|jddlmZddlmZddlZddlZddlmZddlm Z m Z ddl m Z m Z mZddlmZddlmZdd lmZerdd lmZdd lmZdd lmZmZgZe jjje jjje jjj e jjj!e jjj"e jjj#gZ$dGdHdZ%dGdHdZ&dGdHdZ' dGdIdZ( dJdKdZ)dGdHd Z*dGdHd!Z+dGdHd"Z,dGdHd#Z-dGdHd$Z.dGdHd%Z/dGdHd&Z0dGdHd'Z1 dLdMd*Z2dGdNd-Z3dGdHd.Z4dGdHd/Z5dGdHd0Z6dGdHd1Z7dGdHd2Z8dGdHd3Z9ed d4gdGdOd7Z:dGdHd8Z; dGdPd=Z< dQdRdFZ=dS)S) annotations) TYPE_CHECKINGN)_C_ops) check_typecheck_variable_and_dtype)convert_np_dtype_to_dtype_corein_dynamic_or_pir_mode)Variable) LayerHelper)param_one_alias)Sequence)Tensor) DTypeLike ShapeLikexrname str | NonereturncZts Jdtj|S)a Calculate elementwise sin of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = sin(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.sin(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-0.90929741, 0.84147102]) >> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.tan(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[2.18503976, 1.55740774]) r)r r sparse_tanrs rtanr XrrcZts Jdtj|S)a Calculate elementwise asin of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = asin(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.asin(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[nan , 1.57079625]) r)r r sparse_asinrs rasinr#{:: " # #F #  a  rperm Sequence[int]c\ts Jdtj||S)a Changes the perm order of ``x`` without changing its data, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = transpose(x, perm) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64. perm (list|tuple): Permute the input according to the data of perm. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A transposed Sparse Tensor with the same data type as ``x``. Examples: .. code-block:: python >>> # doctest: +SKIP('indices overflow') >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> dense_x = paddle.to_tensor([[-2., 0.], [1., 2.]]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.transpose(sparse_x, [1, 0]) >>> out Tensor(shape=[2, 2], dtype=paddle.float32, place=Place(gpu:0), stop_gradient=True, indices=[[0, 0]], values=[[-2., 0.], [ 1., 2.]]) r)r rsparse_transpose)rr%rs r transposer)s=H " # #F #  "1d + ++rFaxisint | Sequence[int] | NonedtypeDTypeLike | Nonekeepdimboolc 0d}|d}t|}trtj||||S|g}n|g}|||d}|r||j|dt |dgddt|d ttttdtfdd}t|}|r|| } n||j } ||d|id | i| | S) a Computes the sum of sparse tensor elements over the given dimension, requiring x to be a SparseCooTensor or SparseCsrTensor. Args: x (Tensor): An N-D Tensor, the data type is bool, float16, float32, float64, int32 or int64. axis (int|list|tuple|None, optional): The dimensions along which the sum is performed. If :attr:`None`, sum all elements of :attr:`x` and return a Tensor with a single element, otherwise must be in the range :math:`[-rank(x), rank(x))`. If :math:`axis[i] < 0`, the dimension to reduce is :math:`rank + axis[i]`. dtype (str|None, optional): The dtype of output Tensor. The default value is None, the dtype of output is the same as input Tensor `x`. keepdim (bool, optional): Whether to reserve the reduced dimension in the output Tensor. The result Tensor will have one fewer dimension than the :attr:`x` unless :attr:`keepdim` is true, default value is False. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: Tensor: Results of summation operation on the specified axis of input Tensor `x`. if `x.dtype='bool'` or `x.dtype='int32'`, it's data type is `'int64'`, otherwise it's data type is the same as `x`. Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([[-2., 0.], [1., 2.]]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out1 = paddle.sparse.sum(sparse_x) >>> out1 Tensor(shape=[1], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[0], values=1.) >>> out2 = paddle.sparse.sum(sparse_x, axis=0) >>> out2 Tensor(shape=[1, 2], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0]], values=[[-1., 2.]]) >>> out3 = paddle.sparse.sum(sparse_x, axis=-1) >>> out3 Tensor(shape=[2], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 1]], values=[-2., 3.]) >>> out4 = paddle.sparse.sum(sparse_x, axis=1, keepdim=True) >>> out4 Tensor(shape=[2, 1], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 1]], values=[[-2.], [ 3.]]) FNT)r*r,r.)in_dtype out_dtyperr/float32float64int16int32int64 sparse_sumr*r,outtypeinputsoutputsattrs)rr rr9updater,rrintlisttupler=r r )create_sparse_variable_for_type_inference append_op) rr*r,r.r dtype_flagr@op_typehelperr;s rsumrJssvJ  *511) D%999 <DD6D'BB  D LLagEBB C C C        &3eT$ZZBL   W%%  BBBOOCCBBgCC #qE3>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.atan(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-1.10714877, 0.78539819]) r)r r sparse_atanrs ratanrM4r$rcZts Jdtj|S)a Calculate elementwise sinh of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = sinh(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.sinh(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-3.62686038, 1.17520118]) r)r r sparse_sinhrs rsinhrPWr$rcZts Jdtj|S)a Calculate elementwise asinh of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = asinh(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.asinh(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-1.44363546, 0.88137358]) r)r r sparse_asinhrs rasinhrSz:: " # #F #  q ! !!rcZts Jdtj|S)a Calculate elementwise atanh of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = atanh(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.atanh(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[nan , inf.]) r)r r sparse_atanhrs ratanhrWrTrcZts Jdtj|S)a Calculate elementwise tanh of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = tanh(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.tanh(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-0.96402758, 0.76159418]) r)r r sparse_tanhrs rtanhrZr$rcZts Jdtj|S)a Calculate elementwise square of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = square(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.square(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[4., 1.]) r)r r sparse_squarers rsquarer]s:: " # #F #   " ""rcZts Jdtj|S)a Calculate elementwise sqrt of SparseTensor, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = sqrt(x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.sqrt(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[nan, 1. ]) r)r r sparse_sqrtrs rsqrtr`r$rcZts Jdtj|S)a Calculate the natural log of (1+x), requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = ln(1+x) Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2, 0, 1], dtype='float32') >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.log1p(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[nan , 0.69314718]) r)r r sparse_log1prs rlog1prc)rTr index_dtype value_dtypecNts Jd|r:t|tjjtjfst |}|r:t|tjjtjfst |}tj|||S)a? cast non-zero-index of SparseTensor to `index_dtype`, non-zero-element of SparseTensor to `value_dtype` , requiring x to be a SparseCooTensor or SparseCsrTensor. Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64. index_dtype (np.dtype|str, optional): Data type of the index of SparseCooTensor, or crows/cols of SparseCsrTensor. Can be uint8, int8, int16, int32, int64. value_dtype (np.dtype|str, optional): Data type of the value of SparseCooTensor, SparseCsrTensor. Can be bool, float16, float32, float64, int8, int32, int64, uint8. name (str|core.VarDesc.VarType|core.DataType|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2, 0, 1]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.cast(sparse_x, 'int32', 'float64') >>> out Tensor(shape=[3], dtype=paddle.float64, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-2., 1.]) r) r isinstancer VarDescVarTypeDataTyperr sparse_cast)rrdrers rcastrlLsF " # #F #>:dl*DM:>1== >:dl*DM:>1==  ak : ::rfactorfloatcvts Jdtj|t|S)a Calculate elementwise pow of x, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = x^{factor} Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64. factor (float|int): factor of pow. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2, 0, 3], dtype='float32') >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.pow(sparse_x, 2) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[4., 9.]) r)r r sparse_powrn)rrmrs rpowrq}sB< " # #F #  Qf . ..rc`ts Jdtj|dddS)a Calculate elementwise negative of x, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = -x Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2, 0, 3], dtype='float32') >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.neg(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[ 2., -3.]) rgT)r r sparse_scalers rnegrus@: " # #F #  q$T 2 22rcZts Jdtj|S)a Calculate elementwise absolute value of x, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = |x| Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2, 0, 3], dtype='float32') >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.abs(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[2., 3.]) r)r r sparse_absrs rabsrxrrcZts Jdtj|S)a the coalesced operator include sorted and merge, after coalesced, the indices of x is sorted and unique. Parameters: x (Tensor): the input SparseCooTensor. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: Tensor: return the SparseCooTensor after coalesced. Examples: .. code-block:: python >>> import paddle >>> indices = [[0, 0, 1], [1, 1, 2]] >>> values = [1.0, 2.0, 3.0] >>> sp_x = paddle.sparse.sparse_coo_tensor(indices, values) >>> sp_x = paddle.sparse.coalesce(sp_x) >>> print(sp_x.indices()) Tensor(shape=[2, 2], dtype=int64, place=Place(cpu), stop_gradient=True, [[0, 1], [1, 2]]) >>> print(sp_x.values()) Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, [3., 3.]) r)r rsparse_coalescers rcoalescer{s:: " # #F #  !! $ $$rcts Jd|jtvr*tj|dt jjj}tj |dtj z ddS)a Convert each of the elements of input x from radian to degree, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: rad2deg(x) = 180/ \pi * x Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, int32, int64. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([3.142, 0., -3.142]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.rad2deg(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[ 180.02334595, -180.02334595]) rNf@rsT r r, _int_dtype_rrkr rhriFP32rtnppirs rrad2degr so< " # #F # w+  q$ (<(A B B  q%"%-d ; ;;rcts Jd|jtvr*tj|dt jjj}tj |tj dz ddS)a Convert each of the elements of input x from degree to radian, requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: deg2rad(x) = \pi * x / 180 Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, int32, int64. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-180, 0, 180]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.deg2rad(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-3.14159274, 3.14159274]) rNr}rsTr~rs rdeg2radr0so< " # #F # w+  q$ (<(A B B  q"%%-d ; ;;rcZts Jdtj|S)a Calculate elementwise `exp(x)-1` , requiring x to be a SparseCooTensor or SparseCsrTensor. .. math:: out = exp(x) - 1 Parameters: x (Tensor): The input Sparse Tensor with data type float32, float64, complex64, complex128. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same data type and shape as ``x`` . Examples: .. code-block:: python >>> import paddle >>> dense_x = paddle.to_tensor([-2., 0., 1.]) >>> sparse_x = dense_x.to_sparse_coo(1) >>> out = paddle.sparse.expm1(sparse_x) >>> out Tensor(shape=[3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 2]], values=[-0.86466473, 1.71828187]) r)r r sparse_expm1rs rexpm1rVrTrinputshapercHtrtj||St|dgddt |dt t fdd|i}d|i}td}||j }| d|d|i||S)a Changes the shape of ``x`` without changing its value, requiring x to be a SparseCooTensor or SparseCsrTensor. Currently this function can only reshape the sparse dims of ``x`` , but ``shape`` argument must be specified as the shape of the reshaped tensor. Note that if x is a SparseCsrTensor, then len(shape) must be 2 or 3. There are some tricks when specifying the target shape. - 1. -1 means the value of this dimension is inferred from the total element number of x and remaining dimensions. Thus one and only one dimension can be set -1. - 2. 0 means the actual dimension value is going to be copied from the corresponding dimension of x. The indices of 0 in the target shape can not exceed the rank of x. Here are some examples to explain it. - 1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape is [6, 8], the reshape operator will transform x into a 2-D tensor with shape [6, 8] and leaving x's data unchanged. - 2. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape is [2, 3, -1, 2], the reshape operator will transform x into a 4-D tensor with shape [2, 3, 4, 2] and leaving x's data unchanged. In this case, one dimension of the target shape is set to -1, the value of this dimension is inferred from the total element number of x and remaining dimensions. - 3. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape is [-1, 0, 3, 2], the reshape operator will transform x into a 4-D tensor with shape [2, 4, 3, 2] and leaving x's data unchanged. In this case, besides -1, 0 means the actual dimension value is going to be copied from the corresponding dimension of x. .. note:: Alias Support: The parameter name ``input`` can be used as an alias for ``x``. For example, ``reshape(input=tensor_x, ...)`` is equivalent to ``reshape(x=tensor_x, ...)``. Args: x (Tensor): The input sparse tensor with data type ``float32``, ``float64``, ``int32``, ``int64`` or ``bool``. shape (list|tuple): Define the target shape. At most one dimension of the target shape can be -1. The data type is ``int32``. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: Tensor: A reshaped Tensor with the same data type as ``x``. Examples: .. code-block:: pycon >>> import paddle >>> x_shape = [6, 2, 3] >>> new_shape = [1, 0, 2, -1, 3] >>> format = "coo" >>> dense_x = paddle.randint(-100, 100, x_shape) * paddle.randint(0, 2, x_shape) >>> if format == "coo": ... sp_x = dense_x.to_sparse_coo(len(x_shape)) >>> else: ... sp_x = dense_x.to_sparse_csr() >>> sp_out = paddle.sparse.reshape(sp_x, new_shape) >>> print(sp_out.shape) paddle.Size([1, 2, 2, 3, 3]) r)float16r4r5r6r7r8r/uint16reshapersparse_reshaper;r<) r rrrrrCrDr rEr,rF)rrrr>r@rIr;s rrryst$Q...         5'D%=)<<<q% -..>>qwGG!CL     rctrtj|Sd}t|}||j}||d|id|ii|S)a Return whether every element of input tensor is `NaN` or not, requiring x to be a SparseCooTensor or SparseCsrTensor. Args: x (Tensor): The input tensor (SparseCooTensor or SparseCsrTensor), it's data type should be float16, float32, float64, int32, int64. name (str|None, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: A Sparse Tensor with the same shape as ``x``, the bool result which shows every element of `x` whether it is `NaN` or not. Examples: .. code-block:: python >>> import paddle >>> import numpy as np >>> format = "coo" >>> np_x = np.asarray([[[0., 0], [1., 2.]], [[0., 0], [3., float('nan')]]]) >>> dense_x = paddle.to_tensor(np_x) >>> if format == "coo": ... sparse_x = dense_x.to_sparse_coo(len(np_x.shape)) >>> else: ... sparse_x = dense_x.to_sparse_csr() ... >>> sparse_out = paddle.sparse.isnan(sparse_x) >>> print(sparse_out) Tensor(shape=[2, 2, 2], dtype=paddle.bool, place=Place(gpu:0), stop_gradient=True, indices=[[0, 0, 1, 1], [1, 1, 1, 1], [0, 1, 0, 1]], values=[False, False, False, True ]) sparse_isnanrr;r<)r rrr rEr,rF)rrrHrIr;s risnanrsH "1%%% W%%>>qwGG#qE3>> import paddle >>> import numpy as np >>> format = 'coo' >>> np_x = np.asarray([[4, 0, 7, 0], [0, 0, 5, 0], [-4, 2, 0, 0]]) >>> dense_x = paddle.to_tensor(np_x) >>> if format == 'coo': ... sp_x = dense_x.to_sparse_coo(len(np_x.shape)) >>> else: ... sp_x = dense_x.to_sparse_csr() ... >>> axes = [0, 1] >>> starts = [1, 0] >>> ends = [3, -2] >>> sp_out = paddle.sparse.slice(sp_x, axes, starts, ends) >>> # sp_out is x[1:3, 0:-2] >>> print(sp_out) Tensor(shape=[2, 2], dtype=paddle.int64, place=Place(cpu), stop_gradient=True, indices=[[1, 1], [0, 1]], values=[-4, 2]) )rrrrr3 sparse_slicerrrr:r;r<) r rrrrrCrDr rEr,rF) rrrrrr@rHrIr;s rslicersz"1dFD999>>       4$???68dE]NCCC4$??? W%%>>QW>MM#qE3>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> format = "coo" >>> paddle.seed(2023) >>> dense_x = paddle.randn((5, 5), dtype='float64') >>> if format == "coo": ... sparse_x = dense_x.to_sparse_coo(len(dense_x.shape)) >>> else: ... sparse_x = dense_x.to_sparse_csr() >>> print("sparse.pca_lowrank API only support CUDA 11.x") >>> # U, S, V = None, None, None >>> # use code blow when your device CUDA version >= 11.0 >>> U, S, V = paddle.sparse.pca_lowrank(sparse_x) >>> print(U) Tensor(shape=[5, 5], dtype=float64, place=Place(gpu:0), stop_gradient=True, [[-0.31412600, 0.44814876, 0.18390454, -0.19967630, -0.79170452], [-0.31412600, 0.44814876, 0.18390454, -0.58579808, 0.56877700], [-0.31412600, 0.44814876, 0.18390454, 0.78547437, 0.22292751], [-0.38082462, 0.10982129, -0.91810233, 0.00000000, 0.00000000], [ 0.74762770, 0.62082796, -0.23585052, 0.00000000, -0.00000000]]) >>> print(S) Tensor(shape=[5], dtype=float64, place=Place(gpu:0), stop_gradient=True, [1.56031096, 1.12956227, 0.27922715, 0.00000000, 0.00000000]) >>> print(V) Tensor(shape=[5, 5], dtype=float64, place=Place(gpu:0), stop_gradient=True, [[ 0.88568469, -0.29081908, 0.06163676, 0.19597228, -0.29796422], [-0.26169364, -0.27616183, 0.43148760, -0.42522796, -0.69874939], [ 0.28587685, 0.30695344, -0.47790836, -0.76982533, -0.05501437], [-0.23958121, -0.62770647, -0.71141770, 0.11463224, -0.17125926], [ 0.08918713, -0.59238761, 0.27478686, -0.41833534, 0.62498824]]) cv|j}|tjtjtjfvr|StjSN)r,paddlerr4r5)rr,s rget_floating_dtypez'pca_lowrank..get_floating_dtypes. V^V^V^D D DL~rcV|r|S|Sr) is_complexconj)rs r conjugatezpca_lowrank..conjugates$ <<>> 6688Orc.|j}ttdt|}g|dd|d|d}|r t j||St j||S)Nr)rrCrangelen is_sparsersparser))rrr%s rr)zpca_lowrank..transposesE!SZZ(())/crc/DH/d2h/ ;;== 4=**1d33 34(((rc,|Sr)rrr)s r transjugatez pca_lowrank..transjugatesy1&&&rrNc|dn|}|jdd\}}tjj}tj||f|j}|} |} ||tj||d} t|D]`} |tj| | d} |tj|| d} an|} |tj||tj||z d} t|D]} |tj| | tj| | z d} |tj|| tj|| z d} | S)Nrrr:r) rrlinalgqrrandnr,rmatmulr)rrrMmnrRA_tA_HQiM_Hrrr)s rget_approximate_basisz*pca_lowrank..get_approximate_basiss]wrss|1 ]  L!Qqw / / /illinn 96=''1--..q1A5\\ 6 6Bv}++C3344Q7Bv}++Aq112215 6+a..C6=''1-- a0C0CCDDQGA5\\ L LBv}++C33fmC6K6KKLLQOBv}++Aq11FM!Q4G4GGHHKrcD|dn|}|jdd\}}|d}n |}|}||ks||kr)||||}|} |!tj|| } n6tj|| tj|| z } | jd|ksJ| j|f| jd|ksJ| j|f| jd| jdksJ| jtj| d\} } } | }||}n2||||}|} |!tj|| }n6tj|| tj|| z }|} | jd|ksJ| j|f| jd|ksJ| j|f| jd| jdksJ| jtj| d\} } } | }|| } | | |fS)NrrrrrF) full_matrices)rrrrrsvd)rrrrrrM_trrQ_cB_tUSVhVBrrrr)s r svd_lowrankz pca_lowrank..svd_lowranksAAwrss|1 9CC)A,,Cill q55AEE%%c1ESAAAA)A,,Cym**1c22m**1c22V]1c5J5JJ9R=A%%% 1~%%%9R=A%%% 1~%%%9R=CIbM11139111}((E(BBHAq" BA AA%%a%1===A)A,,CyM((c22M((c22V]35L5LL)A,,C9R=A%%% 1~%%%9R=A%%% 1~%%%9R=CIbM11139111}((E(BBHAq" BA A!QwrzInput must be tensor, but got z#Input must be sparse, but got denseFalse.r z-sparse.pca_lowrank API only support CUDA 11.xrzq(=z>) must be non-negative integer and not greater than min(m, n)=zniter(=z) must be non-negative integerrz,input is expected to be 2-dimensional tensor)r*)r,placer:)rN)rrN)r is_tensor ValueErrorr=rversioncudarBsplitrminrrrJvaluessparse_coo_tensorindicesr,rzerosonesrto_dense)rrrrrrr cuda_versionrrr,s_sums_valccolumn_indicesrC_t ones_m1_trrrrr)s @@@@r pca_lowrankr_sV  )))''''''.%%%%%%%%%N  A  EC$q''CCDDD ;;==@>???>&&((L 7 " " |!!#&&q) * *R / /HIII WRSS\FQy 1aLL1ffc!Qii ;! ; ;/21ayy ; ;    QJJH5HHHIII  q ! !E 6{1au5555 17||qGHHH M  ab  ) )E LLNNQ E '' ek (  AYY[[^NlAs>223>;OPPPGGAJ - ) )aV5 *  C 1agcrcl1A1q1???I &-  ::;;A ;q!5A . . ..rr)rrrrrr)rrr%r&rrrr)NNFN) rrr*r+r,r-r.r/rrrr)NNN) rrrdr-rer-rrrr)rrrmrnrrrr)rrrrrrrr) rrrrrrrrrrrr)NTrN) rrrrrr/rrBrrrr)> __future__rtypingrnumpyrrrpaddle.base.data_feederrrpaddle.base.frameworkrr r paddle.common_ops_importr paddle.frameworkr paddle.utils.decorator_utilsr collections.abcrrpaddle._typingrr__all__rhriUINT8INT8INT16INT32INT64BOOLrrr r#r)rJrMrPrSrWrZr]r`rcrlrqrurxr{rrrrrrrrrrrsk#"""""  HHHHHHHH .-----((((((4((((((33333333  LLLLLL        F      F ! ! ! ! !H8<',',',',',X(," iiiiiX ! ! ! ! !F ! ! ! ! !F " " " " "F " " " " "F ! ! ! ! !F # # # # #F ! ! ! ! !F " " " " "J%)$( .;.;.;.;.;b!/!/!/!/!/H 3 3 3 3 3F      F % % % % %F#<#<#<#<#