, -reg, out=K)
    np.exp(K, out=K)

    # print(np.min(K))
    tmp2 = np.empty(b.shape, dtype=M.dtype)

    Kp = (1 / a).reshape(-1, 1) * K
    cpt = 0
    err = 1
    while (err > stopThr and cpt < numItermax):
        uprev = u
        vprev = v

        KtransposeU = np.dot(K.T, u)
        v = np.divide(b, KtransposeU)
        u = 1. / np.dot(Kp, v)

return cupy.empty(shape_out, dtype=c.dtype)
    c_shape = []
    ret_shape = []
    for dim_in, nrep in zip(c.shape, tup):
        if nrep == 1:
            c_shape.append(dim_in)
            ret_shape.append(dim_in)
        elif dim_in == 1:
            c_shape.append(dim_in)
            ret_shape.append(nrep)
        else:
            c_shape.append(1)
            c_shape.append(dim_in)
            ret_shape.append(nrep)
            ret_shape.append(dim_in)
    ret = cupy.empty(ret_shape, dtype=c.dtype)
    if ret.size:
        ret[...] = c.reshape(c_shape)
    return ret.reshape(shape_out)

def backward(self, dhs):
        Wx, Wh, b = self.params
        N, T, H = dhs.shape
        D, H = Wx.shape

        dxs = cp.empty((N, T, D), dtype='f')
        dh = 0
        grads = [0, 0, 0]
        for t in reversed(range(T)):
            layer = self.layers[t]
            dx, dh = layer.backward(dhs[:, t, :] + dh)
            dxs[:, t, :] = dx

            for i, grad in enumerate(layer.grads):
                grads[i] += grad

        for i, grad in enumerate(grads):
            self.grads[i][...] = grad
        self.dh = dh

        return dxs

.. seealso:: :func:`numpy.arange`

    """
    if dtype is None:
        if any(numpy.dtype(type(val)).kind == 'f'
               for val in (start, stop, step)):
            dtype = float
        else:
            dtype = int

    if stop is None:
        stop = start
        start = 0
    size = int(numpy.ceil((stop - start) / step))
    if size <= 0:
        return cupy.empty((0,), dtype=dtype)

    ret = cupy.empty((size,), dtype=dtype)
    typ = numpy.dtype(dtype).type
    _arange_ufunc(typ(start), typ(step), ret, dtype=dtype)
    return ret