repo_id
stringlengths
21
96
file_path
stringlengths
31
155
content
stringlengths
1
92.9M
__index_level_0__
int64
0
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/karate_asymmetric.yaml
name: karate-asymmetric file_type: .csv description: This is an undirected, asymmetric variant of the Karate dataset. The original dataset, which this is based on, was created by Wayne Zachary in 1977. author: Nvidia refs: W. W. Zachary, An information flow model for conflict and fission in small groups, Journal of Anthropological Research 33, 452-473 (1977). delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: false is_directed: false is_multigraph: false is_symmetric: true number_of_edges: 78 number_of_nodes: 34 url: https://data.rapids.ai/cugraph/datasets/karate-asymmetric.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/soc-livejournal1.yaml
name: soc-LiveJournal1 file_type: .csv description: A graph of the LiveJournal social network. author: L. Backstrom, D. Huttenlocher, J. Kleinberg, X. Lan refs: L. Backstrom, D. Huttenlocher, J. Kleinberg, X. Lan. Group Formation in Large Social Networks Membership, Growth, and Evolution. KDD, 2006. delim: " " header: None col_names: - src - dst col_types: - int32 - int32 has_loop: true is_directed: true is_multigraph: false is_symmetric: false number_of_edges: 68993773 number_of_nodes: 4847571 url: https://data.rapids.ai/cugraph/datasets/soc-LiveJournal1.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/karate.yaml
name: karate file_type: .csv description: The graph "karate" contains the network of friendships between the 34 members of a karate club at a US university, as described by Wayne Zachary in 1977. author: Zachary W. refs: W. W. Zachary, An information flow model for conflict and fission in small groups, Journal of Anthropological Research 33, 452-473 (1977). delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: false is_directed: true is_multigraph: false is_symmetric: true number_of_edges: 156 number_of_nodes: 34 url: https://data.rapids.ai/cugraph/datasets/karate.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/toy_graph.yaml
name: toy_graph file_type: .csv description: The `toy_graph` dataset was created by Nvidia for testing and demonstration purposes, and consists of a small (6 nodes) directed graph. author: null refs: null delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: false is_directed: true is_multigraph: false is_symmetric: true number_of_edges: 16 number_of_nodes: 6 url: https://data.rapids.ai/cugraph/datasets/toy_graph.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/netscience.yaml
name: netscience file_type: .csv description: The graph netscience contains a coauthorship network of scientists working on network theory and experiment, as compiled by M. Newman in May 2006. author: Newman, Mark E.J. refs: Finding community structure in networks using the eigenvectors of matrices. delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: false is_directed: true is_multigraph: false is_symmetric: true number_of_edges: 5484 number_of_nodes: 1461 url: https://data.rapids.ai/cugraph/datasets/netscience.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/europe_osm.yaml
name: europe_osm file_type: .csv description: A graph of OpenStreetMap data for Europe. author: M. Kobitzsh / Geofabrik GmbH refs: Rossi, Ryan. Ahmed, Nesreen. The Network Data Respoistory with Interactive Graph Analytics and Visualization. delim: " " header: None col_names: - src - dst col_types: - int32 - int32 has_loop: false is_directed: false is_multigraph: false is_symmetric: true number_of_edges: 54054660 number_of_nodes: 50912018 url: https://data.rapids.ai/cugraph/datasets/europe_osm.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/toy_graph_undirected.yaml
name: toy_graph_undirected file_type: .csv description: The `toy_graph_undirected` dataset was created by Nvidia for testing and demonstration purposes, and consists of a small (6 nodes) undirected graph. author: Nvidia refs: null delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: false is_directed: false is_multigraph: false is_symmetric: true number_of_edges: 8 number_of_nodes: 6 url: https://data.rapids.ai/cugraph/datasets/toy_graph_undirected.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/dining_prefs.yaml
name: dining_prefs file_type: .csv description: Classic social networking dataset describes dining preferences for a dormitory in New York state. author: J.L. Moreno refs: J. L. Moreno (1960). The Sociometry Reader. The Free Press, Glencoe, Illinois, pg.35 delim: " " header: None col_names: - src - dst - wgt col_types: - string - string - int has_loop: false is_directed: false is_multigraph: false is_symmetric: true number_of_edges: 42 number_of_nodes: 26 url: https://data.rapids.ai/cugraph/datasets/dining_prefs.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/email_Eu_core.yaml
name: email-Eu-core file_type: .csv description: The network was generated using anonymized email data from a large European research institution. There is an edge (u, v) in the network if person u sent person v at least one email. The e-mails only represent communication between institution members (the core), and the dataset does not contain incoming messages from or outgoing messages to the rest of the world. author: Jure Leskovec refs: - Hao Yin, Austin R. Benson, Jure Leskovec, and David F. Gleich. 'Local Higher-order Graph Clustering.' In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2017. - J. Leskovec, J. Kleinberg and C. Faloutsos. Graph Evolution. Densification and Shrinking Diameters. ACM Transactions on Knowledge Discovery from Data (ACM TKDD), 1(1), 2007. delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: true is_directed: true is_multigraph: false is_symmetric: false number_of_edges: 25571 number_of_nodes: 1005 url: https://data.rapids.ai/cugraph/datasets/email-Eu-core.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/__init__.py
# Copyright (c) 2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/soc-twitter-2010.yaml
name: soc-twitter-2010 file_type: .csv description: A network of follower relationships from a snapshot of Twitter in 2010, where an edge from i to j indicates that j is a follower of i. author: H. Kwak, C. Lee, H. Park, S. Moon refs: J. Yang, J. Leskovec. Temporal Variation in Online Media. ACM Intl. Conf. on Web Search and Data Mining (WSDM '11), 2011. delim: " " header: None col_names: - src - dst col_types: - int32 - int32 has_loop: false is_directed: false is_multigraph: false is_symmetric: false number_of_edges: 530051354 number_of_nodes: 21297772 url: https://data.rapids.ai/cugraph/datasets/soc-twitter-2010.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/karate_disjoint.yaml
name: karate-disjoint file_type: .csv description: This is disjoint variant of the Karate dataset. The original dataset, which this is based on, was created by Wayne Zachary in 1977. author: Nvidia refs: W. W. Zachary, An information flow model for conflict and fission in small groups, Journal of Anthropological Research 33, 452-473 (1977). delim: " " header: None col_names: - src - dst - wgt col_types: - int32 - int32 - float32 has_loop: false is_directed: true is_multigraph: false is_symmetric: true number_of_edges: 312 number_of_nodes: 68 url: https://data.rapids.ai/cugraph/datasets/karate-disjoint.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets
rapidsai_public_repos/cugraph/python/cugraph/cugraph/datasets/metadata/cit-patents.yaml
name: cit-Patents file_type: .csv description: A citation graph that includes all citations made by patents granted between 1975 and 1999, totaling 16,522,438 citations. author: NBER refs: J. Leskovec, J. Kleinberg and C. Faloutsos. Graphs over Time Densification Laws, Shrinking Diameters and Possible Explanations. ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), 2005. delim: " " header: None col_names: - src - dst col_types: - int32 - int32 has_loop: true is_directed: true is_multigraph: false is_symmetric: false number_of_edges: 16518948 number_of_nodes: 3774768 url: https://data.rapids.ai/cugraph/datasets/cit-Patents.csv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/components/CMakeLists.txt
# ============================================================================= # Copyright (c) 2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except # in compliance with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software distributed under the License # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express # or implied. See the License for the specific language governing permissions and limitations under # the License. # ============================================================================= set(cython_sources connectivity_wrapper.pyx) set(linked_libraries cugraph::cugraph) rapids_cython_create_modules( CXX SOURCE_FILES "${cython_sources}" LINKED_LIBRARIES "${linked_libraries}" MODULE_PREFIX components_ ASSOCIATED_TARGETS cugraph )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/components/connectivity_wrapper.pyx
# Copyright (c) 2019-2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # cython: profile=False # distutils: language = c++ # cython: embedsignature = True # cython: language_level = 3 from cugraph.components.connectivity cimport * from cugraph.structure.graph_primtypes cimport * from cugraph.structure.graph_utilities cimport * from cugraph.structure import utils_wrapper from cugraph.structure import graph_primtypes_wrapper from libc.stdint cimport uintptr_t from cugraph.structure.symmetrize import symmetrize from cugraph.structure.graph_classes import Graph as type_Graph import cudf import numpy as np def strongly_connected_components(input_graph): """ Call connected_components """ if not input_graph.adjlist: input_graph.view_adj_list() [offsets, indices] = graph_primtypes_wrapper.datatype_cast([input_graph.adjlist.offsets, input_graph.adjlist.indices], [np.int32]) num_verts = input_graph.number_of_vertices() num_edges = input_graph.number_of_edges(directed_edges=True) df = cudf.DataFrame() df['vertex'] = cudf.Series(np.zeros(num_verts, dtype=np.int32)) df['labels'] = cudf.Series(np.zeros(num_verts, dtype=np.int32)) cdef uintptr_t c_offsets = offsets.__cuda_array_interface__['data'][0] cdef uintptr_t c_indices = indices.__cuda_array_interface__['data'][0] cdef uintptr_t c_identifier = df['vertex'].__cuda_array_interface__['data'][0]; cdef uintptr_t c_labels_val = df['labels'].__cuda_array_interface__['data'][0]; cdef GraphCSRView[int,int,float] g g = GraphCSRView[int,int,float](<int*>c_offsets, <int*>c_indices, <float*>NULL, num_verts, num_edges) cdef cugraph_cc_t connect_type=CUGRAPH_STRONG connected_components(g, <cugraph_cc_t>connect_type, <int *>c_labels_val) g.get_vertex_identifiers(<int*>c_identifier) return df
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/components/connectivity.pxd
# Copyright (c) 2019-2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # cython: profile=False # distutils: language = c++ # cython: embedsignature = True # cython: language_level = 3 from cugraph.structure.graph_primtypes cimport * from cugraph.structure.graph_utilities cimport * cdef extern from "cugraph/algorithms.hpp" namespace "cugraph": ctypedef enum cugraph_cc_t: CUGRAPH_STRONG "cugraph::cugraph_cc_t::CUGRAPH_STRONG" cdef void connected_components[VT,ET,WT]( const GraphCSRView[VT,ET,WT] &graph, cugraph_cc_t connect_type, VT *labels) except +
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/components/connectivity.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.utilities import ( df_score_to_dictionary, ensure_cugraph_obj, is_matrix_type, is_cp_matrix_type, is_nx_graph_type, cupy_package as cp, ) from cugraph.structure import Graph from cugraph.components import connectivity_wrapper import cudf from pylibcugraph import weakly_connected_components as pylibcugraph_wcc from pylibcugraph import ResourceHandle def _ensure_args(api_name, G, directed, connection, return_labels): """ Ensures the args passed in are usable for the API api_name and returns the args with proper defaults if not specified, or raises TypeError or ValueError if incorrectly specified. """ G_type = type(G) # Check for Graph-type inputs and set defaults if unset if (G_type in [Graph]) or is_nx_graph_type(G_type): exc_value = "'%s' cannot be specified for a Graph-type input" if directed is not None: raise TypeError(exc_value % "directed") if return_labels is not None: raise TypeError(exc_value % "return_labels") directed = True return_labels = True # Check for non-Graph-type inputs and set defaults if unset else: directed = True if (directed is None) else directed return_labels = True if (return_labels is None) else return_labels # Handle connection type, based on API being called if api_name == "strongly_connected_components": if (connection is not None) and (connection != "strong"): raise TypeError("'connection' must be 'strong' for " f"{api_name}()") connection = "strong" elif api_name == "weakly_connected_components": if (connection is not None) and (connection != "weak"): raise TypeError("'connection' must be 'weak' for " f"{api_name}()") connection = "weak" else: raise RuntimeError("invalid API name specified (internal): " f"{api_name}") return (directed, connection, return_labels) def _convert_df_to_output_type(df, input_type, return_labels): """ Given a cudf.DataFrame df, convert it to a new type appropriate for the graph algos in this module, based on input_type. return_labels is only used for return values from cupy/scipy input types. """ if input_type in [Graph]: return df elif is_nx_graph_type(input_type): return df_score_to_dictionary(df, "labels", "vertex") elif is_matrix_type(input_type): # Convert DF of 2 columns (labels, vertices) to the SciPy-style return # value: # n_components: int # The number of connected components (number of unique labels). # labels: ndarray # The length-N array of labels of the connected components. n_components = df["labels"].nunique() sorted_df = df.sort_values("vertex") if return_labels: if is_cp_matrix_type(input_type): labels = cp.from_dlpack(sorted_df["labels"].to_dlpack()) else: labels = sorted_df["labels"].to_numpy() return (n_components, labels) else: return n_components else: raise TypeError(f"input type {input_type} is not a supported type.") def weakly_connected_components(G, directed=None, connection=None, return_labels=None): """ Generate the Weakly Connected Components and attach a component label to each vertex. Parameters ---------- G : cugraph.Graph, networkx.Graph, CuPy or SciPy sparse matrix Graph or matrix object, which should contain the connectivity information (edge weights are not used for this algorithm). If using a graph object, the graph must be undirected where an undirected edge is represented by a directed edge in both directions. The adjacency list will be computed if not already present. The number of vertices should fit into a 32b int. directed : bool, optional (default=None) NOTE For non-Graph-type (eg. sparse matrix) values of G only. Raises TypeError if used with a Graph object. If True, then convert the input matrix to a Graph(directed=True) and only move from point i to point j along paths csgraph[i, j]. If False, then find the shortest path on an undirected graph: the algorithm can progress from point i to j along csgraph[i, j] or csgraph[j, i]. connection : str, optional (default=None) Added for SciPy compatibility, can only be specified for non-Graph-type (eg. sparse matrix) values of G only (raises TypeError if used with a Graph object), and can only be set to "weak" for this API. return_labels : bool, optional (default=True) NOTE For non-Graph-type (eg. sparse matrix) values of G only. Raises TypeError if used with a Graph object. If True, then return the labels for each of the connected components. Returns ------- Return value type is based on the input type. If G is a cugraph.Graph, returns: cudf.DataFrame GPU data frame containing two cudf.Series of size V: the vertex identifiers and the corresponding component identifier. df['vertex'] Contains the vertex identifier df['labels'] The component identifier If G is a networkx.Graph, returns: python dictionary, where keys are vertices and values are the component identifiers. If G is a CuPy or SciPy matrix, returns: CuPy ndarray (if CuPy matrix input) or Numpy ndarray (if SciPy matrix input) of shape (<num vertices>, 2), where column 0 contains component identifiers and column 1 contains vertices. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.weakly_connected_components(G) """ (directed, connection, return_labels) = _ensure_args( "weakly_connected_components", G, directed, connection, return_labels ) # FIXME: allow nx_weight_attr to be specified (G, input_type) = ensure_cugraph_obj( G, nx_weight_attr="weight", matrix_graph_type=Graph(directed=directed) ) if G.is_directed(): raise ValueError("input graph must be undirected") vertex, labels = pylibcugraph_wcc( resource_handle=ResourceHandle(), graph=G._plc_graph, offsets=None, indices=None, weights=None, labels=None, do_expensive_check=False, ) df = cudf.DataFrame() df["vertex"] = vertex df["labels"] = labels if G.renumbered: df = G.unrenumber(df, "vertex") return _convert_df_to_output_type(df, input_type, return_labels) def strongly_connected_components( G, directed=None, connection=None, return_labels=None ): """ Generate the Strongly Connected Components and attach a component label to each vertex. Parameters ---------- G : cugraph.Graph, networkx.Graph, CuPy or SciPy sparse matrix Graph or matrix object, which should contain the connectivity information (edge weights are not used for this algorithm). If using a graph object, the graph can be either directed or undirected where an undirected edge is represented by a directed edge in both directions. The adjacency list will be computed if not already present. The number of vertices should fit into a 32b int. directed : bool, optional (default=True) NOTE For non-Graph-type (eg. sparse matrix) values of G only. Raises TypeError if used with a Graph object. If True, then convert the input matrix to a Graph(directed=True) and only move from point i to point j along paths csgraph[i, j]. If False, then find the shortest path on an undirected graph: the algorithm can progress from point i to j along csgraph[i, j] or csgraph[j, i]. connection : str, optional (default=None) Added for SciPy compatibility, can only be specified for non-Graph-type (eg. sparse matrix) values of G only (raises TypeError if used with a Graph object), and can only be set to "strong" for this API. return_labels : bool, optional (default=True) NOTE For non-Graph-type (eg. sparse matrix) values of G only. Raises TypeError if used with a Graph object. If True, then return the labels for each of the connected components. Returns ------- Return value type is based on the input type. If G is a cugraph.Graph, returns: cudf.DataFrame GPU data frame containing two cudf.Series of size V: the vertex identifiers and the corresponding component identifier. df['vertex'] Contains the vertex identifier df['labels'] The component identifier If G is a networkx.Graph, returns: python dictionary, where keys are vertices and values are the component identifiers. If G is a CuPy or SciPy matrix, returns: CuPy ndarray (if CuPy matrix input) or Numpy ndarray (if SciPy matrix input) of shape (<num vertices>, 2), where column 0 contains component identifiers and column 1 contains vertices. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.strongly_connected_components(G) """ (directed, connection, return_labels) = _ensure_args( "strongly_connected_components", G, directed, connection, return_labels ) # FIXME: allow nx_weight_attr to be specified (G, input_type) = ensure_cugraph_obj( G, nx_weight_attr="weight", matrix_graph_type=Graph(directed=directed) ) # Renumber the vertices so that they are contiguous (required) # FIXME: Remove 'renumbering' once the algo leverage the CAPI graph if not G.renumbered: edgelist = G.edgelist.edgelist_df renumbered_edgelist_df, renumber_map = G.renumber_map.renumber( edgelist, ["src"], ["dst"] ) renumbered_src_col_name = renumber_map.renumbered_src_col_name renumbered_dst_col_name = renumber_map.renumbered_dst_col_name G.edgelist.edgelist_df = renumbered_edgelist_df.rename( columns={renumbered_src_col_name: "src", renumbered_dst_col_name: "dst"} ) G.properties.renumbered = True G.renumber_map = renumber_map df = connectivity_wrapper.strongly_connected_components(G) if G.renumbered: df = G.unrenumber(df, "vertex") return _convert_df_to_output_type(df, input_type, return_labels) def connected_components(G, directed=None, connection="weak", return_labels=None): """ Generate either the strongly or weakly connected components and attach a component label to each vertex. Parameters ---------- G : cugraph.Graph, networkx.Graph, CuPy or SciPy sparse matrix Graph or matrix object, which should contain the connectivity information (edge weights are not used for this algorithm). If using a graph object, the graph can be either directed or undirected where an undirected edge is represented by a directed edge in both directions. The adjacency list will be computed if not already present. The number of vertices should fit into a 32b int. directed : bool, optional (default=True) NOTE For non-Graph-type (eg. sparse matrix) values of G only. Raises TypeError if used with a Graph object. If True, then convert the input matrix to a Graph(directed=True) and only move from point i to point j along paths csgraph[i, j]. If False, then find the shortest path on an undirected graph: the algorithm can progress from point i to j along csgraph[i, j] or csgraph[j, i]. connection : str, optional (default='weak') NOTE For Graph-type values of G, weak components are only supported for undirected graphs. [‘weak’|’strong’]. Return either weakly or strongly connected components. return_labels : bool, optional (default=True) NOTE For non-Graph-type (eg. sparse matrix) values of G only. Raises TypeError if used with a Graph object. If True, then return the labels for each of the connected components. Returns ------- Return value type is based on the input type. If G is a cugraph.Graph, returns: cudf.DataFrame GPU data frame containing two cudf.Series of size V: the vertex identifiers and the corresponding component identifier. df['vertex'] Contains the vertex identifier df['labels'] The component identifier If G is a networkx.Graph, returns: python dictionary, where keys are vertices and values are the component identifiers. If G is a CuPy or SciPy matrix, returns: CuPy ndarray (if CuPy matrix input) or Numpy ndarray (if SciPy matrix input) of shape (<num vertices>, 2), where column 0 contains component identifiers and column 1 contains vertices. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.connected_components(G, connection="weak") """ if connection == "weak": return weakly_connected_components(G, directed, connection, return_labels) elif connection == "strong": return strongly_connected_components(G, directed, connection, return_labels) else: raise ValueError( f"invalid connection type: {connection}, " "must be either 'strong' or 'weak'" )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/components/__init__.py
# Copyright (c) 2019-2021, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.components.connectivity import connected_components from cugraph.components.connectivity import weakly_connected_components from cugraph.components.connectivity import strongly_connected_components
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/testing/mg_utils.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import os import tempfile from pprint import pformat import time from dask.distributed import wait, default_client from dask import persist from dask.distributed import Client from dask.base import is_dask_collection from dask_cuda import LocalCUDACluster from dask_cuda.initialize import initialize from cugraph.dask.comms import comms as Comms from cugraph.dask.common.mg_utils import get_visible_devices from cugraph.generators import rmat import numpy as np def start_dask_client( protocol=None, rmm_async=False, rmm_pool_size=None, dask_worker_devices=None, jit_unspill=False, device_memory_limit=0.8, ): """ Creates a new dask client, and possibly also a cluster, and returns them as a tuple (client, cluster). If the env var SCHEDULER_FILE is set, it is assumed to contain the path to a JSON file generated by a running dask scheduler that can be used to configure the new dask client (the new client object returned will be a client to that scheduler), and the value of cluster will be None. If SCHEDULER_FILE is not set, a new LocalCUDACluster will be created and returned as the value of cluster. If the env var DASK_WORKER_DEVICES is set, it will be assumed to be a list of comma-separated GPU devices (ex. "0,1,2" for those 3 devices) for the LocalCUDACluster to use when setting up individual workers (1 worker per device). If not set, the parameter dask_worker_devices will be used the same way instead. If neither are set, the new LocalCUDACluster instance will default to one worker per device visible to this process. If the env var DASK_LOCAL_DIRECTORY is set, it will be used as the "local_directory" arg to LocalCUDACluster, for all temp files generated. Upon successful creation of a client (either to a LocalCUDACluster or otherwise), the cugraph.dask.comms.comms singleton is initialized using "p2p=True". Parameters ---------- protocol : str or None, default None The "protocol" arg to LocalCUDACluster (ex. "tcp"), see docs for dask_cuda.LocalCUDACluster for details. This parameter is ignored if the env var SCHEDULER_FILE is set which implies the dask cluster has already been created. rmm_pool_size : int, str or None, default None The "rmm_pool_size" arg to LocalCUDACluster (ex. "20GB"), see docs for dask_cuda.LocalCUDACluster for details. This parameter is ignored if the env var SCHEDULER_FILE is set which implies the dask cluster has already been created. dask_worker_devices : str, list of int, or None, default None GPUs to restrict activity to. Can be a string (like ``"0,1,2,3"``), list (like ``[0, 1, 2, 3]``), or ``None`` to use all available GPUs. This parameter is overridden by the value of env var DASK_WORKER_DEVICES. This parameter is ignored if the env var SCHEDULER_FILE is set which implies the dask cluster has already been created. jit_unspill : bool or None, default None The "jit_unspill" arg to LocalCUDACluster to enable just-in-time spilling, see docs for dask_cuda.LocalCUDACluster for details. This parameter is ignored if the env var SCHEDULER_FILE is set which implies the dask cluster has already been created. device_memory_limit : int, float, str, or None, default 0.8 The "device_memory_limit" arg to LocalCUDACluster to determine when workers start spilling to host memory, see docs for dask_cuda.LocalCUDACluster for details. This parameter is ignored if the env var SCHEDULER_FILE is set which implies the dask cluster has already been created. """ dask_scheduler_file = os.environ.get("SCHEDULER_FILE") dask_local_directory = os.getenv("DASK_LOCAL_DIRECTORY") # Allow the DASK_WORKER_DEVICES env var to override a value passed in. If # neither are set, this will be None. dask_worker_devices = os.getenv("DASK_WORKER_DEVICES", dask_worker_devices) cluster = None client = None tempdir_object = None if dask_scheduler_file: if protocol is not None: print( f"WARNING: {protocol=} is ignored in start_dask_client() when using " "dask SCHEDULER_FILE" ) if rmm_pool_size is not None: print( f"WARNING: {rmm_pool_size=} is ignored in start_dask_client() when " "using dask SCHEDULER_FILE" ) if dask_worker_devices is not None: print( f"WARNING: {dask_worker_devices=} is ignored in start_dask_client() " "when using dask SCHEDULER_FILE" ) initialize() client = Client(scheduler_file=dask_scheduler_file) # FIXME: use proper logging, INFO or DEBUG level print("\nDask client created using " f"{dask_scheduler_file}") else: if dask_local_directory is None: # The tempdir created by tempdir_object should be cleaned up once # tempdir_object is deleted. tempdir_object = tempfile.TemporaryDirectory() local_directory = tempdir_object.name else: local_directory = dask_local_directory cluster = LocalCUDACluster( local_directory=local_directory, protocol=protocol, rmm_pool_size=rmm_pool_size, rmm_async=rmm_async, CUDA_VISIBLE_DEVICES=dask_worker_devices, jit_unspill=jit_unspill, device_memory_limit=device_memory_limit, ) client = Client(cluster) if dask_worker_devices is None: num_workers = len(get_visible_devices()) else: if isinstance(dask_worker_devices, list): num_workers = len(dask_worker_devices) else: # FIXME: this assumes a properly formatted string with commas num_workers = len(dask_worker_devices.split(",")) client.wait_for_workers(num_workers) # Add a reference to tempdir_object to the client to prevent it from # being deleted when this function returns. This will be deleted in # stop_dask_client() client.tempdir_object = tempdir_object # FIXME: use proper logging, INFO or DEBUG level print("\nDask client/cluster created using LocalCUDACluster") Comms.initialize(p2p=True) return (client, cluster) def stop_dask_client(client, cluster=None): """ Shutdown/cleanup a client and possibly cluster object returned from start_dask_client(). This also stops the cugraph.dask.comms.comms singleton. """ Comms.destroy() client.close() if cluster: cluster.close() # Remove a TemporaryDirectory object that may have been assigned to the # client, which should remove it and all the contents from disk. if hasattr(client, "tempdir_object"): del client.tempdir_object # FIXME: use proper logging, INFO or DEBUG level print("\nDask client closed.") def restart_client(client): """ Restart the Dask client """ Comms.destroy() client.restart() client = client.run(enable_spilling) Comms.initialize(p2p=True) def enable_spilling(): import cudf cudf.set_option("spill", True) def generate_edgelist_rmat( scale, edgefactor, seed=None, unweighted=False, mg=True, ): """ Returns a cudf/dask_cudf DataFrame created using the R-MAT graph generator. The resulting graph is weighted with random values of a uniform distribution from the interval [0, 1) Args: scale: scale is used to determine the number of vertices to be generated (num_verts = 2^scale), which is also used to determine the data type for the vertex ID values in the DataFrame. edgefactor: edgefactor determies the number of edges (num_edges = num_edges*edgefactor) seed: seed, if specified, will be used as the seed to the RNG. unweighted: unweighted determines if the resulting edgelist will have randomly-generated weights ranging in value between [0, 1). If True, an edgelist with only 2 columns is returned. mg: mg determines if the resulting edgelist will be a multi-GPU edgelist. If True, returns a dask_cudf.DataFrame and if False, returns a cudf.DataFrame. """ ddf = rmat( scale, (2**scale) * edgefactor, 0.57, # from Graph500 0.19, # from Graph500 0.19, # from Graph500 seed or 42, clip_and_flip=False, scramble_vertex_ids=True, create_using=None, # return edgelist instead of Graph instance mg=mg, ) if not unweighted: rng = np.random.default_rng(seed) ddf["weight"] = ddf.map_partitions(lambda df: rng.random(size=len(df))) return ddf def set_statistics_adaptor(): """ Sets the current device resource to a StatisticsResourceAdaptor """ import rmm rmm.mr.set_current_device_resource( rmm.mr.StatisticsResourceAdaptor(rmm.mr.get_current_device_resource()) ) def _get_allocation_counts(): """ Returns the allocation counts from the current device resource """ import rmm mr = rmm.mr.get_current_device_resource() if not hasattr(mr, "allocation_counts"): if hasattr(mr, "upstream_mr"): return _get_allocation_counts(mr.upstream_mr) else: return -1 else: return mr.allocation_counts def persist_dask_object(arg): """ Persist if it is a dask object """ if is_dask_collection(arg) or hasattr(arg, "persist"): arg = persist(arg) wait(arg) arg = arg[0] return arg # Function to convert bytes into human readable format def sizeof_fmt(num, suffix="B"): if isinstance(num, str): if num[-2:] == "GB": return num[:-2] + "G" elif num[-2:] == "MB": return num[:-2] + "M" elif num[-2:] == "KB": return num[:-2] + "K" else: raise ValueError("unknown unit") for unit in ["", "K", "M", "G", "T", "P", "E", "Z"]: if abs(num) < 1024.0: return "%3.1f%s%s" % (num, unit, suffix) num /= 1024.0 return "%.1f%s%s" % (num, "Yi", suffix) def _parse_allocation_counts(allocation_counts): """ Parses the allocation counts from the current device resource into human readable format """ return {k: sizeof_fmt(v) for k, v in allocation_counts.items() if "bytes" in k} # Decorator to set the statistics adaptor # and calls the allocation_counts function def get_allocation_counts_dask_lazy(return_allocations=False, logging=True): def decorator(func): def wrapper(*args, **kwargs): client = default_client() client.run(set_statistics_adaptor) st = time.time() return_val = func(*args, **kwargs) et = time.time() allocation_counts = client.run(_get_allocation_counts) if logging: _print_allocation_statistics( func, args, kwargs, et - st, allocation_counts ) client.run(set_statistics_adaptor) if return_allocations: return return_val, allocation_counts else: return return_val return wrapper return decorator def get_allocation_counts_dask_persist(return_allocations=False, logging=True): def decorator(func): def wrapper(*args, **kwargs): args = [persist_dask_object(a) for a in args] kwargs = {k: persist_dask_object(v) for k, v in kwargs.items()} client = default_client() client.run(set_statistics_adaptor) st = time.time() return_val = func(*args, **kwargs) return_val = persist_dask_object(return_val) if isinstance(return_val, (list, tuple)): return_val = [persist_dask_object(d) for d in return_val] et = time.time() allocation_counts = client.run(_get_allocation_counts) if logging: _print_allocation_statistics( func, args, kwargs, et - st, allocation_counts ) client.run(set_statistics_adaptor) if return_allocations: return return_val, allocation_counts else: return return_val return wrapper return decorator def _get_allocation_stats_string(func, args, kwargs, execution_time, allocation_counts): allocation_counts_parsed = { worker_id: _parse_allocation_counts(worker_allocations) for worker_id, worker_allocations in allocation_counts.items() } return ( f"function: {func.__name__}\n" + f"function args: {args} kwargs: {kwargs}\n" + f"execution_time: {execution_time}\n" + "allocation_counts:\n" + f"{pformat(allocation_counts_parsed, indent=4, width=1, compact=True)}" ) def _print_allocation_statistics(func, args, kwargs, execution_time, allocation_counts): print( _get_allocation_stats_string( func, args, kwargs, execution_time, allocation_counts ) ) def get_peak_output_ratio_across_workers(allocation_counts): peak_ratio = -1 for w_allocations in allocation_counts.values(): w_peak_ratio = w_allocations["peak_bytes"] / w_allocations["current_bytes"] peak_ratio = max(w_peak_ratio, peak_ratio) return peak_ratio
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/testing/resultset.py
# Copyright (c) 2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import tarfile import urllib.request import cudf from cugraph.datasets.dataset import ( DefaultDownloadDir, default_download_dir, ) # results_dir_path = utils.RAPIDS_DATASET_ROOT_DIR_PATH / "tests" / "resultsets" class Resultset: """ A Resultset Object, used to store golden results to easily run tests that need to access said results without the overhead of running an algorithm to get the results. Parameters ---------- data_dictionary : dict The existing algorithm output, expected as a dictionary """ def __init__(self, data_dictionary): self._data_dictionary = data_dictionary def get_cudf_dataframe(self): """ Converts the existing algorithm output from a dictionary to a cudf.DataFrame before writing the DataFrame to output into a csv """ return cudf.DataFrame(self._data_dictionary) _resultsets = {} def get_resultset(resultset_name, **kwargs): """ Returns the golden results for a specific test. Parameters ---------- resultset_name : String Name of the test's module (currently just 'traversal' is supported) kwargs : All distinct test details regarding the choice of algorithm, dataset, and graph """ arg_dict = dict(kwargs) arg_dict["resultset_name"] = resultset_name # Example: # {'a': 1, 'z': 9, 'c': 5, 'b': 2} becomes 'a-1-b-2-c-5-z-9' resultset_key = "-".join( [ str(val) for arg_dict_pair in sorted(arg_dict.items()) for val in arg_dict_pair ] ) uuid = _resultsets.get(resultset_key) if uuid is None: raise KeyError(f"results for {arg_dict} not found") results_dir_path = default_resultset_download_dir.path results_filename = results_dir_path / (uuid + ".csv") return cudf.read_csv(results_filename) default_resultset_download_dir = DefaultDownloadDir(subdir="tests/resultsets") def load_resultset(resultset_name, resultset_download_url): """ Read a mapping file (<resultset_name>.csv) in the _results_dir and save the mappings between each unique set of args/identifiers to UUIDs to the _resultsets dictionary. If <resultset_name>.csv does not exist in _results_dir, use resultset_download_url to download a file to install/unpack/etc. to _results_dir first. """ # curr_resultset_download_dir = get_resultset_download_dir() curr_resultset_download_dir = default_resultset_download_dir.path # curr_download_dir = path curr_download_dir = default_download_dir.path mapping_file_path = curr_resultset_download_dir / (resultset_name + "_mappings.csv") if not mapping_file_path.exists(): # Downloads a tar gz from s3 bucket, then unpacks the results files compressed_file_dir = curr_download_dir / "tests" compressed_file_path = compressed_file_dir / "resultsets.tar.gz" if not curr_resultset_download_dir.exists(): curr_resultset_download_dir.mkdir(parents=True, exist_ok=True) if not compressed_file_path.exists(): urllib.request.urlretrieve(resultset_download_url, compressed_file_path) tar = tarfile.open(str(compressed_file_path), "r:gz") tar.extractall(str(curr_resultset_download_dir)) tar.close() # FIXME: This assumes separator is " ", but should this be configurable? sep = " " with open(mapping_file_path) as mapping_file: for line in mapping_file.readlines(): if line.startswith("#"): continue (uuid, *row_args) = line.split(sep) if (len(row_args) % 2) != 0: raise ValueError( f'bad row in {mapping_file_path}: "{line}", must ' "contain UUID followed by an even number of items" ) row_keys = row_args[::2] row_vals = row_args[1::2] row_keys = " ".join(row_keys).split() row_vals = " ".join(row_vals).split() arg_dict = dict(zip(row_keys, row_vals)) arg_dict["resultset_name"] = resultset_name # Create a unique string key for the _resultsets dict based on # sorted row_keys. Looking up results based on args will also have # to sort, but this will ensure results can looked up without # requiring maintaining a specific order. Example: # {'a': 1, 'z': 9, 'c': 5, 'b': 2} becomes 'a-1-b-2-c-5-z-9' resultset_key = "-".join( [ str(val) for arg_dict_pair in sorted(arg_dict.items()) for val in arg_dict_pair ] ) _resultsets[resultset_key] = uuid
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/testing/generate_resultsets.py
# Copyright (c) 2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from tempfile import NamedTemporaryFile import random import numpy as np import networkx as nx import cudf import cugraph from cugraph.datasets import dolphins, netscience, karate_disjoint, karate # from cugraph.testing import utils, Resultset, SMALL_DATASETS, results_dir_path from cugraph.testing import ( utils, Resultset, SMALL_DATASETS, default_resultset_download_dir, ) _resultsets = {} def add_resultset(result_data_dictionary, **kwargs): rs = Resultset(result_data_dictionary) hashable_dict_repr = tuple((k, kwargs[k]) for k in sorted(kwargs.keys())) _resultsets[hashable_dict_repr] = rs if __name__ == "__main__": # ============================================================================= # Parameters # ============================================================================= SEEDS = [42] DIRECTED_GRAPH_OPTIONS = [True, False] DEPTH_LIMITS = [None, 1, 5, 18] DATASETS = [dolphins, netscience, karate_disjoint] # ============================================================================= # tests/traversal/test_bfs.py # ============================================================================= test_bfs_results = {} for ds in DATASETS + [karate]: for seed in SEEDS: for depth_limit in DEPTH_LIMITS: for dirctd in DIRECTED_GRAPH_OPTIONS: # this is used for get_cu_graph_golden_results_and_params Gnx = utils.generate_nx_graph_from_file( ds.get_path(), directed=dirctd ) random.seed(seed) start_vertex = random.sample(list(Gnx.nodes()), 1)[0] golden_values = nx.single_source_shortest_path_length( Gnx, start_vertex, cutoff=depth_limit ) vertices = cudf.Series(golden_values.keys()) distances = cudf.Series(golden_values.values()) add_resultset( {"vertex": vertices, "distance": distances}, graph_dataset=ds.metadata["name"], graph_directed=str(dirctd), algo="single_source_shortest_path_length", start_vertex=str(start_vertex), cutoff=str(depth_limit), ) # these are pandas dataframes for dirctd in DIRECTED_GRAPH_OPTIONS: Gnx = utils.generate_nx_graph_from_file(karate.get_path(), directed=dirctd) golden_result = cugraph.bfs_edges(Gnx, source=7) cugraph_df = cudf.from_pandas(golden_result) add_resultset( cugraph_df, graph_dataset="karate", graph_directed=str(dirctd), algo="bfs_edges", source="7", ) # ============================================================================= # tests/traversal/test_sssp.py # ============================================================================= test_sssp_results = {} SOURCES = [1] for ds in SMALL_DATASETS: for source in SOURCES: Gnx = utils.generate_nx_graph_from_file(ds.get_path(), directed=True) golden_paths = nx.single_source_dijkstra_path_length(Gnx, source) vertices = cudf.Series(golden_paths.keys()) distances = cudf.Series(golden_paths.values()) add_resultset( {"vertex": vertices, "distance": distances}, graph_dataset=ds.metadata["name"], graph_directed="True", algo="single_source_dijkstra_path_length", source=str(source), ) M = utils.read_csv_for_nx(ds.get_path(), read_weights_in_sp=True) edge_attr = "weight" Gnx = nx.from_pandas_edgelist( M, source="0", target="1", edge_attr=edge_attr, create_using=nx.DiGraph(), ) M["weight"] = M["weight"].astype(np.int32) Gnx = nx.from_pandas_edgelist( M, source="0", target="1", edge_attr="weight", create_using=nx.DiGraph(), ) golden_paths_datatypeconv = nx.single_source_dijkstra_path_length( Gnx, source ) vertices_datatypeconv = cudf.Series(golden_paths_datatypeconv.keys()) distances_datatypeconv = cudf.Series(golden_paths_datatypeconv.values()) add_resultset( {"vertex": vertices_datatypeconv, "distance": distances_datatypeconv}, graph_dataset=ds.metadata["name"], graph_directed="True", algo="single_source_dijkstra_path_length", test="data_type_conversion", source=str(source), ) for dirctd in DIRECTED_GRAPH_OPTIONS: for source in SOURCES: Gnx = utils.generate_nx_graph_from_file( karate.get_path(), directed=dirctd, edgevals=True ) add_resultset( cugraph.sssp(Gnx, source), graph_dataset="karate", graph_directed=str(dirctd), algo="sssp_nonnative", source=str(source), ) Gnx = nx.Graph() Gnx.add_edge(0, 1, other=10) Gnx.add_edge(1, 2, other=20) df = cugraph.sssp(Gnx, 0, edge_attr="other") add_resultset(df, algo="sssp_nonnative", test="network_edge_attr") # ============================================================================= # tests/traversal/test_paths.py # ============================================================================= CONNECTED_GRAPH = """1,5,3 1,4,1 1,2,1 1,6,2 1,7,2 4,5,1 2,3,1 7,6,2 """ DISCONNECTED_GRAPH = CONNECTED_GRAPH + "8,9,4" paths = [("1", "1"), ("1", "5"), ("1", "3"), ("1", "6")] invalid_paths = { "connected": [("-1", "1"), ("0", "42")], "disconnected": [("1", "10"), ("1", "8")], } with NamedTemporaryFile(mode="w+", suffix=".csv") as graph_tf: graph_tf.writelines(DISCONNECTED_GRAPH) graph_tf.seek(0) Gnx_DIS = nx.read_weighted_edgelist(graph_tf.name, delimiter=",") res1 = nx.shortest_path_length(Gnx_DIS, source="1", weight="weight") vertices = cudf.Series(res1.keys()) distances = cudf.Series(res1.values()) add_resultset( {"vertex": vertices, "distance": distances}, algo="shortest_path_length", graph_dataset="DISCONNECTED", graph_directed="True", source="1", weight="weight", ) # NOTE: Currently, only traversal result files are generated random.seed(24) traversal_mappings = cudf.DataFrame( columns=[ "#UUID", "arg0", "arg0val", "arg1", "arg1val", "arg2", "arg2val", "arg3", "arg3val", "arg4", "arg4val", "arg5", "arg5val", "arg6", "arg6val", "arg7", "arg7val", "arg8", "arg8val", "arg9", "arg9val", ] ) # Generating ALL results files results_dir_path = default_resultset_download_dir.path if not results_dir_path.exists(): results_dir_path.mkdir(parents=True, exist_ok=True) for temp in _resultsets: res = _resultsets[temp].get_cudf_dataframe() temp_filename = str(random.getrandbits(50)) temp_dict = dict(temp) argnames, argvals = [t for t in temp_dict.keys()], [ t for t in temp_dict.values() ] single_mapping = np.empty(21, dtype=object) dict_length = len(argnames) single_mapping[0] = temp_filename for i in np.arange(dict_length): single_mapping[2 * i + 1] = argnames[i] single_mapping[2 * i + 2] = argvals[i] temp_mapping = cudf.DataFrame( [single_mapping], columns=[ "#UUID", "arg0", "arg0val", "arg1", "arg1val", "arg2", "arg2val", "arg3", "arg3val", "arg4", "arg4val", "arg5", "arg5val", "arg6", "arg6val", "arg7", "arg7val", "arg8", "arg8val", "arg9", "arg9val", ], ) traversal_mappings = cudf.concat( [traversal_mappings, temp_mapping], axis=0, ignore_index=True ) res.to_csv(results_dir_path / (temp_filename + ".csv"), index=False) traversal_mappings.to_csv( results_dir_path / "traversal_mappings.csv", index=False, sep=" " )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/testing/__init__.py
# Copyright (c) 2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.testing.utils import ( RAPIDS_DATASET_ROOT_DIR_PATH, RAPIDS_DATASET_ROOT_DIR, ) from cugraph.testing.resultset import ( Resultset, load_resultset, get_resultset, default_resultset_download_dir, ) from cugraph.datasets import ( cyber, dining_prefs, dolphins, karate, karate_disjoint, polbooks, netscience, small_line, small_tree, email_Eu_core, toy_graph, toy_graph_undirected, soc_livejournal, cit_patents, europe_osm, hollywood, # twitter, ) # # Moved Dataset Batches # UNDIRECTED_DATASETS = [karate, dolphins] SMALL_DATASETS = [karate, dolphins, polbooks] WEIGHTED_DATASETS = [ dining_prefs, dolphins, karate, karate_disjoint, netscience, polbooks, small_line, small_tree, ] ALL_DATASETS = [ dining_prefs, dolphins, karate, karate_disjoint, polbooks, netscience, small_line, small_tree, email_Eu_core, toy_graph, toy_graph_undirected, ] DEFAULT_DATASETS = [dolphins, netscience, karate_disjoint] BENCHMARKING_DATASETS = [soc_livejournal, cit_patents, europe_osm, hollywood]
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/testing/utils.py
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import os # Assume test environment has the following dependencies installed import pytest import pandas as pd import networkx as nx import numpy as np import cupy as cp from cupyx.scipy.sparse import coo_matrix as cp_coo_matrix from cupyx.scipy.sparse import csr_matrix as cp_csr_matrix from cupyx.scipy.sparse import csc_matrix as cp_csc_matrix from scipy.sparse import coo_matrix as sp_coo_matrix from scipy.sparse import csr_matrix as sp_csr_matrix from scipy.sparse import csc_matrix as sp_csc_matrix from pathlib import Path import cudf import dask_cudf import cugraph from cugraph.dask.common.mg_utils import get_client CUPY_MATRIX_TYPES = [cp_coo_matrix, cp_csr_matrix, cp_csc_matrix] SCIPY_MATRIX_TYPES = [sp_coo_matrix, sp_csr_matrix, sp_csc_matrix] RAPIDS_DATASET_ROOT_DIR = os.getenv( "RAPIDS_DATASET_ROOT_DIR", os.path.join(os.path.dirname(__file__), "../datasets") ) RAPIDS_DATASET_ROOT_DIR_PATH = Path(RAPIDS_DATASET_ROOT_DIR) # # Datasets # DATASETS_UNDIRECTED = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["karate.csv", "dolphins.csv"] ] DATASETS_UNDIRECTED_WEIGHTS = [RAPIDS_DATASET_ROOT_DIR_PATH / "netscience.csv"] DATASETS_UNRENUMBERED = [Path(RAPIDS_DATASET_ROOT_DIR) / "karate-disjoint.csv"] DATASETS = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["karate-disjoint.csv", "dolphins.csv", "netscience.csv"] ] DATASETS_MULTI_EDGES = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["karate_multi_edge.csv", "dolphins_multi_edge.csv"] ] DATASETS_STR_ISLT_V = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["karate_mod.mtx", "karate_str.mtx"] ] DATASETS_SELF_LOOPS = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["karate_s_loop.csv", "dolphins_s_loop.csv"] ] # '../datasets/email-Eu-core.csv'] STRONGDATASETS = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["dolphins.csv", "netscience.csv", "email-Eu-core.csv"] ] DATASETS_KTRUSS = [ ( RAPIDS_DATASET_ROOT_DIR_PATH / "polbooks.csv", RAPIDS_DATASET_ROOT_DIR_PATH / "ref/ktruss/polbooks.csv", ) ] DATASETS_TSPLIB = [ (RAPIDS_DATASET_ROOT_DIR_PATH / f,) + (d,) for (f, d) in [ ("gil262.tsp", 2378), ("eil51.tsp", 426), ("kroA100.tsp", 21282), ("tsp225.tsp", 3916), ] ] DATASETS_SMALL = [ RAPIDS_DATASET_ROOT_DIR_PATH / f for f in ["karate.csv", "dolphins.csv", "polbooks.csv"] ] MATRIX_INPUT_TYPES = [ pytest.param(cp_coo_matrix, marks=pytest.mark.matrix_types, id="CuPy.coo_matrix"), pytest.param(cp_csr_matrix, marks=pytest.mark.matrix_types, id="CuPy.csr_matrix"), pytest.param(cp_csc_matrix, marks=pytest.mark.matrix_types, id="CuPy.csc_matrix"), ] NX_INPUT_TYPES = [ pytest.param(nx.Graph, marks=pytest.mark.nx_types, id="nx.Graph"), ] NX_DIR_INPUT_TYPES = [ pytest.param(nx.Graph, marks=pytest.mark.nx_types, id="nx.DiGraph"), ] CUGRAPH_INPUT_TYPES = [ pytest.param(cugraph.Graph(), marks=pytest.mark.cugraph_types, id="cugraph.Graph"), ] CUGRAPH_DIR_INPUT_TYPES = [ pytest.param( cugraph.Graph(directed=True), marks=pytest.mark.cugraph_types, id="cugraph.Graph(directed=True)", ), ] def read_csv_for_nx(csv_file, read_weights_in_sp=True, read_weights=True): if read_weights: if read_weights_in_sp is True: df = pd.read_csv( csv_file, delimiter=" ", header=None, names=["0", "1", "weight"], dtype={"0": "int32", "1": "int32", "weight": "float32"}, ) else: df = pd.read_csv( csv_file, delimiter=" ", header=None, names=["0", "1", "weight"], dtype={"0": "int32", "1": "int32", "weight": "float64"}, ) else: df = pd.read_csv( csv_file, delimiter=" ", header=None, names=["0", "1"], usecols=["0", "1"], dtype={"0": "int32", "1": "int32"}, ) return df def create_obj_from_csv( csv_file_name, obj_type, csv_has_weights=True, edgevals=False, directed=False ): """ Return an object based on obj_type populated with the contents of csv_file_name """ if obj_type in [cugraph.Graph]: return generate_cugraph_graph_from_file( csv_file_name, directed=directed, edgevals=edgevals, ) elif isinstance(obj_type, cugraph.Graph): return generate_cugraph_graph_from_file( csv_file_name, directed=directed, edgevals=edgevals, ) elif obj_type in SCIPY_MATRIX_TYPES + CUPY_MATRIX_TYPES: # FIXME: assuming float32 if csv_has_weights: (rows, cols, weights) = np.genfromtxt( csv_file_name, delimiter=" ", dtype=np.float32, unpack=True ) else: (rows, cols) = np.genfromtxt( csv_file_name, delimiter=" ", dtype=np.float32, unpack=True ) if (csv_has_weights is False) or (edgevals is False): # COO matrices must have a value array. Also if edgevals are to be # ignored (False), reset all weights to 1. weights = np.array([1] * len(rows)) if obj_type in CUPY_MATRIX_TYPES: coo = cp_coo_matrix( (cp.asarray(weights), (cp.asarray(rows), cp.asarray(cols))), dtype=np.float32, ) else: coo = sp_coo_matrix( (weights, (np.array(rows, dtype=int), np.array(cols, dtype=int))), ) if obj_type in [cp_csr_matrix, sp_csr_matrix]: return coo.tocsr(copy=False) elif obj_type in [cp_csc_matrix, sp_csc_matrix]: return coo.tocsc(copy=False) else: return coo elif obj_type in [nx.Graph, nx.DiGraph]: return generate_nx_graph_from_file( csv_file_name, directed=(obj_type is nx.DiGraph), edgevals=edgevals ) else: raise TypeError(f"unsupported type: {obj_type}") def read_csv_file(csv_file, read_weights_in_sp=True): if read_weights_in_sp is True: return cudf.read_csv( csv_file, delimiter=" ", dtype={"0": "int32", "1": "int32", "2": "float32"}, header=None, ) else: return cudf.read_csv( csv_file, delimiter=" ", dtype={"0": "int32", "1": "int32", "2": "float64"}, header=None, ) def read_dask_cudf_csv_file(csv_file, read_weights_in_sp=True, single_partition=True): if read_weights_in_sp is True: if single_partition: chunksize = os.path.getsize(csv_file) return dask_cudf.read_csv( csv_file, chunksize=chunksize, delimiter=" ", names=["src", "dst", "weight"], dtype=["int32", "int32", "float32"], header=None, ) else: return dask_cudf.read_csv( csv_file, delimiter=" ", names=["src", "dst", "weight"], dtype=["int32", "int32", "float32"], header=None, ) else: if single_partition: chunksize = os.path.getsize(csv_file) return dask_cudf.read_csv( csv_file, chunksize=chunksize, delimiter=" ", names=["src", "dst", "weight"], dtype=["int32", "int32", "float32"], header=None, ) else: return dask_cudf.read_csv( csv_file, delimiter=" ", names=["src", "dst", "weight"], dtype=["int32", "int32", "float64"], header=None, ) def generate_nx_graph_from_file(graph_file, directed=True, edgevals=False): M = read_csv_for_nx(graph_file, read_weights_in_sp=edgevals) edge_attr = "weight" if edgevals else None Gnx = nx.from_pandas_edgelist( M, create_using=(nx.DiGraph() if directed else nx.Graph()), source="0", target="1", edge_attr=edge_attr, ) return Gnx def generate_cugraph_graph_from_file(graph_file, directed=True, edgevals=False): cu_M = read_csv_file(graph_file) G = cugraph.Graph(directed=directed) if edgevals: G.from_cudf_edgelist(cu_M, source="0", destination="1", edge_attr="2") else: G.from_cudf_edgelist(cu_M, source="0", destination="1") return G def generate_mg_batch_cugraph_graph_from_file(graph_file, directed=True): client = get_client() _ddf = read_dask_cudf_csv_file(graph_file) ddf = client.persist(_ddf) G = cugraph.Graph(directed=directed) G.from_dask_cudf_edgelist(ddf) return G def build_cu_and_nx_graphs(graph_file, directed=True, edgevals=False): G = generate_cugraph_graph_from_file( graph_file, directed=directed, edgevals=edgevals ) Gnx = generate_nx_graph_from_file(graph_file, directed=directed, edgevals=edgevals) return G, Gnx def build_mg_batch_cu_and_nx_graphs(graph_file, directed=True): G = generate_mg_batch_cugraph_graph_from_file(graph_file, directed=directed) Gnx = generate_nx_graph_from_file(graph_file, directed=directed) return G, Gnx def random_edgelist( e=1024, ef=16, dtypes={"src": np.int32, "dst": np.int32, "val": float}, drop_duplicates=True, seed=None, ): """Create a random edge list Parameters ---------- e : int Number of edges ef : int Edge factor (average number of edges per vertex) dtypes : dict Mapping of column names to types. Supported type is {"src": int, "dst": int, "val": float} drop_duplicates Drop duplicates seed : int (optional) Randomstate seed Examples -------- >>> from cugraph.testing import utils >>> # genrates 20 df with 100M edges each and write to disk >>> for x in range(20): >>> df = utils.random_edgelist(e=100000000, ef=64, >>> dtypes={'src':np.int32, 'dst':np.int32}, >>> seed=x) >>> df.to_csv('df'+str(x), header=False, index=False) >>> #df.to_parquet('files_parquet/df'+str(x), index=False) """ state = np.random.RandomState(seed) columns = dict((k, make[dt](e // ef, e, state)) for k, dt in dtypes.items()) df = pd.DataFrame(columns) if drop_duplicates: df = df.drop_duplicates(subset=["src", "dst"]) print("Generated " + str(df.shape[0]) + " edges") return df def make_int32(v, e, rstate): return rstate.randint(low=0, high=v, size=e, dtype=np.int32) def make_int64(v, e, rstate): return rstate.randint(low=0, high=v, size=e, dtype=np.int64) def make_float(v, e, rstate): return rstate.rand(e) make = {float: make_float, np.int32: make_int32, np.int64: make_int64} # shared between min and max spanning tree tests def compare_mst(mst_cugraph, mst_nx): mst_nx_df = nx.to_pandas_edgelist(mst_nx) edgelist_df = mst_cugraph.view_edge_list() assert len(mst_nx_df) == len(edgelist_df) # check cycles Gnx = nx.from_pandas_edgelist( edgelist_df.to_pandas(), create_using=nx.Graph(), source="src", target="dst", ) try: lc = nx.find_cycle(Gnx, source=None, orientation="ignore") print(lc) except nx.NetworkXNoCycle: pass # check total weight cg_sum = edgelist_df[mst_cugraph.weight_column].sum() nx_sum = mst_nx_df["weight"].sum() print(cg_sum) print(nx_sum) assert np.isclose(cg_sum, nx_sum)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/__init__.py
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from .link_analysis.pagerank import pagerank from .link_analysis.hits import hits from .traversal.bfs import bfs from .traversal.sssp import sssp from .common.read_utils import get_chunksize from .common.read_utils import get_n_workers from .community.louvain import louvain from .community.triangle_count import triangle_count from .community.egonet import ego_graph from .community.induced_subgraph import induced_subgraph from .centrality.katz_centrality import katz_centrality from .components.connectivity import weakly_connected_components from .sampling.uniform_neighbor_sample import uniform_neighbor_sample from .sampling.random_walks import random_walks from .centrality.eigenvector_centrality import eigenvector_centrality from .cores.core_number import core_number from .centrality.betweenness_centrality import betweenness_centrality from .centrality.betweenness_centrality import edge_betweenness_centrality from .cores.k_core import k_core from .link_prediction.jaccard import jaccard from .link_prediction.sorensen import sorensen from .link_prediction.overlap import overlap from .community.leiden import leiden
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/components/CMakeLists.txt
# ============================================================================= # Copyright (c) 2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except # in compliance with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software distributed under the License # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express # or implied. See the License for the specific language governing permissions and limitations under # the License. # ============================================================================= set(cython_sources mg_connectivity_wrapper.pyx) set(linked_libraries cugraph::cugraph) rapids_cython_create_modules( CXX SOURCE_FILES "${cython_sources}" LINKED_LIBRARIES "${linked_libraries}" MODULE_PREFIX components_ ASSOCIATED_TARGETS cugraph )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/components/connectivity.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from pylibcugraph import ResourceHandle from pylibcugraph import weakly_connected_components as pylibcugraph_wcc def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_vertex, cupy_labels = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertex df["labels"] = cupy_labels return df def _call_plc_wcc(sID, mg_graph_x, do_expensive_check): return pylibcugraph_wcc( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, offsets=None, indices=None, weights=None, labels=None, do_expensive_check=do_expensive_check, ) def weakly_connected_components(input_graph): """ Generate the Weakly Connected Components and attach a component label to each vertex. Parameters ---------- input_graph : cugraph.Graph The graph descriptor should contain the connectivity information and weights. The adjacency list will be computed if not already present. The current implementation only supports undirected graphs. Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing 2 dask_cudf.Series ddf['vertex']: dask_cudf.Series Contains the vertex identifiers ddf['labels']: dask_cudf.Series Contains the wcc labels Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=False) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst', ... edge_attr='value') >>> result = dcg.weakly_connected_components(dg) """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") # Initialize dask client client = default_client() do_expensive_check = False result = [ client.submit( _call_plc_wcc, Comms.get_session_id(), input_graph._plc_graph[w], do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/components/__init__.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/community/triangle_count.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from pylibcugraph import ResourceHandle, triangle_count as pylibcugraph_triangle_count def _call_triangle_count( sID, mg_graph_x, start_list, do_expensive_check, ): return pylibcugraph_triangle_count( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, start_list=start_list, do_expensive_check=do_expensive_check, ) def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_vertices, cupy_counts = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertices df["counts"] = cupy_counts return df def triangle_count(input_graph, start_list=None): """ Computes the number of triangles (cycles of length three) and the number per vertex in the input graph. Parameters ---------- input_graph : cugraph.graph cuGraph graph descriptor, should contain the connectivity information, (edge weights are not used in this algorithm). The current implementation only supports undirected graphs. start_list : list or cudf.Series list of vertices for triangle count. if None the entire set of vertices in the graph is processed Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing 2 dask_cudf.Series ddf['vertex']: dask_cudf.Series Contains the triangle counting vertices ddf['counts']: dask_cudf.Series Contains the triangle counting counts """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") # Initialize dask client client = default_client() if start_list is not None: if isinstance(start_list, int): start_list = [start_list] if isinstance(start_list, list): start_list = cudf.Series(start_list) if not isinstance(start_list, cudf.Series): raise TypeError( f"'start_list' must be either a list or a cudf.Series," f"got: {start_list.dtype}" ) # start_list uses "external" vertex IDs, but since the graph has been # renumbered, the start vertex IDs must also be renumbered. if input_graph.renumbered: start_list = input_graph.lookup_internal_vertex_id(start_list).compute() do_expensive_check = False result = [ client.submit( _call_triangle_count, Comms.get_session_id(), input_graph._plc_graph[w], start_list, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/community/induced_subgraph.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf import cupy as cp from cugraph.dask.common.input_utils import get_distributed_data from typing import Union, Tuple from pylibcugraph import ( ResourceHandle, induced_subgraph as pylibcugraph_induced_subgraph, ) def _call_induced_subgraph( sID: bytes, mg_graph_x, vertices: cudf.Series, offsets: cudf.Series, do_expensive_check: bool, ) -> Tuple[cp.ndarray, cp.ndarray, cp.ndarray, cp.ndarray]: return pylibcugraph_induced_subgraph( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, subgraph_vertices=vertices, subgraph_offsets=offsets, do_expensive_check=do_expensive_check, ) def consolidate_results(df: cudf.DataFrame, offsets: cudf.Series) -> cudf.DataFrame: """ Each rank returns its induced_subgraph dataframe with its corresponding offsets array. This is ideal if the user operates on distributed memory but when attempting to bring the result into a single machine, the induced_subgraph dataframes generated from each seed cannot be extracted using the offsets array. This function consolidate the final result by performing segmented copies. Returns: consolidated induced_subgraph dataframe """ for i in range(len(offsets) - 1): df_tmp = df[offsets[i] : offsets[i + 1]] df_tmp["labels"] = i if i == 0: df_consolidate = df_tmp else: df_consolidate = cudf.concat([df_consolidate, df_tmp]) return df_consolidate def convert_to_cudf(cp_arrays: cp.ndarray) -> cudf.DataFrame: cp_src, cp_dst, cp_weight, cp_offsets = cp_arrays df = cudf.DataFrame() df["src"] = cp_src df["dst"] = cp_dst df["weight"] = cp_weight offsets = cudf.Series(cp_offsets) return consolidate_results(df, offsets) def induced_subgraph( input_graph, vertices: Union[cudf.Series, cudf.DataFrame], offsets: Union[list, cudf.Series] = None, ) -> Tuple[dask_cudf.DataFrame, dask_cudf.Series]: """ Compute a subgraph of the existing graph including only the specified vertices. This algorithm works with both directed and undirected graphs and does not actually traverse the edges, but instead simply pulls out any edges that are incident on vertices that are both contained in the vertices list. If no subgraph can be extracted from the vertices provided, a 'None' value will be returned. Parameters ---------- input_graph : cugraph.Graph Graph or matrix object, which should contain the connectivity information. Edge weights, if present, should be single or double precision floating point values. vertices : cudf.Series or cudf.DataFrame Specifies the vertices of the induced subgraph. For multi-column vertices, vertices should be provided as a cudf.DataFrame offsets : list or cudf.Series, optional Specifies the subgraph offsets into the subgraph vertices. If no offsets array is provided, a default array [0, len(vertices)] will be used. Returns ------- ego_edge_lists : dask_cudf.DataFrame Distributed GPU data frame containing all induced sources identifiers, destination identifiers, edge weights seeds_offsets: dask_cudf.Series Distributed Series containing the starting offset in the returned edge list for each seed. """ # Initialize dask client client = default_client() if isinstance(vertices, (int, list)): vertices = cudf.Series(vertices) elif not isinstance( vertices, (cudf.Series, dask_cudf.Series, cudf.DataFrame, dask_cudf.DataFrame) ): raise TypeError( f"'vertices' must be either an integer or a list or a " f"cudf or dask_cudf Series or DataFrame, got: {type(vertices)}" ) if isinstance(offsets, list): offsets = cudf.Series(offsets) if offsets is None: offsets = cudf.Series([0, len(vertices)]) if not isinstance(offsets, cudf.Series): raise TypeError( f"'offsets' must be either 'None', a list or a " f"cudf Series, got: {type(offsets)}" ) # vertices uses "external" vertex IDs, but since the graph has been # renumbered, the node ID must also be renumbered. if input_graph.renumbered: vertices = input_graph.lookup_internal_vertex_id(vertices) vertices_type = input_graph.edgelist.edgelist_df.dtypes[0] else: vertices_type = input_graph.input_df.dtypes[0] if isinstance(vertices, (cudf.Series, cudf.DataFrame)): vertices = dask_cudf.from_cudf( vertices, npartitions=min(input_graph._npartitions, len(vertices)) ) vertices = vertices.astype(vertices_type) vertices = get_distributed_data(vertices) wait(vertices) vertices = vertices.worker_to_parts do_expensive_check = False result = [ client.submit( _call_induced_subgraph, Comms.get_session_id(), input_graph._plc_graph[w], vertices[w][0], offsets, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) if len(ddf) == 0: return None, None wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) ddf = ddf.sort_values("labels") # extract offsets from segmented induced_subgraph dataframes offsets = ddf["labels"].value_counts().compute().sort_index() offsets = cudf.concat([cudf.Series(0), offsets]) offsets = ( dask_cudf.from_cudf( offsets, npartitions=min(input_graph._npartitions, len(vertices)) ) .cumsum() .astype(vertices_type) ) ddf = ddf.drop(columns="labels") if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "src") ddf = input_graph.unrenumber(ddf, "dst") return ddf, offsets
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/community/louvain.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from __future__ import annotations from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import dask from dask import delayed import cudf import cupy as cp import numpy from pylibcugraph import ResourceHandle from pylibcugraph import louvain as pylibcugraph_louvain from typing import Tuple, TYPE_CHECKING import warnings if TYPE_CHECKING: from cugraph import Graph def convert_to_cudf(result: cp.ndarray) -> Tuple[cudf.DataFrame, float]: """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_vertex, cupy_partition, modularity = result df = cudf.DataFrame() df["vertex"] = cupy_vertex df["partition"] = cupy_partition return df, modularity def _call_plc_louvain( sID: bytes, mg_graph_x, max_level: int, threshold: float, resolution: float, do_expensive_check: bool, ) -> Tuple[cp.ndarray, cp.ndarray, float]: return pylibcugraph_louvain( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, max_level=max_level, threshold=threshold, resolution=resolution, do_expensive_check=do_expensive_check, ) # FIXME: max_level should default to 100 once max_iter is removed def louvain( input_graph: Graph, max_level: int = None, max_iter: int = None, resolution: float = 1.0, threshold: float = 1e-7, ) -> Tuple[dask_cudf.DataFrame, float]: """ Compute the modularity optimizing partition of the input graph using the Louvain method It uses the Louvain method described in: VD Blondel, J-L Guillaume, R Lambiotte and E Lefebvre: Fast unfolding of community hierarchies in large networks, J Stat Mech P10008 (2008), http://arxiv.org/abs/0803.0476 Parameters ---------- G : cugraph.Graph The graph descriptor should contain the connectivity information and weights. The adjacency list will be computed if not already present. The current implementation only supports undirected graphs. max_level : integer, optional (default=100) This controls the maximum number of levels of the Louvain algorithm. When specified the algorithm will terminate after no more than the specified number of levels. No error occurs when the algorithm terminates early in this manner. max_iter : integer, optional (default=None) This parameter is deprecated in favor of max_level. Previously it was used to control the maximum number of levels of the Louvain algorithm. resolution: float, optional (default=1.0) Called gamma in the modularity formula, this changes the size of the communities. Higher resolutions lead to more smaller communities, lower resolutions lead to fewer larger communities. threshold: float, optional (default=1e-7) Modularity gain threshold for each level. If the gain of modularity between 2 levels of the algorithm is less than the given threshold then the algorithm stops and returns the resulting communities. Returns ------- parts : dask_cudf.DataFrame GPU data frame of size V containing two columns the vertex id and the partition id it is assigned to. ddf['vertex'] : cudf.Series Contains the vertex identifiers ddf['partition'] : cudf.Series Contains the partition assigned to the vertices modularity_score : float a floating point number containing the global modularity score of the partitioning. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(fetch=True) >>> parts = cugraph.louvain(G) """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") # FIXME: This max_iter logic and the max_level defaulting can be deleted # in favor of defaulting max_level in call once max_iter is deleted if max_iter: if max_level: raise ValueError( "max_iter is deprecated. Cannot specify both max_iter and max_level" ) warning_msg = ( "max_iter has been renamed max_level. Use of max_iter is " "deprecated and will no longer be supported in the next releases. " ) warnings.warn(warning_msg, FutureWarning) max_level = max_iter if max_level is None: max_level = 100 # Initialize dask client client = default_client() do_expensive_check = False result = [ client.submit( _call_plc_louvain, Comms.get_session_id(), input_graph._plc_graph[w], max_level, threshold, resolution, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) part_mod_score = [client.submit(convert_to_cudf, r) for r in result] wait(part_mod_score) vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] empty_df = cudf.DataFrame( { "vertex": numpy.empty(shape=0, dtype=vertex_dtype), "partition": numpy.empty(shape=0, dtype="int32"), } ) part_mod_score = [delayed(lambda x: x, nout=2)(r) for r in part_mod_score] ddf = dask_cudf.from_delayed( [r[0] for r in part_mod_score], meta=empty_df, verify_meta=False ).persist() mod_score = dask.array.from_delayed( part_mod_score[0][1], shape=(1,), dtype=float ).compute() wait(ddf) wait(mod_score) wait([r.release() for r in part_mod_score]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") return ddf, mod_score
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/community/__init__.py
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from .louvain import louvain from .triangle_count import triangle_count from .induced_subgraph import induced_subgraph from .leiden import leiden
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/community/leiden.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from __future__ import annotations from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import dask from dask import delayed import cudf from pylibcugraph import ResourceHandle from pylibcugraph import leiden as pylibcugraph_leiden import numpy import cupy as cp from typing import Tuple, TYPE_CHECKING if TYPE_CHECKING: from cugraph import Graph def convert_to_cudf(result: cp.ndarray) -> Tuple[cudf.DataFrame, float]: """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_vertex, cupy_partition, modularity = result df = cudf.DataFrame() df["vertex"] = cupy_vertex df["partition"] = cupy_partition return df, modularity def _call_plc_leiden( sID: bytes, mg_graph_x, max_iter: int, resolution: int, random_state: int, theta: int, do_expensive_check: bool, ) -> Tuple[cp.ndarray, cp.ndarray, float]: return pylibcugraph_leiden( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), random_state=random_state, graph=mg_graph_x, max_level=max_iter, resolution=resolution, theta=theta, do_expensive_check=do_expensive_check, ) def leiden( input_graph: Graph, max_iter: int = 100, resolution: int = 1.0, random_state: int = None, theta: int = 1.0, ) -> Tuple[dask_cudf.DataFrame, float]: """ Compute the modularity optimizing partition of the input graph using the Leiden method Traag, V. A., Waltman, L., & van Eck, N. J. (2019). From Louvain to Leiden: guaranteeing well-connected communities. Scientific reports, 9(1), 5233. doi: 10.1038/s41598-019-41695-z Parameters ---------- G : cugraph.Graph The graph descriptor should contain the connectivity information and weights. The adjacency list will be computed if not already present. The current implementation only supports undirected graphs. max_iter : integer, optional (default=100) This controls the maximum number of levels/iterations of the Leiden algorithm. When specified the algorithm will terminate after no more than the specified number of iterations. No error occurs when the algorithm terminates early in this manner. resolution: float, optional (default=1.0) Called gamma in the modularity formula, this changes the size of the communities. Higher resolutions lead to more smaller communities, lower resolutions lead to fewer larger communities. Defaults to 1. random_state: int, optional(default=None) Random state to use when generating samples. Optional argument, defaults to a hash of process id, time, and hostname. theta: float, optional (default=1.0) Called theta in the Leiden algorithm, this is used to scale modularity gain in Leiden refinement phase, to compute the probability of joining a random leiden community. Returns ------- parts : dask_cudf.DataFrame GPU data frame of size V containing two columns the vertex id and the partition id it is assigned to. ddf['vertex'] : cudf.Series Contains the vertex identifiers ddf['partition'] : cudf.Series Contains the partition assigned to the vertices modularity_score : float a floating point number containing the global modularity score of the partitioning. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(fetch=True) >>> parts, modularity_score = cugraph.leiden(G) """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") # Return a client if one has started client = default_client() do_expensive_check = False result = [ client.submit( _call_plc_leiden, Comms.get_session_id(), input_graph._plc_graph[w], max_iter, resolution, random_state, theta, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) part_mod_score = [client.submit(convert_to_cudf, r) for r in result] wait(part_mod_score) vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] empty_df = cudf.DataFrame( { "vertex": numpy.empty(shape=0, dtype=vertex_dtype), "partition": numpy.empty(shape=0, dtype="int32"), } ) part_mod_score = [delayed(lambda x: x, nout=2)(r) for r in part_mod_score] ddf = dask_cudf.from_delayed( [r[0] for r in part_mod_score], meta=empty_df, verify_meta=False ).persist() mod_score = dask.array.from_delayed( part_mod_score[0][1], shape=(1,), dtype=float ).compute() wait(ddf) wait(mod_score) wait([r.release() for r in part_mod_score]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") return ddf, mod_score
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/community/egonet.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from cugraph.dask.common.input_utils import get_distributed_data from pylibcugraph import ResourceHandle, ego_graph as pylibcugraph_ego_graph def _call_ego_graph( sID, mg_graph_x, n, radius, do_expensive_check, ): return pylibcugraph_ego_graph( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, source_vertices=n, radius=radius, do_expensive_check=do_expensive_check, ) def consolidate_results(df, offsets): """ Each rank returns its ego_graph dataframe with its corresponding offsets array. This is ideal if the user operates on distributed memory but when attempting to bring the result into a single machine, the ego_graph dataframes generated from each seed cannot be extracted using the offsets array. This function consolidate the final result by performing segmented copies. Returns: consolidated ego_graph dataframe """ for i in range(len(offsets) - 1): df_tmp = df[offsets[i] : offsets[i + 1]] df_tmp["labels"] = i if i == 0: df_consolidate = df_tmp else: df_consolidate = cudf.concat([df_consolidate, df_tmp]) return df_consolidate def convert_to_cudf(cp_arrays): cp_src, cp_dst, cp_weight, cp_offsets = cp_arrays df = cudf.DataFrame() df["src"] = cp_src df["dst"] = cp_dst if cp_weight is None: df["weight"] = None else: df["weight"] = cp_weight offsets = cudf.Series(cp_offsets) return consolidate_results(df, offsets) def ego_graph(input_graph, n, radius=1, center=True): """ Compute the induced subgraph of neighbors centered at node n, within a given radius. Parameters ---------- input_graph : cugraph.Graph Graph or matrix object, which should contain the connectivity information. Edge weights, if present, should be single or double precision floating point values. n : int, list or cudf Series or Dataframe, dask_cudf Series or DataFrame A node or a list or cudf.Series of nodes or a cudf.DataFrame if nodes are represented with multiple columns. If a cudf.DataFrame is provided, only the first row is taken as the node input. radius: integer, optional (default=1) Include all neighbors of distance<=radius from n. center: bool, optional Defaults to True. False is not supported Returns ------- ego_edge_lists : dask_cudf.DataFrame Distributed GPU data frame containing all induced sources identifiers, destination identifiers, edge weights seeds_offsets: dask_cudf.Series Distributed Series containing the starting offset in the returned edge list for each seed. """ # Initialize dask client client = default_client() if isinstance(n, (int, list)): n = cudf.Series(n) elif not isinstance( n, (cudf.Series, dask_cudf.Series, cudf.DataFrame, dask_cudf.DataFrame) ): raise TypeError( f"'n' must be either an integer or a list or a " f"cudf or dask_cudf Series or DataFrame, got: {type(n)}" ) # n uses "external" vertex IDs, but since the graph has been # renumbered, the node ID must also be renumbered. if input_graph.renumbered: n = input_graph.lookup_internal_vertex_id(n) n_type = input_graph.edgelist.edgelist_df.dtypes[0] else: n_type = input_graph.input_df.dtypes[0] if isinstance(n, (cudf.Series, cudf.DataFrame)): n = dask_cudf.from_cudf(n, npartitions=min(input_graph._npartitions, len(n))) n = n.astype(n_type) n = get_distributed_data(n) wait(n) n = n.worker_to_parts do_expensive_check = False result = [ client.submit( _call_ego_graph, Comms.get_session_id(), input_graph._plc_graph[w], n[w][0], radius, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) ddf = ddf.sort_values("labels") # extract offsets from segmented ego_graph dataframes offsets = ddf["labels"].value_counts().compute().sort_index() offsets = cudf.concat([cudf.Series(0), offsets]) offsets = ( dask_cudf.from_cudf(offsets, npartitions=min(input_graph._npartitions, len(n))) .cumsum() .astype(n_type) ) ddf = ddf.drop(columns="labels") if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "src") ddf = input_graph.unrenumber(ddf, "dst") return ddf, offsets
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/comms/comms_wrapper.pyx
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # cython: profile=False # distutils: language = c++ # cython: embedsignature = True # cython: language_level = 3 from pylibraft.common.handle cimport * from cugraph.dask.comms.comms cimport init_subcomm as c_init_subcomm def init_subcomms(handle, row_comm_size): cdef size_t handle_size_t = <size_t>handle.getHandle() handle_ = <handle_t*>handle_size_t c_init_subcomm(handle_[0], row_comm_size)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/comms/CMakeLists.txt
# ============================================================================= # Copyright (c) 2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except # in compliance with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software distributed under the License # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express # or implied. See the License for the specific language governing permissions and limitations under # the License. # ============================================================================= set(cython_sources comms_wrapper.pyx) set(linked_libraries cugraph::cugraph) rapids_cython_create_modules( CXX SOURCE_FILES "${cython_sources}" LINKED_LIBRARIES "${linked_libraries}" MODULE_PREFIX comms_ ASSOCIATED_TARGETS cugraph )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/comms/comms.py
# Copyright (c) 2018-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # FIXME: these raft imports break the library if ucx-py is # not available. They are necessary only when doing MG work. from cugraph.dask.common.read_utils import MissingUCXPy try: from raft_dask.common.comms import Comms as raftComms from raft_dask.common.comms import get_raft_comm_state except ImportError as err: # FIXME: Generalize since err.name is arr when # libnuma.so.1 is not available if err.name == "ucp" or err.name == "arr": raftComms = MissingUCXPy() get_raft_comm_state = MissingUCXPy() else: raise from pylibraft.common.handle import Handle from cugraph.dask.comms.comms_wrapper import init_subcomms as c_init_subcomms from dask.distributed import default_client, get_worker from cugraph.dask.common import read_utils import math __instance = None __default_handle = None __subcomm = None def __get_2D_div(ngpus): prows = int(math.sqrt(ngpus)) while ngpus % prows != 0: prows = prows - 1 return prows, int(ngpus / prows) def subcomm_init(prows, pcols, partition_type): sID = get_session_id() ngpus = get_n_workers() if prows is None and pcols is None: if partition_type == 1: pcols, prows = __get_2D_div(ngpus) else: prows, pcols = __get_2D_div(ngpus) else: if prows is not None and pcols is not None: if ngpus != prows * pcols: raise Exception( "prows*pcols should be equal to the\ number of processes" ) elif prows is not None: if ngpus % prows != 0: raise Exception( "prows must be a factor of the number\ of processes" ) pcols = int(ngpus / prows) elif pcols is not None: if ngpus % pcols != 0: raise Exception( "pcols must be a factor of the number\ of processes" ) prows = int(ngpus / pcols) client = default_client() client.run(_subcomm_init, sID, pcols) global __subcomm __subcomm = (prows, pcols, partition_type) def _subcomm_init(sID, partition_row_size, dask_worker=None): handle = get_handle(sID, dask_worker) c_init_subcomms(handle, partition_row_size) def initialize(comms=None, p2p=False, prows=None, pcols=None, partition_type=1): """ Initialize a communicator for multi-node/multi-gpu communications. It is expected to be called right after client initialization for running multi-GPU algorithms (this wraps raft comms that manages underlying NCCL and UCX comms handles across the workers of a Dask cluster). It is recommended to also call `destroy()` when the comms are no longer needed so the underlying resources can be cleaned up. Parameters ---------- comms : raft Comms, optional (default=None) A pre-initialized raft communicator. If provided, this is used for mnmg communications. If not provided, default comms are initialized as per client information. p2p : bool, optional (default=False) Initialize UCX endpoints if True. prows : int, optional (default=None) Specifies the number of rows when performing a 2D partitioning of the input graph. If specified, this must be a factor of the total number of parallel processes. When specified with pcols, prows*pcols should be equal to the total number of parallel processes. pcols : int, optional (default=None) Specifies the number of columns when performing a 2D partitioning of the input graph. If specified, this must be a factor of the total number of parallel processes. When specified with prows, prows*pcols should be equal to the total number of parallel processes. partition_type : int, optional (default=1) Valid values are currently 1 or any int other than 1. A value of 1 (the default) represents a partitioning resulting in prows*pcols partitions. A non-1 value currently results in a partitioning of p*pcols partitions, where p is the number of GPUs. Examples -------- >>> from dask.distributed import Client >>> from dask_cuda import LocalCUDACluster >>> import cugraph.dask.comms as Comms >>> cluster = LocalCUDACluster() >>> client = Client(cluster) >>> Comms.initialize(p2p=True) >>> # DO WORK HERE >>> # All done, clean up >>> Comms.destroy() >>> client.close() >>> cluster.close() """ global __instance if __instance is None: global __default_handle __default_handle = None if comms is None: # Initialize communicator if not p2p: raise Exception("Set p2p to True for running mnmg algorithms") __instance = raftComms(comms_p2p=p2p) __instance.init() # Initialize subcommunicator subcomm_init(prows, pcols, partition_type) else: __instance = comms else: raise Exception("Communicator is already initialized") def is_initialized(): """ Returns True if comms was initialized, False otherwise. """ global __instance if __instance is not None: return True else: return False def get_comms(): """ Returns raft Comms instance """ global __instance return __instance def get_workers(): """ Returns the workers in the Comms instance, or None if Comms is not initialized. """ if is_initialized(): global __instance return __instance.worker_addresses def get_session_id(): """ Returns the sessionId for finding sessionstate of workers, or None if Comms is not initialized. """ if is_initialized(): global __instance return __instance.sessionId def get_2D_partition(): """ Returns a tuple representing the 2D partition information: (prows, pcols, partition_type) """ global __subcomm if __subcomm is not None: return __subcomm def destroy(): """ Shuts down initialized comms and cleans up resources. """ global __instance if is_initialized(): __instance.destroy() __instance = None def get_default_handle(): """ Returns the default handle. This does not perform nccl initialization. """ global __default_handle if __default_handle is None: __default_handle = Handle() return __default_handle # Functions to be called from within workers def get_handle(sID, dask_worker=None): """ Returns the handle from within the worker using the sessionstate. """ if dask_worker is None: dask_worker = get_worker() sessionstate = get_raft_comm_state(sID, dask_worker) return sessionstate["handle"] def get_worker_id(sID, dask_worker=None): """ Returns the worker's sessionId from within the worker. """ if dask_worker is None: dask_worker = get_worker() sessionstate = get_raft_comm_state(sID, dask_worker) return sessionstate["wid"] # FIXME: There are several similar instances of utility functions for getting # the number of workers, including: # * get_n_workers() (from cugraph.dask.common.read_utils) # * len(get_visible_devices()) # * len(numba.cuda.gpus) # Consider consolidating these or emphasizing why different # functions/techniques are needed. def get_n_workers(sID=None, dask_worker=None): if sID is None: return read_utils.get_n_workers() else: if dask_worker is None: dask_worker = get_worker() sessionstate = get_raft_comm_state(sID, dask_worker) return sessionstate["nworkers"]
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/comms/comms.pxd
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # cython: profile=False # distutils: language = c++ # cython: embedsignature = True # cython: language_level = 3 from pylibraft.common.handle cimport * cdef extern from "cugraph/partition_manager.hpp" namespace "cugraph::partition_manager": cdef void init_subcomm(handle_t &handle, size_t row_comm_size)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/comms/__init__.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/sampling/random_walks.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from dask.distributed import wait, default_client import dask_cudf import cudf import operator as op from pylibcugraph import ResourceHandle from pylibcugraph import ( uniform_random_walks as pylibcugraph_uniform_random_walks, ) from cugraph.dask.comms import comms as Comms from cugraph.dask.common.input_utils import get_distributed_data def convert_to_cudf(cp_paths, number_map=None, is_vertex_paths=False): """ Creates cudf Series from cupy arrays from pylibcugraph wrapper """ if is_vertex_paths and len(cp_paths) > 0: if number_map.implementation.numbered: df_ = cudf.DataFrame() df_["vertex_paths"] = cp_paths df_ = number_map.unrenumber( df_, "vertex_paths", preserve_order=True ).compute() vertex_paths = cudf.Series(df_["vertex_paths"]).fillna(-1) return vertex_paths return cudf.Series(cp_paths) def _call_plc_uniform_random_walks(sID, mg_graph_x, st_x, max_depth): return pylibcugraph_uniform_random_walks( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), input_graph=mg_graph_x, start_vertices=st_x, max_length=max_depth, ) def random_walks( input_graph, random_walks_type="uniform", start_vertices=None, max_depth=None, use_padding=None, legacy_result_type=None, ): """ compute random walks for each nodes in 'start_vertices' and returns a padded result along with the maximum path length. Vertices with no outgoing edges will be padded with -1. parameters ---------- input_graph : cuGraph.Graph The graph can be either directed or undirected. random_walks_type : str, optional (default='uniform') Type of random walks: 'uniform', 'biased', 'node2vec'. Only 'uniform' random walks is currently supported start_vertices : int or list or cudf.Series or cudf.DataFrame A single node or a list or a cudf.Series of nodes from which to run the random walks. In case of multi-column vertices it should be a cudf.DataFrame max_depth : int The maximum depth of the random walks use_padding : bool This parameter is here for SG compatibility and ignored legacy_result_type : bool This parameter is here for SG compatibility and ignored Returns ------- vertex_paths : dask_cudf.Series or dask_cudf.DataFrame Series containing the vertices of edges/paths in the random walk. edge_weight_paths: dask_cudf.Series Series containing the edge weights of edges represented by the returned vertex_paths max_path_length : int The maximum path length """ if isinstance(start_vertices, int): start_vertices = [start_vertices] if isinstance(start_vertices, list): start_vertices = cudf.Series(start_vertices) # start_vertices uses "external" vertex IDs, but if the graph has been # renumbered, the start vertex IDs must also be renumbered. if input_graph.renumbered: # FIXME: This should match start_vertices type to the renumbered df type # but verify that. If not retrieve the type and cast it when creating # the dask_cudf from a cudf start_vertices = input_graph.lookup_internal_vertex_id(start_vertices).compute() start_vertices_type = input_graph.edgelist.edgelist_df.dtypes[0] else: # FIXME: Get the 'src' column names instead and retrieve the type start_vertices_type = input_graph.input_df.dtypes[0] start_vertices = dask_cudf.from_cudf( start_vertices, npartitions=min(input_graph._npartitions, len(start_vertices)) ) start_vertices = start_vertices.astype(start_vertices_type) start_vertices = get_distributed_data(start_vertices) wait(start_vertices) start_vertices = start_vertices.worker_to_parts client = default_client() result = [ client.submit( _call_plc_uniform_random_walks, Comms.get_session_id(), input_graph._plc_graph[w], start_vertices[w][0], max_depth, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) result_vertex_paths = [client.submit(op.getitem, f, 0) for f in result] result_edge_wgt_paths = [client.submit(op.getitem, f, 1) for f in result] max_path_length = [client.submit(op.getitem, f, 2) for f in result] cudf_vertex_paths = [ client.submit(convert_to_cudf, cp_vertex_paths, input_graph.renumber_map, True) for cp_vertex_paths in result_vertex_paths ] cudf_edge_wgt_paths = [ client.submit(convert_to_cudf, cp_edge_wgt_paths) for cp_edge_wgt_paths in result_edge_wgt_paths ] wait([cudf_vertex_paths, cudf_edge_wgt_paths]) max_path_length = max_path_length[0].result() ddf_vertex_paths = dask_cudf.from_delayed(cudf_vertex_paths).persist() ddf_edge_wgt_paths = dask_cudf.from_delayed(cudf_edge_wgt_paths).persist() wait([ddf_vertex_paths, ddf_edge_wgt_paths]) # Wait until the inactive futures are released wait( [ (r.release(), c_v.release(), c_e.release()) for r, c_v, c_e in zip(result, cudf_vertex_paths, cudf_edge_wgt_paths) ] ) return ddf_vertex_paths, ddf_edge_wgt_paths, max_path_length
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/sampling/uniform_neighbor_sample.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from __future__ import annotations import warnings import numpy from dask import delayed from dask.distributed import Lock, get_client, wait import dask_cudf import cudf import cupy as cp from pylibcugraph import ResourceHandle from pylibcugraph import uniform_neighbor_sample as pylibcugraph_uniform_neighbor_sample from cugraph.dask.comms import comms as Comms from cugraph.dask import get_n_workers from typing import Sequence, List, Union, Tuple from typing import TYPE_CHECKING from cugraph.dask.common.part_utils import ( get_persisted_df_worker_map, persist_dask_df_equal_parts_per_worker, ) if TYPE_CHECKING: from cugraph import Graph src_n = "sources" dst_n = "destinations" indices_n = "indices" weight_n = "weight" edge_id_n = "edge_id" edge_type_n = "edge_type" batch_id_n = "batch_id" offsets_n = "offsets" hop_id_n = "hop_id" map_n = "map" map_offsets_n = "renumber_map_offsets" start_col_name = "_START_" batch_col_name = "_BATCH_" def create_empty_df(indices_t, weight_t): df = cudf.DataFrame( { src_n: numpy.empty(shape=0, dtype=indices_t), dst_n: numpy.empty(shape=0, dtype=indices_t), indices_n: numpy.empty(shape=0, dtype=weight_t), } ) return df def create_empty_df_with_edge_props( indices_t, weight_t, return_offsets=False, renumber=False, use_legacy_names=True, include_hop_column=True, compression="COO", ): if compression != "COO": majors_name = "major_offsets" else: majors_name = src_n if use_legacy_names else "majors" minors_name = dst_n if use_legacy_names else "minors" if renumber: empty_df_renumber = cudf.DataFrame( { map_n: numpy.empty(shape=0, dtype=indices_t), map_offsets_n: numpy.empty(shape=0, dtype="int32"), } ) if return_offsets: df = cudf.DataFrame( { majors_name: numpy.empty(shape=0, dtype=indices_t), minors_name: numpy.empty(shape=0, dtype=indices_t), weight_n: numpy.empty(shape=0, dtype=weight_t), edge_id_n: numpy.empty(shape=0, dtype=indices_t), edge_type_n: numpy.empty(shape=0, dtype="int32"), } ) if include_hop_column: df[hop_id_n] = numpy.empty(shape=0, dtype="int32") empty_df_offsets = cudf.DataFrame( { offsets_n: numpy.empty(shape=0, dtype="int32"), batch_id_n: numpy.empty(shape=0, dtype="int32"), } ) if renumber: return df, empty_df_offsets, empty_df_renumber else: return df, empty_df_offsets else: df = cudf.DataFrame( { majors_name: numpy.empty(shape=0, dtype=indices_t), minors_name: numpy.empty(shape=0, dtype=indices_t), weight_n: numpy.empty(shape=0, dtype=weight_t), edge_id_n: numpy.empty(shape=0, dtype=indices_t), edge_type_n: numpy.empty(shape=0, dtype="int32"), batch_id_n: numpy.empty(shape=0, dtype="int32"), hop_id_n: numpy.empty(shape=0, dtype="int32"), } ) if renumber: return df, empty_df_renumber else: return df def __get_label_to_output_comm_rank(min_batch_id, max_batch_id, n_workers): num_batches = max_batch_id - min_batch_id + 1 num_batches = int(num_batches) z = cp.zeros(num_batches, dtype="int32") s = cp.array_split(cp.arange(num_batches), n_workers) for i, t in enumerate(s): z[t] = i return z def _call_plc_uniform_neighbor_sample( sID, mg_graph_x, st_x, keep_batches_together, n_workers, min_batch_id, max_batch_id, fanout_vals, with_replacement, weight_t, with_edge_properties, random_state=None, return_offsets=False, return_hops=True, prior_sources_behavior=None, deduplicate_sources=False, renumber=False, use_legacy_names=True, include_hop_column=True, compress_per_hop=False, compression="COO", ): st_x = st_x[0] start_list_x = st_x[start_col_name] batch_id_list_x = st_x[batch_col_name] if batch_col_name in st_x else None label_list = None label_to_output_comm_rank = None if keep_batches_together: label_list = cp.arange(min_batch_id, max_batch_id + 1, dtype="int32") label_to_output_comm_rank = __get_label_to_output_comm_rank( min_batch_id, max_batch_id, n_workers ) cupy_array_dict = pylibcugraph_uniform_neighbor_sample( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), input_graph=mg_graph_x, start_list=start_list_x, label_list=label_list, label_to_output_comm_rank=label_to_output_comm_rank, h_fan_out=fanout_vals, with_replacement=with_replacement, do_expensive_check=False, with_edge_properties=with_edge_properties, batch_id_list=batch_id_list_x, random_state=random_state, prior_sources_behavior=prior_sources_behavior, deduplicate_sources=deduplicate_sources, return_hops=return_hops, renumber=renumber, compression=compression, compress_per_hop=compress_per_hop, return_dict=True, ) # have to import here due to circular import issue from cugraph.sampling.sampling_utilities import ( sampling_results_from_cupy_array_dict, ) return sampling_results_from_cupy_array_dict( cupy_array_dict, weight_t, len(fanout_vals), with_edge_properties=with_edge_properties, return_offsets=return_offsets, renumber=renumber, use_legacy_names=use_legacy_names, include_hop_column=include_hop_column, ) def _mg_call_plc_uniform_neighbor_sample( client, session_id, input_graph, ddf, keep_batches_together, min_batch_id, max_batch_id, fanout_vals, with_replacement, weight_t, indices_t, with_edge_properties, random_state, return_offsets=False, return_hops=True, prior_sources_behavior=None, deduplicate_sources=False, renumber=False, use_legacy_names=True, include_hop_column=True, compress_per_hop=False, compression="COO", ): n_workers = None if keep_batches_together: n_workers = get_n_workers() if hasattr(min_batch_id, "compute"): min_batch_id = min_batch_id.compute() if hasattr(max_batch_id, "compute"): max_batch_id = max_batch_id.compute() result = [ client.submit( _call_plc_uniform_neighbor_sample, session_id, input_graph._plc_graph[w], starts, keep_batches_together, n_workers, min_batch_id, max_batch_id, fanout_vals, with_replacement, weight_t=weight_t, with_edge_properties=with_edge_properties, # FIXME accept and properly transmute a numpy/cupy random state. random_state=hash((random_state, w)), return_offsets=return_offsets, return_hops=return_hops, prior_sources_behavior=prior_sources_behavior, deduplicate_sources=deduplicate_sources, renumber=renumber, use_legacy_names=use_legacy_names, # remove in 23.12 include_hop_column=include_hop_column, # remove in 23.12 compress_per_hop=compress_per_hop, compression=compression, allow_other_workers=False, pure=False, ) for w, starts in ddf.items() ] del ddf empty_df = ( create_empty_df_with_edge_props( indices_t, weight_t, return_offsets=return_offsets, renumber=renumber, use_legacy_names=use_legacy_names, compression=compression, include_hop_column=include_hop_column, ) if with_edge_properties else create_empty_df(indices_t, weight_t) ) if not isinstance(empty_df, (list, tuple)): empty_df = [empty_df] wait(result) nout = 1 if return_offsets: nout += 1 if renumber: nout += 1 result_split = [delayed(lambda x: x, nout=nout)(r) for r in result] ddf = dask_cudf.from_delayed( [r[0] for r in result_split], meta=empty_df[0], verify_meta=False ).persist() return_dfs = [ddf] if return_offsets: ddf_offsets = dask_cudf.from_delayed( [r[1] for r in result_split], meta=empty_df[1], verify_meta=False ).persist() return_dfs.append(ddf_offsets) if renumber: ddf_renumber = dask_cudf.from_delayed( [r[-1] for r in result_split], meta=empty_df[-1], verify_meta=False ).persist() return_dfs.append(ddf_renumber) wait(return_dfs) wait([r.release() for r in result_split]) wait([r.release() for r in result]) del result if len(return_dfs) == 1: return return_dfs[0] else: return tuple(return_dfs) def uniform_neighbor_sample( input_graph: Graph, start_list: Sequence, fanout_vals: List[int], *, with_replacement: bool = True, with_edge_properties: bool = False, # deprecated with_batch_ids: bool = False, keep_batches_together=False, min_batch_id=None, max_batch_id=None, random_state: int = None, return_offsets: bool = False, return_hops: bool = True, include_hop_column: bool = True, # deprecated prior_sources_behavior: str = None, deduplicate_sources: bool = False, renumber: bool = False, use_legacy_names=True, # deprecated compress_per_hop=False, compression="COO", _multiple_clients: bool = False, ) -> Union[dask_cudf.DataFrame, Tuple[dask_cudf.DataFrame, dask_cudf.DataFrame]]: """ Does neighborhood sampling, which samples nodes from a graph based on the current node's neighbors, with a corresponding fanout value at each hop. Parameters ---------- input_graph : cugraph.Graph cuGraph graph, which contains connectivity information as dask cudf edge list dataframe start_list : int, list, cudf.Series, or dask_cudf.Series (int32 or int64) a list of starting vertices for sampling fanout_vals : list List of branching out (fan-out) degrees per starting vertex for each hop level. with_replacement: bool, optional (default=True) Flag to specify if the random sampling is done with replacement with_edge_properties: bool, optional (default=False) Deprecated. Flag to specify whether to return edge properties (weight, edge id, edge type, batch id, hop id) with the sampled edges. with_batch_ids: bool, optional (default=False) Flag to specify whether batch ids are present in the start_list keep_batches_together: bool (optional, default=False) If True, will ensure that the returned samples for each batch are on the same partition. min_batch_id: int (optional, default=None) Required for the keep_batches_together option. The minimum batch id. max_batch_id: int (optional, default=None) Required for the keep_batches_together option. The maximum batch id. random_state: int, optional Random seed to use when making sampling calls. return_offsets: bool, optional (default=False) Whether to return the sampling results with batch ids included as one dataframe, or to instead return two dataframes, one with sampling results and one with batch ids and their start offsets per rank. return_hops: bool, optional (default=True) Whether to return the sampling results with hop ids corresponding to the hop where the edge appeared. Defaults to True. include_hop_column: bool, optional (default=True) Deprecated. Defaults to True. If True, will include the hop column even if return_offsets is True. This option will be removed in release 23.12. prior_sources_behavior: str (Optional) Options are "carryover", and "exclude". Default will leave the source list as-is. Carryover will carry over sources from previous hops to the current hop. Exclude will exclude sources from previous hops from reappearing as sources in future hops. deduplicate_sources: bool, optional (default=False) Whether to first deduplicate the list of possible sources from the previous destinations before performing next hop. renumber: bool, optional (default=False) Whether to renumber on a per-batch basis. If True, will return the renumber map and renumber map offsets as an additional dataframe. use_legacy_names: bool, optional (default=True) Whether to use the legacy column names (sources, destinations). If True, will use "sources" and "destinations" as the column names. If False, will use "majors" and "minors" as the column names. Deprecated. Will be removed in release 23.12 in favor of always using the new names "majors" and "minors". compress_per_hop: bool, optional (default=False) Whether to compress globally (default), or to produce a separate compressed edgelist per hop. compression: str, optional (default=COO) Sets the compression type for the output minibatches. Valid options are COO (default), CSR, CSC, DCSR, and DCSC. _multiple_clients: bool, optional (default=False) internal flag to ensure sampling works with multiple dask clients set to True to prevent hangs in multi-client environment Returns ------- result : dask_cudf.DataFrame or Tuple[dask_cudf.DataFrame, dask_cudf.DataFrame] GPU distributed data frame containing several dask_cudf.Series If with_edge_properties=True: ddf['sources']: dask_cudf.Series Contains the source vertices from the sampling result ddf['destinations']: dask_cudf.Series Contains the destination vertices from the sampling result ddf['indices']: dask_cudf.Series Contains the indices from the sampling result for path reconstruction If with_edge_properties=False: If return_offsets=False: df['sources']: dask_cudf.Series Contains the source vertices from the sampling result df['destinations']: dask_cudf.Series Contains the destination vertices from the sampling result df['edge_weight']: dask_cudf.Series Contains the edge weights from the sampling result df['edge_id']: dask_cudf.Series Contains the edge ids from the sampling result df['edge_type']: dask_cudf.Series Contains the edge types from the sampling result df['batch_id']: dask_cudf.Series Contains the batch ids from the sampling result df['hop_id']: dask_cudf.Series Contains the hop ids from the sampling result If renumber=True: (adds the following dataframe) renumber_df['map']: dask_cudf.Series Contains the renumber maps for each batch renumber_df['offsets']: dask_cudf.Series Contains the batch offsets for the renumber maps If return_offsets=True: df['sources']: dask_cudf.Series Contains the source vertices from the sampling result df['destinations']: dask_cudf.Series Contains the destination vertices from the sampling result df['edge_weight']: dask_cudf.Series Contains the edge weights from the sampling result df['edge_id']: dask_cudf.Series Contains the edge ids from the sampling result df['edge_type']: dask_cudf.Series Contains the edge types from the sampling result df['hop_id']: dask_cudf.Series Contains the hop ids from the sampling result offsets_df['batch_id']: dask_cudf.Series Contains the batch ids from the sampling result offsets_df['offsets']: dask_cudf.Series Contains the offsets of each batch in the sampling result If renumber=True: (adds the following dataframe) renumber_df['map']: dask_cudf.Series Contains the renumber maps for each batch renumber_df['offsets']: dask_cudf.Series Contains the batch offsets for the renumber maps """ if compression not in ["COO", "CSR", "CSC", "DCSR", "DCSC"]: raise ValueError("compression must be one of COO, CSR, CSC, DCSR, or DCSC") if with_edge_properties: warning_msg = ( "The with_edge_properties flag is deprecated" " and will be removed in the next release." ) warnings.warn(warning_msg, FutureWarning) if ( (compression != "COO") and (not compress_per_hop) and prior_sources_behavior != "exclude" ): raise ValueError( "hop-agnostic compression is only supported with" " the exclude prior sources behavior due to limitations " "of the libcugraph C++ API" ) if compress_per_hop and prior_sources_behavior != "carryover": raise ValueError( "Compressing the edgelist per hop is only supported " "with the carryover prior sources behavior due to limitations" " of the libcugraph C++ API" ) if include_hop_column: warning_msg = ( "The include_hop_column flag is deprecated and will be" " removed in the next release in favor of always " "excluding the hop column when return_offsets is True" ) warnings.warn(warning_msg, FutureWarning) if compression != "COO": raise ValueError( "Including the hop id column is only supported with COO compression." ) if isinstance(start_list, int): start_list = [start_list] if isinstance(start_list, list): start_list = cudf.Series( start_list, dtype=input_graph.edgelist.edgelist_df[ input_graph.renumber_map.renumbered_src_col_name ].dtype, ) elif with_edge_properties and not with_batch_ids: if isinstance(start_list, (cudf.DataFrame, dask_cudf.DataFrame)): raise ValueError("expected 1d input for start list without batch ids") start_list = start_list.to_frame() if isinstance(start_list, dask_cudf.DataFrame): start_list = start_list.map_partitions( lambda df: df.assign( **{batch_id_n: cudf.Series(cp.zeros(len(df), dtype="int32"))} ) ).persist() else: start_list = start_list.reset_index(drop=True).assign( **{batch_id_n: cudf.Series(cp.zeros(len(start_list), dtype="int32"))} ) if keep_batches_together and min_batch_id is None: raise ValueError( "must provide min_batch_id if using keep_batches_together option" ) if keep_batches_together and max_batch_id is None: raise ValueError( "must provide max_batch_id if using keep_batches_together option" ) if renumber and not keep_batches_together: raise ValueError( "mg uniform_neighbor_sample requires that keep_batches_together=True " "when performing renumbering." ) # fanout_vals must be passed to pylibcugraph as a host array if isinstance(fanout_vals, numpy.ndarray): fanout_vals = fanout_vals.astype("int32") elif isinstance(fanout_vals, list): fanout_vals = numpy.asarray(fanout_vals, dtype="int32") elif isinstance(fanout_vals, cp.ndarray): fanout_vals = fanout_vals.get().astype("int32") elif isinstance(fanout_vals, cudf.Series): fanout_vals = fanout_vals.values_host.astype("int32") else: raise TypeError("fanout_vals must be a sequence, " f"got: {type(fanout_vals)}") if "value" in input_graph.edgelist.edgelist_df: weight_t = input_graph.edgelist.edgelist_df["value"].dtype else: weight_t = "float32" if "_SRC_" in input_graph.edgelist.edgelist_df: indices_t = input_graph.edgelist.edgelist_df["_SRC_"].dtype elif src_n in input_graph.edgelist.edgelist_df: indices_t = input_graph.edgelist.edgelist_df[src_n].dtype else: indices_t = numpy.int32 if isinstance(start_list, (cudf.Series, dask_cudf.Series)): start_list = start_list.rename(start_col_name) ddf = start_list.to_frame() else: ddf = start_list columns = ddf.columns ddf = ddf.rename( columns={columns[0]: start_col_name, columns[-1]: batch_col_name} ) if input_graph.renumbered: ddf = input_graph.lookup_internal_vertex_id(ddf, column_name=start_col_name) client = get_client() session_id = Comms.get_session_id() n_workers = get_n_workers() if isinstance(ddf, cudf.DataFrame): ddf = dask_cudf.from_cudf(ddf, npartitions=n_workers) ddf = ddf.repartition(npartitions=n_workers) ddf = persist_dask_df_equal_parts_per_worker(ddf, client) ddf = get_persisted_df_worker_map(ddf, client) sample_call_kwargs = { "client": client, "session_id": session_id, "input_graph": input_graph, "ddf": ddf, "keep_batches_together": keep_batches_together, "min_batch_id": min_batch_id, "max_batch_id": max_batch_id, "fanout_vals": fanout_vals, "with_replacement": with_replacement, "weight_t": weight_t, "indices_t": indices_t, "with_edge_properties": with_edge_properties, "random_state": random_state, "return_offsets": return_offsets, "return_hops": return_hops, "prior_sources_behavior": prior_sources_behavior, "deduplicate_sources": deduplicate_sources, "renumber": renumber, "use_legacy_names": use_legacy_names, "include_hop_column": include_hop_column, "compress_per_hop": compress_per_hop, "compression": compression, } if _multiple_clients: # Distributed centralized lock to allow # two disconnected processes (clients) to coordinate a lock # https://docs.dask.org/en/stable/futures.html?highlight=lock#distributed.Lock lock = Lock("plc_graph_access") if lock.acquire(timeout=100): try: ddf = _mg_call_plc_uniform_neighbor_sample(**sample_call_kwargs) finally: lock.release() else: raise RuntimeError( "Failed to acquire lock(plc_graph_access) while trying to sampling" ) else: ddf = _mg_call_plc_uniform_neighbor_sample(**sample_call_kwargs) if return_offsets: if renumber: ddf, offsets_df, renumber_df = ddf else: ddf, offsets_ddf = ddf else: if renumber: ddf, renumber_df = ddf if input_graph.renumbered and not renumber: if use_legacy_names: ddf = input_graph.unrenumber(ddf, "sources", preserve_order=True) ddf = input_graph.unrenumber(ddf, "destinations", preserve_order=True) else: ddf = input_graph.unrenumber(ddf, "majors", preserve_order=True) ddf = input_graph.unrenumber(ddf, "minors", preserve_order=True) if return_offsets: if renumber: return ddf, offsets_df, renumber_df else: return ddf, offsets_ddf if renumber: return ddf, renumber_df return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/sampling/__init__.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/cores/core_number.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from pylibcugraph import ResourceHandle, core_number as pylibcugraph_core_number def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_vertices, cupy_core_number = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertices df["core_number"] = cupy_core_number return df def _call_plc_core_number(sID, mg_graph_x, dt_x, do_expensive_check): return pylibcugraph_core_number( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, degree_type=dt_x, do_expensive_check=do_expensive_check, ) def core_number(input_graph, degree_type="bidirectional"): """ Compute the core numbers for the nodes of the graph G. A k-core of a graph is a maximal subgraph that contains nodes of degree k or more. A node has a core number of k if it belongs a k-core but not to k+1-core. This call does not support a graph with self-loops and parallel edges. Parameters ---------- input_graph : cugraph.graph cuGraph graph descriptor, should contain the connectivity information, (edge weights are not used in this algorithm). The current implementation only supports undirected graphs. degree_type: str, (default="bidirectional") This option determines if the core number computation should be based on input, output, or both directed edges, with valid values being "incoming", "outgoing", and "bidirectional" respectively. Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing 2 dask_cudf.Series ddf['vertex']: dask_cudf.Series Contains the core number vertices ddf['core_number']: dask_cudf.Series Contains the core number of vertices """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") if degree_type not in ["incoming", "outgoing", "bidirectional"]: raise ValueError( f"'degree_type' must be either incoming, " f"outgoing or bidirectional, got: {degree_type}" ) # Initialize dask client client = default_client() do_expensive_check = False result = [ client.submit( _call_plc_core_number, Comms.get_session_id(), input_graph._plc_graph[w], degree_type, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/cores/k_core.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms from cugraph.dask.common.input_utils import get_distributed_data import dask_cudf import cudf import cugraph.dask as dcg from pylibcugraph import ResourceHandle, k_core as pylibcugraph_k_core def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_src_vertices, cupy_dst_vertices, cupy_weights = cp_arrays df = cudf.DataFrame() df["src"] = cupy_src_vertices df["dst"] = cupy_dst_vertices df["weights"] = cupy_weights return df def _call_plc_k_core(sID, mg_graph_x, k, degree_type, core_result, do_expensive_check): return pylibcugraph_k_core( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, k=k, degree_type=degree_type, core_result=core_result, do_expensive_check=do_expensive_check, ) def k_core(input_graph, k=None, core_number=None, degree_type="bidirectional"): """ Compute the k-core of the graph G based on the out degree of its nodes. A k-core of a graph is a maximal subgraph that contains nodes of degree k or more. This call does not support a graph with self-loops and parallel edges. Parameters ---------- input_graph : cuGraph.Graph cuGraph graph descriptor with connectivity information. The graph should contain undirected edges where undirected edges are represented as directed edges in both directions. While this graph can contain edge weights, they don't participate in the calculation of the k-core. The current implementation only supports undirected graphs. k : int, optional (default=None) Order of the core. This value must not be negative. If set to None, the main core is returned. degree_type: str (default="bidirectional") This option determines if the core number computation should be based on input, output, or both directed edges, with valid values being "incoming", "outgoing", and "bidirectional" respectively. core_number : cudf.DataFrame or das_cudf.DataFrame, optional (default=None) Precomputed core number of the nodes of the graph G containing two cudf.Series of size V: the vertex identifiers and the corresponding core number values. If set to None, the core numbers of the nodes are calculated internally. core_number['vertex'] : cudf.Series or dask_cudf.Series Contains the vertex identifiers core_number['values'] : cudf.Series or dask_cudf.Series Contains the core number of vertices Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing the K Core of the input graph ddf['src']: dask_cudf.Series Contains sources of the K Core ddf['dst']: dask_cudf.Series Contains destinations of the K Core and/or ddf['weights']: dask_cudf.Series Contains weights of the K Core Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=False) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst', ... edge_attr='value') >>> KCore_df = dcg.k_core(dg) """ if degree_type not in ["incoming", "outgoing", "bidirectional"]: raise ValueError( f"'degree_type' must be either incoming, " f"outgoing or bidirectional, got: {degree_type}" ) if input_graph.is_directed(): raise ValueError("input graph must be undirected") if core_number is None: core_number = dcg.core_number(input_graph) if input_graph.renumbered is True: if len(input_graph.renumber_map.implementation.col_names) > 1: cols = core_number.columns[:-1].to_list() else: cols = "vertex" core_number = input_graph.add_internal_vertex_id( core_number, "vertex", cols ) if not isinstance(core_number, dask_cudf.DataFrame): if isinstance(core_number, cudf.DataFrame): # convert to dask_cudf in order to distribute the edges core_number = dask_cudf.from_cudf(core_number, input_graph._npartitions) else: raise TypeError( f"'core_number' must be either None or of" f"type cudf/dask_cudf, got: {type(core_number)}" ) core_number = core_number.rename(columns={"core_number": "values"}) if k is None: k = core_number["values"].max().compute() core_number = get_distributed_data(core_number) wait(core_number) core_number = core_number.worker_to_parts client = default_client() do_expensive_check = False result = [ client.submit( _call_plc_k_core, Comms.get_session_id(), input_graph._plc_graph[w], k, degree_type, core_number[w][0], do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # FIXME: Dask doesn't always release it fast enough. # For instance if the algo is run several times with # the same PLC graph, the current iteration might try to cache # the past iteration's futures and this can cause a hang if some # of those futures get released midway del result del cudf_result if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "src") ddf = input_graph.unrenumber(ddf, "dst") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/cores/__init__.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from .core_number import core_number from .k_core import k_core
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/centrality/eigenvector_centrality.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client from pylibcugraph import ( eigenvector_centrality as pylib_eigen, ResourceHandle, ) import cugraph.dask.comms.comms as Comms import dask_cudf import cudf import warnings def _call_plc_eigenvector_centrality( sID, mg_graph_x, max_iterations, epsilon, do_expensive_check, ): return pylib_eigen( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, epsilon=epsilon, max_iterations=max_iterations, do_expensive_check=do_expensive_check, ) def convert_to_cudf(cp_arrays): """ create a cudf DataFrame from cupy arrays """ cupy_vertices, cupy_values = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertices df["eigenvector_centrality"] = cupy_values return df def eigenvector_centrality(input_graph, max_iter=100, tol=1.0e-6): """ Compute the eigenvector centrality for a graph G. Eigenvector centrality computes the centrality for a node based on the centrality of its neighbors. The eigenvector centrality for node i is the i-th element of the vector x defined by the eigenvector equation. Parameters ---------- input_graph : cuGraph.Graph or networkx.Graph cuGraph graph descriptor with connectivity information. The graph can contain either directed or undirected edges. max_iter : int, optional (default=100) The maximum number of iterations before an answer is returned. This can be used to limit the execution time and do an early exit before the solver reaches the convergence tolerance. tol : float, optional (default=1e-6) Set the tolerance the approximation, this parameter should be a small magnitude value. The lower the tolerance the better the approximation. If this value is 0.0f, cuGraph will use the default value which is 1.0e-6. Setting too small a tolerance can lead to non-convergence due to numerical roundoff. Usually values between 1e-2 and 1e-6 are acceptable. normalized : not supported If True normalize the resulting eigenvector centrality values Returns ------- df : dask_cudf.DataFrame GPU data frame containing two cudf.Series of size V: the vertex identifiers and the corresponding eigenvector centrality values. df['vertex'] : cudf.Series Contains the vertex identifiers df['eigenvector_centrality'] : cudf.Series Contains the eigenvector centrality of vertices Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph() >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst', ... edge_attr='value') >>> ec = dcg.eigenvector_centrality(dg) """ client = default_client() if input_graph.store_transposed is False: warning_msg = ( "Eigenvector centrality expects the 'store_transposed' " "flag to be set to 'True' for optimal performance " "during the graph creation" ) warnings.warn(warning_msg, UserWarning) # FIXME: should we add this parameter as an option? do_expensive_check = False cupy_result = [ client.submit( _call_plc_eigenvector_centrality, Comms.get_session_id(), input_graph._plc_graph[w], max_iter, tol, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(cupy_result) cudf_result = [ client.submit( convert_to_cudf, cp_arrays, workers=client.who_has(cp_arrays)[cp_arrays.key] ) for cp_arrays in cupy_result ] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(cupy_result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/centrality/betweenness_centrality.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, get_client from pylibcugraph import ( ResourceHandle, betweenness_centrality as pylibcugraph_betweenness_centrality, edge_betweenness_centrality as pylibcugraph_edge_betweenness_centrality, ) import cugraph.dask.comms.comms as Comms from cugraph.dask.common.input_utils import get_distributed_data import dask_cudf import cudf import cupy as cp import warnings import dask from typing import Union def convert_to_cudf(cp_arrays: cp.ndarray, edge_bc: bool) -> cudf.DataFrame: """ create a cudf DataFrame from cupy arrays """ df = cudf.DataFrame() if edge_bc: cupy_src_vertices, cupy_dst_vertices, cupy_values, cupy_edge_ids = cp_arrays df["src"] = cupy_src_vertices df["dst"] = cupy_dst_vertices df["betweenness_centrality"] = cupy_values if cupy_edge_ids is not None: df["edge_id"] = cupy_edge_ids else: cupy_vertices, cupy_values = cp_arrays df["vertex"] = cupy_vertices df["betweenness_centrality"] = cupy_values return df def _call_plc_betweenness_centrality( mg_graph_x, sID: bytes, k: Union[int, cudf.Series], random_state: int, normalized: bool, endpoints: bool, do_expensive_check: bool, edge_bc: bool, ) -> cudf.DataFrame: if edge_bc: cp_arrays = pylibcugraph_edge_betweenness_centrality( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, k=k, random_state=random_state, normalized=normalized, do_expensive_check=do_expensive_check, ) else: cp_arrays = pylibcugraph_betweenness_centrality( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, k=k, random_state=random_state, normalized=normalized, include_endpoints=endpoints, do_expensive_check=do_expensive_check, ) return convert_to_cudf(cp_arrays, edge_bc) def _mg_call_plc_betweenness_centrality( input_graph, client: dask.distributed.client.Client, sID: bytes, k: dict, random_state: int, normalized: bool, do_expensive_check: bool, endpoints: bool = False, edge_bc: bool = False, ) -> dask_cudf.DataFrame: result = [ client.submit( _call_plc_betweenness_centrality, input_graph._plc_graph[w], sID, k if isinstance(k, (int, type(None))) else k[w][0], hash((random_state, i)), normalized, endpoints, do_expensive_check, edge_bc, workers=[w], allow_other_workers=False, pure=False, ) for i, w in enumerate(Comms.get_workers()) ] wait(result) ddf = dask_cudf.from_delayed(result, verify_meta=False).persist() wait(ddf) wait([r.release() for r in result]) return ddf def betweenness_centrality( input_graph, k: Union[ int, list, cudf.Series, cudf.DataFrame, dask_cudf.Series, dask_cudf.DataFrame ] = None, normalized: bool = True, weight: cudf.DataFrame = None, endpoints: bool = False, random_state: int = None, ) -> dask_cudf.DataFrame: """ Compute the betweenness centrality for all vertices of the graph G. Betweenness centrality is a measure of the number of shortest paths that pass through a vertex. A vertex with a high betweenness centrality score has more paths passing through it and is therefore believed to be more important. To improve performance. rather than doing an all-pair shortest path, a sample of k starting vertices can be used. CuGraph does not currently support 'weight' parameters. Parameters ---------- input_graph: cuGraph.Graph The graph can be either directed (Graph(directed=True)) or undirected. The current implementation uses a parallel variation of the Brandes Algorithm (2001) to compute exact or approximate betweenness. If weights are provided in the edgelist, they will not be used. k : int, list or (dask)cudf object or None, optional (default=None) If k is not None, use k node samples to estimate betweenness. Higher values give better approximation. If k is either a list, a cudf DataFrame, or a dask_cudf DataFrame, then its contents are assumed to be vertex identifiers to be used for estimation. If k is None (the default), all the vertices are used to estimate betweenness. Vertices obtained through sampling or defined as a list will be used as sources for traversals inside the algorithm. normalized : bool, optional (default=True) If True, normalize the resulting betweenness centrality values by __2 / ((n - 1) * (n - 2))__ for undirected Graphs, and __1 / ((n - 1) * (n - 2))__ for directed Graphs where n is the number of nodes in G. Normalization will ensure that values are in [0, 1], this normalization scales for the highest possible value where one node is crossed by every single shortest path. weight : (dask)cudf.DataFrame, optional (default=None) Specifies the weights to be used for each edge. Should contain a mapping between edges and weights. (Not Supported) endpoints : bool, optional (default=False) If true, include the endpoints in the shortest path counts. random_state : int, optional (default=None) if k is specified and k is an integer, use random_state to initialize the random number generator. Using None defaults to a hash of process id, time, and hostname If k is either None or list or cudf objects: random_state parameter is ignored. Returns ------- betweenness_centrality : dask_cudf.DataFrame GPU distributed data frame containing two dask_cudf.Series of size V: the vertex identifiers and the corresponding betweenness centrality values. ddf['vertex'] : dask_cudf.Series Contains the vertex identifiers ddf['betweenness_centrality'] : dask_cudf.Series Contains the betweenness centrality of vertices Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst') >>> pr = dcg.betweenness_centrality(dg) """ if input_graph.store_transposed is True: warning_msg = ( "Betweenness centrality expects the 'store_transposed' flag " "to be set to 'False' for optimal performance during " "the graph creation" ) warnings.warn(warning_msg, UserWarning) if weight is not None: raise NotImplementedError( "weighted implementation of betweenness " "centrality not currently supported" ) if not isinstance(k, (dask_cudf.DataFrame, dask_cudf.Series)): if isinstance(k, (cudf.DataFrame, cudf.Series, list)): if isinstance(k, list): k_dtype = input_graph.nodes().dtype k = cudf.Series(k, dtype=k_dtype) if isinstance(k, (cudf.Series, cudf.DataFrame)): splits = cp.array_split(cp.arange(len(k)), len(Comms.get_workers())) k = {w: [k.iloc[splits[i]]] for i, w in enumerate(Comms.get_workers())} else: if k is not None: k = get_distributed_data(k) wait(k) k = k.worker_to_parts if input_graph.renumbered: if isinstance(k, dask_cudf.DataFrame): tmp_col_names = k.columns elif isinstance(k, dask_cudf.Series): tmp_col_names = None if isinstance(k, (dask_cudf.DataFrame, dask_cudf.Series)): k = input_graph.lookup_internal_vertex_id(k, tmp_col_names) # FIXME: should we add this parameter as an option? do_expensive_check = False client = get_client() ddf = _mg_call_plc_betweenness_centrality( input_graph=input_graph, client=client, sID=Comms.get_session_id(), k=k, random_state=random_state, normalized=normalized, endpoints=endpoints, do_expensive_check=do_expensive_check, ) if input_graph.renumbered: return input_graph.unrenumber(ddf, "vertex") return ddf def edge_betweenness_centrality( input_graph, k: Union[ int, list, cudf.Series, cudf.DataFrame, dask_cudf.Series, dask_cudf.DataFrame ] = None, normalized: bool = True, weight: cudf.DataFrame = None, random_state: int = None, ) -> dask_cudf.DataFrame: """ Compute the edge betweenness centrality for all edges of the graph G. Betweenness centrality is a measure of the number of shortest paths that pass over an edge. An edge with a high betweenness centrality score has more paths passing over it and is therefore believed to be more important. To improve performance. rather than doing an all-pair shortest path, a sample of k starting vertices can be used. CuGraph does not currently support the 'weight' parameter. Parameters ---------- input_graph: cuGraph.Graph The graph can be either directed (Graph(directed=True)) or undirected. The current implementation uses a parallel variation of the Brandes Algorithm (2001) to compute exact or approximate betweenness. If weights are provided in the edgelist, they will not be used. k : int, list or (dask)cudf object or None, optional (default=None) If k is not None, use k node samples to estimate betweenness. Higher values give better approximation. If k is either a list, a cudf DataFrame, or a dask_cudf DataFrame, then its contents are assumed to be vertex identifiers to be used for estimation. If k is None (the default), all the vertices are used to estimate betweenness. Vertices obtained through sampling or defined as a list will be used as sources for traversals inside the algorithm. normalized : bool, optional (default=True) If True, normalize the resulting betweenness centrality values by __2 / (n * (n - 1))__ for undirected Graphs, and __1 / (n * (n - 1))__ for directed Graphs where n is the number of nodes in G. Normalization will ensure that values are in [0, 1], this normalization scales for the highest possible value where one edge is crossed by every single shortest path. weight : (dask)cudf.DataFrame, optional (default=None) Specifies the weights to be used for each edge. Should contain a mapping between edges and weights. (Not Supported) random_state : int, optional (default=None) if k is specified and k is an integer, use random_state to initialize the random number generator. Using None defaults to a hash of process id, time, and hostname If k is either None or list or cudf objects: random_state parameter is ignored. Returns ------- betweenness_centrality : dask_cudf.DataFrame GPU distributed data frame containing two dask_cudf.Series of size V: the vertex identifiers and the corresponding betweenness centrality values. ddf['src'] : dask_cudf.Series Contains the vertex identifiers of the source of each edge ddf['dst'] : dask_cudf.Series Contains the vertex identifiers of the destination of each edge ddf['betweenness_centrality'] : dask_cudf.Series Contains the betweenness centrality of edges ddf["edge_id"] : dask_cudf.Series Contains the edge ids of edges if present. Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst') >>> pr = dcg.edge_betweenness_centrality(dg) """ if input_graph.store_transposed is True: warning_msg = ( "Betweenness centrality expects the 'store_transposed' flag " "to be set to 'False' for optimal performance during " "the graph creation" ) warnings.warn(warning_msg, UserWarning) if weight is not None: raise NotImplementedError( "weighted implementation of edge betweenness " "centrality not currently supported" ) if not isinstance(k, (dask_cudf.DataFrame, dask_cudf.Series)): if isinstance(k, (cudf.DataFrame, cudf.Series, list)): if isinstance(k, list): k_dtype = input_graph.nodes().dtype k = cudf.Series(k, dtype=k_dtype) if isinstance(k, (cudf.Series, cudf.DataFrame)): splits = cp.array_split(cp.arange(len(k)), len(Comms.get_workers())) k = {w: [k.iloc[splits[i]]] for i, w in enumerate(Comms.get_workers())} else: if k is not None: k = get_distributed_data(k) wait(k) k = k.worker_to_parts if input_graph.renumbered: if isinstance(k, dask_cudf.DataFrame): tmp_col_names = k.columns elif isinstance(k, dask_cudf.Series): tmp_col_names = None if isinstance(k, (dask_cudf.DataFrame, dask_cudf.Series)): k = input_graph.lookup_internal_vertex_id(k, tmp_col_names) # FIXME: should we add this parameter as an option? do_expensive_check = False client = get_client() ddf = _mg_call_plc_betweenness_centrality( input_graph=input_graph, client=client, sID=Comms.get_session_id(), k=k, random_state=random_state, normalized=normalized, do_expensive_check=do_expensive_check, edge_bc=True, ) if input_graph.renumbered: return input_graph.unrenumber(ddf, "vertex") if input_graph.is_directed() is False: # swap the src and dst vertices for the lower triangle only. Because # this is a symmeterized graph, this operation results in a df with # multiple src/dst entries. ddf["src"], ddf["dst"] = ddf[["src", "dst"]].min(axis=1), ddf[ ["src", "dst"] ].max(axis=1) # overwrite the df with the sum of the values for all alike src/dst # vertex pairs, resulting in half the edges of the original df from the # symmeterized graph. ddf = ddf.groupby(by=["src", "dst"]).sum().reset_index() return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/centrality/__init__.py
# Copyright (c) 2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.centrality.betweenness_centrality import ( betweenness_centrality, edge_betweenness_centrality, )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/centrality/katz_centrality.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client from pylibcugraph import ResourceHandle, katz_centrality as pylibcugraph_katz import cugraph.dask.comms.comms as Comms import dask_cudf import cudf import warnings def _call_plc_katz_centrality( sID, mg_graph_x, betas, alpha, beta, epsilon, max_iterations, do_expensive_check ): return pylibcugraph_katz( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, betas=betas, alpha=alpha, beta=beta, epsilon=epsilon, max_iterations=max_iterations, do_expensive_check=do_expensive_check, ) def convert_to_cudf(cp_arrays): """ create a cudf DataFrame from cupy arrays """ cupy_vertices, cupy_values = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertices df["katz_centrality"] = cupy_values return df def katz_centrality( input_graph, alpha=None, beta=1.0, max_iter=100, tol=1.0e-6, nstart=None, normalized=True, ): """ Compute the Katz centrality for the nodes of the graph G. Parameters ---------- input_graph : cuGraph.Graph cuGraph graph descriptor with connectivity information. The graph can contain either directed or undirected edges. alpha : float, optional (default=None) Attenuation factor. If alpha is not specified then it is internally calculated as 1/(degree_max) where degree_max is the maximum out degree. NOTE The maximum acceptable value of alpha for convergence alpha_max = 1/(lambda_max) where lambda_max is the largest eigenvalue of the graph. Since lambda_max is always lesser than or equal to degree_max for a graph, alpha_max will always be greater than or equal to (1/degree_max). Therefore, setting alpha to (1/degree_max) will guarantee that it will never exceed alpha_max thus in turn fulfilling the requirement for convergence. beta : float, optional (default=None) Weight scalar added to each vertex's new Katz Centrality score in every iteration. If beta is not specified then it is set as 1.0. max_iter : int, optional (default=100) The maximum number of iterations before an answer is returned. This can be used to limit the execution time and do an early exit before the solver reaches the convergence tolerance. If this value is lower or equal to 0 cuGraph will use the default value, which is 100. tol : float, optional (default=1.0e-5) Set the tolerance the approximation, this parameter should be a small magnitude value. The lower the tolerance the better the approximation. If this value is 0.0f, cuGraph will use the default value which is 1.0e-6. Setting too small a tolerance can lead to non-convergence due to numerical roundoff. Usually values between 1e-2 and 1e-6 are acceptable. nstart : dask_cudf.Dataframe, optional (default=None) Distributed GPU Dataframe containing the initial guess for katz centrality. nstart['vertex'] : dask_cudf.Series Contains the vertex identifiers nstart['values'] : dask_cudf.Series Contains the katz centrality values of vertices normalized : not supported If True normalize the resulting katz centrality values Returns ------- katz_centrality : dask_cudf.DataFrame GPU distributed data frame containing two dask_cudf.Series of size V: the vertex identifiers and the corresponding katz centrality values. ddf['vertex'] : dask_cudf.Series Contains the vertex identifiers ddf['katz_centrality'] : dask_cudf.Series Contains the katz centrality of vertices Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst') >>> pr = dcg.katz_centrality(dg) """ client = default_client() if input_graph.store_transposed is False: warning_msg = ( "Katz centrality expects the 'store_transposed' flag " "to be set to 'True' for optimal performance during " "the graph creation" ) warnings.warn(warning_msg, UserWarning) if alpha is None: degree_max = input_graph.degree()["degree"].max().compute() alpha = 1 / (degree_max) if (alpha is not None) and (alpha <= 0.0): raise ValueError(f"'alpha' must be a positive float or None, " f"got: {alpha}") # FIXME: should we add this parameter as an option? do_expensive_check = False initial_hubs_guess_values = None if nstart: if input_graph.renumbered: if len(input_graph.renumber_map.implementation.col_names) > 1: cols = nstart.columns[:-1].to_list() else: cols = "vertex" nstart = input_graph.add_internal_vertex_id(nstart, "vertex", cols) initial_hubs_guess_values = nstart[nstart.columns[0]].compute() else: initial_hubs_guess_values = nstart["values"] if isinstance(nstart, dask_cudf.DataFrame): initial_hubs_guess_values = initial_hubs_guess_values.compute() cupy_result = [ client.submit( _call_plc_katz_centrality, Comms.get_session_id(), input_graph._plc_graph[w], initial_hubs_guess_values, alpha, beta, tol, max_iter, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(cupy_result) cudf_result = [ client.submit( convert_to_cudf, cp_arrays, workers=client.who_has(cp_arrays)[cp_arrays.key] ) for cp_arrays in cupy_result ] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(cupy_result, cudf_result)]) if input_graph.renumbered: return input_graph.unrenumber(ddf, "vertex") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/traversal/bfs.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from pylibcugraph import ResourceHandle, bfs as pylibcugraph_bfs from dask.distributed import wait, default_client from cugraph.dask.common.input_utils import get_distributed_data import cugraph.dask.comms.comms as Comms import cudf import dask_cudf import warnings def convert_to_cudf(cp_arrays): """ create a cudf DataFrame from cupy arrays """ cupy_distances, cupy_predecessors, cupy_vertices = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertices df["distance"] = cupy_distances df["predecessor"] = cupy_predecessors return df def _call_plc_bfs( sID, mg_graph_x, st_x, depth_limit=None, direction_optimizing=False, return_distances=True, do_expensive_check=False, ): return pylibcugraph_bfs( ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, sources=st_x, direction_optimizing=direction_optimizing, depth_limit=depth_limit if depth_limit is not None else 0, compute_predecessors=return_distances, do_expensive_check=do_expensive_check, ) def bfs(input_graph, start, depth_limit=None, return_distances=True, check_start=True): """ Find the distances and predecessors for a breadth-first traversal of a graph. The input graph must contain edge list as a dask-cudf dataframe with one partition per GPU. Parameters ---------- input_graph : cugraph.Graph cuGraph graph instance, should contain the connectivity information as dask cudf edge list dataframe (edge weights are not used for this algorithm). start : Integer or list or cudf object or dask_cudf object The id(s) of the graph vertex from which the traversal begins in each component of the graph. Only one vertex per connected component of the graph is allowed. depth_limit : Integer or None, optional (default=None) Limit the depth of the search return_distances : bool, optional (default=True) Indicates if distances should be returned check_start : bool, optional (default=True) If True, performs more extensive tests on the start vertices to ensure validitity, at the expense of increased run time. Returns ------- df : dask_cudf.DataFrame df['vertex'] gives the vertex id df['distance'] gives the path distance from the starting vertex (Only if return_distances is True) df['predecessor'] gives the vertex it was reached from in the traversal Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst', ... edge_attr='value') >>> df = dcg.bfs(dg, 0) """ client = default_client() invalid_dtype = False if not isinstance(start, (dask_cudf.DataFrame, dask_cudf.Series)): if not isinstance(start, (cudf.DataFrame, cudf.Series)): vertex_dtype = input_graph.nodes().dtype start = cudf.Series(start, dtype=vertex_dtype) # convert into a dask_cudf start = dask_cudf.from_cudf(start, input_graph._npartitions) if check_start: if isinstance(start, dask_cudf.Series): vertex_dtype = input_graph.nodes().dtype if start.dtype is not vertex_dtype: invalid_dtype = True else: # Multicolumn vertices case start_dtype = start.dtypes.reset_index(drop=True) vertex_dtype = input_graph.nodes().dtypes.reset_index(drop=True) if not start_dtype.equals(vertex_dtype): invalid_dtype = True if invalid_dtype: warning_msg = ( "The 'start' values dtype must match " "the graph's vertices dtype." ) warnings.warn(warning_msg, UserWarning) if isinstance(start, dask_cudf.Series): start = start.astype(vertex_dtype) else: start = start.astype(vertex_dtype[0]) is_valid_vertex = input_graph.has_node(start) if not is_valid_vertex: raise ValueError("At least one start vertex provided was invalid") if input_graph.renumbered: if isinstance(start, dask_cudf.DataFrame): tmp_col_names = start.columns elif isinstance(start, dask_cudf.Series): tmp_col_names = None start = input_graph.lookup_internal_vertex_id(start, tmp_col_names) data_start = get_distributed_data(start) do_expensive_check = False # FIXME: Why is 'direction_optimizing' not part of the python cugraph API # and why is it set to 'False' by default direction_optimizing = False cupy_result = [ client.submit( _call_plc_bfs, Comms.get_session_id(), input_graph._plc_graph[w], st[0], depth_limit, direction_optimizing, return_distances, do_expensive_check, workers=[w], allow_other_workers=False, ) for w, st in data_start.worker_to_parts.items() ] wait(cupy_result) cudf_result = [ client.submit(convert_to_cudf, cp_arrays) for cp_arrays in cupy_result ] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(cupy_result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") ddf = input_graph.unrenumber(ddf, "predecessor") ddf = ddf.fillna(-1) return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/traversal/__init__.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/traversal/sssp.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import cupy import cudf import dask_cudf from pylibcugraph import sssp as pylibcugraph_sssp, ResourceHandle def _call_plc_sssp( sID, mg_graph_x, source, cutoff, compute_predecessors, do_expensive_check ): vertices, distances, predecessors = pylibcugraph_sssp( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, source=source, cutoff=cutoff, compute_predecessors=compute_predecessors, do_expensive_check=do_expensive_check, ) return cudf.DataFrame( { "distance": cudf.Series(distances), "vertex": cudf.Series(vertices), "predecessor": cudf.Series(predecessors), } ) def sssp(input_graph, source, cutoff=None, check_source=True): """ Compute the distance and predecessors for shortest paths from the specified source to all the vertices in the input_graph. The distances column will store the distance from the source to each vertex. The predecessors column will store each vertex's predecessor in the shortest path. Vertices that are unreachable will have a distance of infinity denoted by the maximum value of the data type and the predecessor set as -1. The source vertex's predecessor is also set to -1. The input graph must contain edge list as dask-cudf dataframe with one partition per GPU. Parameters ---------- input_graph : cugraph.Graph cuGraph graph descriptor, should contain the connectivity information as dask cudf edge list dataframe. source : Integer Specify source vertex cutoff : double, optional (default = None) Maximum edge weight sum considered by the algorithm check_source : bool, optional (default=True) If True, performs more extensive tests on the start vertices to ensure validitity, at the expense of increased run time. Returns ------- df : dask_cudf.DataFrame df['vertex'] gives the vertex id df['distance'] gives the path distance from the starting vertex df['predecessor'] gives the vertex id it was reached from in the traversal Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst', ... edge_attr='value') >>> df = dcg.sssp(dg, 0) """ # FIXME: Implement a better way to check if the graph is weighted similar # to 'simpleGraph' if not input_graph.weighted: err_msg = ( "'SSSP' requires the input graph to be weighted." "'BFS' should be used instead of 'SSSP' for unweighted graphs." ) raise ValueError(err_msg) client = default_client() def check_valid_vertex(G, source): is_valid_vertex = G.has_node(source) if not is_valid_vertex: raise ValueError("Invalid source vertex") if check_source: check_valid_vertex(input_graph, source) if cutoff is None: cutoff = cupy.inf if input_graph.renumbered: source = ( input_graph.lookup_internal_vertex_id(cudf.Series([source])) .fillna(-1) .compute() ) source = source.iloc[0] do_expensive_check = False compute_predecessors = True result = [ client.submit( _call_plc_sssp, Comms.get_session_id(), input_graph._plc_graph[w], source, cutoff, compute_predecessors, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) ddf = dask_cudf.from_delayed(result).persist() wait(ddf) # Wait until the inactive futures are released wait([r.release() for r in result]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") ddf = input_graph.unrenumber(ddf, "predecessor") ddf["predecessor"] = ddf["predecessor"].fillna(-1) return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_prediction/overlap.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from cugraph.dask.common.input_utils import get_distributed_data from cugraph.utilities import renumber_vertex_pair from pylibcugraph import ( overlap_coefficients as pylibcugraph_overlap_coefficients, ) from pylibcugraph import ResourceHandle def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_first, cupy_second, cupy_similarity = cp_arrays df = cudf.DataFrame() df["first"] = cupy_first df["second"] = cupy_second df["overlap_coeff"] = cupy_similarity return df def _call_plc_overlap( sID, mg_graph_x, vertex_pair, use_weight, do_expensive_check, vertex_pair_col_name ): first = vertex_pair[vertex_pair_col_name[0]] second = vertex_pair[vertex_pair_col_name[1]] return pylibcugraph_overlap_coefficients( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, first=first, second=second, use_weight=use_weight, do_expensive_check=do_expensive_check, ) def overlap(input_graph, vertex_pair=None, use_weight=False): """ Compute the Overlap Coefficient between each pair of vertices connected by an edge, or between arbitrary pairs of vertices specified by the user. Overlap Coefficient is defined between two sets as the ratio of the volume of their intersection divided by the smaller of their two volumes. In the context of graphs, the neighborhood of a vertex is seen as a set. The Overlap Coefficient weight of each edge represents the strength of connection between vertices based on the relative similarity of their neighbors. If first is specified but second is not, or vice versa, an exception will be thrown. cugraph.overlap, in the absence of a specified vertex pair list, will compute the two_hop_neighbors of the entire graph to construct a vertex pair list and will return the Overlap coefficient for those vertex pairs. This is not advisable as the vertex_pairs can grow exponentially with respect to the size of the datasets Parameters ---------- input_graph : cugraph.Graph cuGraph Graph instance, should contain the connectivity information as an edge list (edge weights are not supported yet for this algorithm). The graph should be undirected where an undirected edge is represented by a directed edge in both direction. The adjacency list will be computed if not already present. This implementation only supports undirected, unweighted Graph. vertex_pair : cudf.DataFrame, optional (default=None) A GPU dataframe consisting of two columns representing pairs of vertices. If provided, the Overlap coefficient is computed for the given vertex pairs. If the vertex_pair is not provided then the current implementation computes the Overlap coefficient for all adjacent vertices in the graph. use_weight : bool, optional (default=False) Flag to indicate whether to compute weighted overlap (if use_weight==True) or un-weighted overlap (if use_weight==False). 'input_graph' must be weighted if 'use_weight=True'. Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing 2 dask_cudf.Series ddf['first']: dask_cudf.Series The first vertex ID of each pair(will be identical to first if specified). ddf['second']: dask_cudf.Series The second vertex ID of each pair(will be identical to second if specified). ddf['overlap_coeff']: dask_cudf.Series The computed overlap coefficient between the first and the second vertex ID. """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") if vertex_pair is None: # Call two_hop neighbor of the entire graph vertex_pair = input_graph.get_two_hop_neighbors() vertex_pair_col_name = vertex_pair.columns if isinstance(vertex_pair, (dask_cudf.DataFrame, cudf.DataFrame)): vertex_pair = renumber_vertex_pair(input_graph, vertex_pair) elif vertex_pair is not None: raise ValueError("vertex_pair must be a dask_cudf or cudf dataframe") if not isinstance(vertex_pair, (dask_cudf.DataFrame)): vertex_pair = dask_cudf.from_cudf( vertex_pair, npartitions=len(Comms.get_workers()) ) vertex_pair = get_distributed_data(vertex_pair) wait(vertex_pair) vertex_pair = vertex_pair.worker_to_parts # Initialize dask client client = default_client() do_expensive_check = False if vertex_pair is not None: result = [ client.submit( _call_plc_overlap, Comms.get_session_id(), input_graph._plc_graph[w], vertex_pair[w][0], use_weight, do_expensive_check, vertex_pair_col_name, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "first") ddf = input_graph.unrenumber(ddf, "second") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_prediction/jaccard.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from cugraph.dask.common.input_utils import get_distributed_data from cugraph.utilities import renumber_vertex_pair from pylibcugraph import ( jaccard_coefficients as pylibcugraph_jaccard_coefficients, ) from pylibcugraph import ResourceHandle def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_first, cupy_second, cupy_similarity = cp_arrays df = cudf.DataFrame() df["first"] = cupy_first df["second"] = cupy_second df["jaccard_coeff"] = cupy_similarity return df def _call_plc_jaccard( sID, mg_graph_x, vertex_pair, use_weight, do_expensive_check, vertex_pair_col_name ): first = vertex_pair[vertex_pair_col_name[0]] second = vertex_pair[vertex_pair_col_name[1]] return pylibcugraph_jaccard_coefficients( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, first=first, second=second, use_weight=use_weight, do_expensive_check=do_expensive_check, ) def jaccard(input_graph, vertex_pair=None, use_weight=False): """ Compute the Jaccard similarity between each pair of vertices connected by an edge, or between arbitrary pairs of vertices specified by the user. Jaccard similarity is defined between two sets as the ratio of the volume of their intersection divided by the volume of their union. In the context of graphs, the neighborhood of a vertex is seen as a set. The Jaccard similarity weight of each edge represents the strength of connection between vertices based on the relative similarity of their neighbors. If first is specified but second is not, or vice versa, an exception will be thrown. NOTE: If the vertex_pair parameter is not specified then the behavior of cugraph.jaccard is different from the behavior of networkx.jaccard_coefficient. cugraph.dask.jaccard, in the absence of a specified vertex pair list, will compute the two_hop_neighbors of the entire graph to construct a vertex pair list and will return the jaccard coefficient for those vertex pairs. This is not advisable as the vertex_pairs can grow exponentially with respect to the size of the datasets networkx.jaccard_coefficient, in the absence of a specified vertex pair list, will return an upper triangular dense matrix, excluding the diagonal as well as vertex pairs that are directly connected by an edge in the graph, of jaccard coefficients. Technically, networkx returns a lazy iterator across this upper triangular matrix where the actual jaccard coefficient is computed when the iterator is dereferenced. Computing a dense matrix of results is not feasible if the number of vertices in the graph is large (100,000 vertices would result in 4.9 billion values in that iterator). If your graph is small enough (or you have enough memory and patience) you can get the interesting (non-zero) values that are part of the networkx solution by doing the following: But please remember that cugraph will fill the dataframe with the entire solution you request, so you'll need enough memory to store the 2-hop neighborhood dataframe. Parameters ---------- input_graph : cugraph.Graph cuGraph Graph instance, should contain the connectivity information as an edge list (edge weights are not supported yet for this algorithm). The graph should be undirected where an undirected edge is represented by a directed edge in both direction. The adjacency list will be computed if not already present. This implementation only supports undirected, unweighted Graph. vertex_pair : cudf.DataFrame, optional (default=None) A GPU dataframe consisting of two columns representing pairs of vertices. If provided, the jaccard coefficient is computed for the given vertex pairs. If the vertex_pair is not provided then the current implementation computes the jaccard coefficient for all adjacent vertices in the graph. use_weight : bool, optional (default=False) Flag to indicate whether to compute weighted jaccard (if use_weight==True) or un-weighted jaccard (if use_weight==False). 'input_graph' must be weighted if 'use_weight=True'. Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing 2 dask_cudf.Series ddf['first']: dask_cudf.Series The first vertex ID of each pair (will be identical to first if specified). ddf['second']: dask_cudf.Series The second vertex ID of each pair (will be identical to second if specified). ddf['jaccard_coeff']: dask_cudf.Series The computed jaccard coefficient between the first and the second vertex ID. """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") if vertex_pair is None: # Call two_hop neighbor of the entire graph vertex_pair = input_graph.get_two_hop_neighbors() vertex_pair_col_name = vertex_pair.columns if isinstance(vertex_pair, (dask_cudf.DataFrame, cudf.DataFrame)): vertex_pair = renumber_vertex_pair(input_graph, vertex_pair) elif vertex_pair is not None: raise ValueError("vertex_pair must be a dask_cudf or cudf dataframe") if not isinstance(vertex_pair, (dask_cudf.DataFrame)): vertex_pair = dask_cudf.from_cudf( vertex_pair, npartitions=len(Comms.get_workers()) ) vertex_pair = get_distributed_data(vertex_pair) wait(vertex_pair) vertex_pair = vertex_pair.worker_to_parts # Initialize dask client client = default_client() do_expensive_check = False if vertex_pair is not None: result = [ client.submit( _call_plc_jaccard, Comms.get_session_id(), input_graph._plc_graph[w], vertex_pair[w][0], use_weight, do_expensive_check, vertex_pair_col_name, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "first") ddf = input_graph.unrenumber(ddf, "second") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_prediction/__init__.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_prediction/sorensen.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf from cugraph.dask.common.input_utils import get_distributed_data from cugraph.utilities import renumber_vertex_pair from pylibcugraph import ( sorensen_coefficients as pylibcugraph_sorensen_coefficients, ) from pylibcugraph import ResourceHandle def convert_to_cudf(cp_arrays): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ cupy_first, cupy_second, cupy_similarity = cp_arrays df = cudf.DataFrame() df["first"] = cupy_first df["second"] = cupy_second df["sorensen_coeff"] = cupy_similarity return df def _call_plc_sorensen( sID, mg_graph_x, vertex_pair, use_weight, do_expensive_check, vertex_pair_col_name ): first = vertex_pair[vertex_pair_col_name[0]] second = vertex_pair[vertex_pair_col_name[1]] return pylibcugraph_sorensen_coefficients( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, first=first, second=second, use_weight=use_weight, do_expensive_check=do_expensive_check, ) def sorensen(input_graph, vertex_pair=None, use_weight=False): """ Compute the Sorensen coefficient between each pair of vertices connected by an edge, or between arbitrary pairs of vertices specified by the user. Sorensen coefficient is defined between two sets as the ratio of twice the volume of their intersection divided by the volume of each set. If first is specified but second is not, or vice versa, an exception will be thrown. cugraph.dask.sorensen, in the absence of a specified vertex pair list, will compute the two_hop_neighbors of the entire graph to construct a vertex pair list and will return the sorensen coefficient for those vertex pairs. This is not advisable as the vertex_pairs can grow exponentially with respect to the size of the datasets Parameters ---------- input_graph : cugraph.Graph cuGraph Graph instance, should contain the connectivity information as an edge list (edge weights are not supported yet for this algorithm). The graph should be undirected where an undirected edge is represented by a directed edge in both direction. The adjacency list will be computed if not already present. This implementation only supports undirected, unweighted Graph. vertex_pair : cudf.DataFrame, optional (default=None) A GPU dataframe consisting of two columns representing pairs of vertices. If provided, the sorensen coefficient is computed for the given vertex pairs. If the vertex_pair is not provided then the current implementation computes the sorensen coefficient for all adjacent vertices in the graph. use_weight : bool, optional (default=False) Flag to indicate whether to compute weighted sorensen (if use_weight==True) or un-weighted sorensen (if use_weight==False). 'input_graph' must be weighted if 'use_weight=True'. Returns ------- result : dask_cudf.DataFrame GPU distributed data frame containing 2 dask_cudf.Series ddf['first']: dask_cudf.Series The first vertex ID of each pair(will be identical to first if specified). ddf['second']: dask_cudf.Series The second vertex ID of each pair(will be identical to second if specified). ddf['sorensen_coeff']: dask_cudf.Series The computed sorensen coefficient between the first and the second vertex ID. """ if input_graph.is_directed(): raise ValueError("input graph must be undirected") if vertex_pair is None: # Call two_hop neighbor of the entire graph vertex_pair = input_graph.get_two_hop_neighbors() vertex_pair_col_name = vertex_pair.columns if isinstance(vertex_pair, (dask_cudf.DataFrame, cudf.DataFrame)): vertex_pair = renumber_vertex_pair(input_graph, vertex_pair) elif vertex_pair is not None: raise ValueError("vertex_pair must be a dask_cudf or cudf dataframe") if not isinstance(vertex_pair, (dask_cudf.DataFrame)): vertex_pair = dask_cudf.from_cudf( vertex_pair, npartitions=len(Comms.get_workers()) ) vertex_pair = get_distributed_data(vertex_pair) wait(vertex_pair) vertex_pair = vertex_pair.worker_to_parts # Initialize dask client client = default_client() do_expensive_check = False if vertex_pair is not None: result = [ client.submit( _call_plc_sorensen, Comms.get_session_id(), input_graph._plc_graph[w], vertex_pair[w][0], use_weight, do_expensive_check, vertex_pair_col_name, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) cudf_result = [client.submit(convert_to_cudf, cp_arrays) for cp_arrays in result] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, cudf_result)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "first") ddf = input_graph.unrenumber(ddf, "second") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_analysis/hits.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from dask.distributed import wait, default_client import cugraph.dask.comms.comms as Comms import dask_cudf import cudf import warnings from pylibcugraph import ResourceHandle, hits as pylibcugraph_hits def _call_plc_hits( sID, mg_graph_x, tol, max_iter, initial_hubs_guess_vertices, initial_hubs_guess_values, normalized, do_expensive_check, ): return pylibcugraph_hits( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, tol=tol, max_iter=max_iter, initial_hubs_guess_vertices=initial_hubs_guess_vertices, initial_hubs_guess_values=initial_hubs_guess_values, normalized=normalized, do_expensive_check=do_expensive_check, ) def convert_to_cudf(cp_arrays): """ create a cudf DataFrame from cupy arrays """ cupy_vertices, cupy_hubs, cupy_authorities = cp_arrays df = cudf.DataFrame() df["vertex"] = cupy_vertices df["hubs"] = cupy_hubs df["authorities"] = cupy_authorities return df def hits(input_graph, tol=1.0e-5, max_iter=100, nstart=None, normalized=True): """ Compute HITS hubs and authorities values for each vertex The HITS algorithm computes two numbers for a node. Authorities estimates the node value based on the incoming links. Hubs estimates the node value based on outgoing links. Both cuGraph and networkx implementation use a 1-norm. Parameters ---------- input_graph : cugraph.Graph cuGraph graph descriptor, should contain the connectivity information as an edge list (edge weights are not used for this algorithm). The adjacency list will be computed if not already present. tol : float, optional (default=1.0e-5) Set the tolerance of the approximation, this parameter should be a small magnitude value. max_iter : int, optional (default=100) The maximum number of iterations before an answer is returned. nstart : cudf.Dataframe, optional (default=None) The initial hubs guess vertices along with their initial hubs guess value nstart['vertex'] : cudf.Series Initial hubs guess vertices nstart['values'] : cudf.Series Initial hubs guess values normalized : bool, optional (default=True) A flag to normalize the results Returns ------- HubsAndAuthorities : dask_cudf.DataFrame GPU distributed data frame containing three dask_cudf.Series of size V: the vertex identifiers and the corresponding hubs and authorities values. df['vertex'] : dask_cudf.Series Contains the vertex identifiers df['hubs'] : dask_cudf.Series Contains the hubs score df['authorities'] : dask_cudf.Series Contains the authorities score Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst', ... edge_attr='value') >>> hits = dcg.hits(dg, max_iter = 50) """ client = default_client() if input_graph.store_transposed is False: warning_msg = ( "HITS expects the 'store_transposed' flag " "to be set to 'True' for optimal performance during " "the graph creation" ) warnings.warn(warning_msg, UserWarning) do_expensive_check = False initial_hubs_guess_vertices = None initial_hubs_guess_values = None if nstart is not None: initial_hubs_guess_vertices = nstart["vertex"] initial_hubs_guess_values = nstart["values"] cupy_result = [ client.submit( _call_plc_hits, Comms.get_session_id(), input_graph._plc_graph[w], tol, max_iter, initial_hubs_guess_vertices, initial_hubs_guess_values, normalized, do_expensive_check, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(cupy_result) cudf_result = [ client.submit( convert_to_cudf, cp_arrays, workers=client.who_has(cp_arrays)[cp_arrays.key] ) for cp_arrays in cupy_result ] wait(cudf_result) ddf = dask_cudf.from_delayed(cudf_result).persist() wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(cupy_result, cudf_result)]) if input_graph.renumbered: return input_graph.unrenumber(ddf, "vertex") return ddf
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_analysis/__init__.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/link_analysis/pagerank.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import warnings import dask from dask.distributed import wait, default_client import dask_cudf import cudf import numpy as np from pylibcugraph import ( pagerank as plc_pagerank, personalized_pagerank as plc_p_pagerank, exceptions as plc_exceptions, ResourceHandle, ) import cugraph.dask.comms.comms as Comms from cugraph.dask.common.input_utils import get_distributed_data from cugraph.exceptions import FailedToConvergeError def convert_to_return_tuple(plc_pr_retval): """ Using the PLC pagerank return tuple, creates a cudf DataFrame from the cupy arrays and extracts the (optional) bool. """ if len(plc_pr_retval) == 3: cupy_vertices, cupy_pagerank, converged = plc_pr_retval else: cupy_vertices, cupy_pagerank = plc_pr_retval converged = True df = cudf.DataFrame() df["vertex"] = cupy_vertices df["pagerank"] = cupy_pagerank return (df, converged) # FIXME: Move this function to the utility module so that it can be # shared by other algos def ensure_valid_dtype(input_graph, input_df, input_df_name): if input_graph.properties.weighted is False: # If the graph is not weighted, an artificial weight column # of type 'float32' is added and it must match the user # personalization/nstart values. edge_attr_dtype = np.float32 else: edge_attr_dtype = input_graph.input_df["value"].dtype if "values" in input_df.columns: input_df_values_dtype = input_df["values"].dtype if input_df_values_dtype != edge_attr_dtype: warning_msg = ( f"PageRank requires '{input_df_name}' values " "to match the graph's 'edge_attr' type. " f"edge_attr type is: {edge_attr_dtype} and got " f"'{input_df_name}' values of type: " f"{input_df_values_dtype}." ) warnings.warn(warning_msg, UserWarning) input_df = input_df.astype({"values": edge_attr_dtype}) vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] input_df_vertex_dtype = input_df["vertex"].dtype if input_df_vertex_dtype != vertex_dtype: warning_msg = ( f"PageRank requires '{input_df_name}' vertex " "to match the graph's 'vertex' type. " f"input graph's vertex type is: {vertex_dtype} and got " f"'{input_df_name}' vertex of type: " f"{input_df_vertex_dtype}." ) warnings.warn(warning_msg, UserWarning) input_df = input_df.astype({"vertex": vertex_dtype}) return input_df def renumber_vertices(input_graph, input_df): input_df = input_graph.add_internal_vertex_id( input_df, "vertex", "vertex" ).compute() return input_df def _call_plc_pagerank( sID, mg_graph_x, pre_vtx_o_wgt_vertices, pre_vtx_o_wgt_sums, initial_guess_vertices, initial_guess_values, alpha, epsilon, max_iterations, do_expensive_check, fail_on_nonconvergence, ): try: return plc_pagerank( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, precomputed_vertex_out_weight_vertices=pre_vtx_o_wgt_vertices, precomputed_vertex_out_weight_sums=pre_vtx_o_wgt_sums, initial_guess_vertices=initial_guess_vertices, initial_guess_values=initial_guess_values, alpha=alpha, epsilon=epsilon, max_iterations=max_iterations, do_expensive_check=do_expensive_check, fail_on_nonconvergence=fail_on_nonconvergence, ) # Re-raise this as a cugraph exception so users trying to catch this do not # have to know to import another package. except plc_exceptions.FailedToConvergeError as exc: raise FailedToConvergeError from exc def _call_plc_personalized_pagerank( sID, mg_graph_x, pre_vtx_o_wgt_vertices, pre_vtx_o_wgt_sums, data_personalization, initial_guess_vertices, initial_guess_values, alpha, epsilon, max_iterations, do_expensive_check, fail_on_nonconvergence, ): personalization_vertices = data_personalization["vertex"] personalization_values = data_personalization["values"] try: return plc_p_pagerank( resource_handle=ResourceHandle(Comms.get_handle(sID).getHandle()), graph=mg_graph_x, precomputed_vertex_out_weight_vertices=pre_vtx_o_wgt_vertices, precomputed_vertex_out_weight_sums=pre_vtx_o_wgt_sums, personalization_vertices=personalization_vertices, personalization_values=personalization_values, initial_guess_vertices=initial_guess_vertices, initial_guess_values=initial_guess_values, alpha=alpha, epsilon=epsilon, max_iterations=max_iterations, do_expensive_check=do_expensive_check, fail_on_nonconvergence=fail_on_nonconvergence, ) # Re-raise this as a cugraph exception so users trying to catch this do not # have to know to import another package. except plc_exceptions.FailedToConvergeError as exc: raise FailedToConvergeError from exc def pagerank( input_graph, alpha=0.85, personalization=None, precomputed_vertex_out_weight=None, max_iter=100, tol=1.0e-5, nstart=None, fail_on_nonconvergence=True, ): """ Find the PageRank values for each vertex in a graph using multiple GPUs. cuGraph computes an approximation of the Pagerank using the power method. The input graph must contain edge list as dask-cudf dataframe with one partition per GPU. All edges will have an edge_attr value of 1.0 if not provided. Parameters ---------- input_graph : cugraph.Graph cuGraph graph descriptor, should contain the connectivity information as dask cudf edge list dataframe(edge weights are not used for this algorithm). alpha : float, optional (default=0.85) The damping factor alpha represents the probability to follow an outgoing edge, standard value is 0.85. Thus, 1.0-alpha is the probability to “teleport” to a random vertex. Alpha should be greater than 0.0 and strictly lower than 1.0. personalization : cudf.Dataframe, optional (default=None) GPU Dataframe containing the personalization information. (a performance optimization) personalization['vertex'] : cudf.Series Subset of vertices of graph for personalization personalization['values'] : cudf.Series Personalization values for vertices precomputed_vertex_out_weight : cudf.Dataframe, optional (default=None) GPU Dataframe containing the precomputed vertex out weight (a performance optimization) information. precomputed_vertex_out_weight['vertex'] : cudf.Series Subset of vertices of graph for precomputed_vertex_out_weight precomputed_vertex_out_weight['sums'] : cudf.Series Corresponding precomputed sum of outgoing vertices weight max_iter : int, optional (default=100) The maximum number of iterations before an answer is returned. This can be used to limit the execution time and do an early exit before the solver reaches the convergence tolerance. If this value is lower or equal to 0 cuGraph will use the default value, which is 100. tol : float, optional (default=1e-05) Set the tolerance the approximation, this parameter should be a small magnitude value. The lower the tolerance the better the approximation. If this value is 0.0f, cuGraph will use the default value which is 1.0E-5. Setting too small a tolerance can lead to non-convergence due to numerical roundoff. Usually values between 0.01 and 0.00001 are acceptable. nstart : cudf.Dataframe, optional (default=None) GPU Dataframe containing the initial guess for pagerank. (a performance optimization) nstart['vertex'] : cudf.Series Subset of vertices of graph for initial guess for pagerank values nstart['values'] : cudf.Series Pagerank values for vertices fail_on_nonconvergence : bool (default=True) If the solver does not reach convergence, raise an exception if fail_on_nonconvergence is True. If fail_on_nonconvergence is False, the return value is a tuple of (pagerank, converged) where pagerank is a cudf.DataFrame as described below, and converged is a boolean indicating if the solver converged (True) or not (False). Returns ------- The return value varies based on the value of the fail_on_nonconvergence paramter. If fail_on_nonconvergence is True: PageRank : dask_cudf.DataFrame GPU data frame containing two dask_cudf.Series of size V: the vertex identifiers and the corresponding PageRank values. NOTE: if the input cugraph.Graph was created using the renumber=False option of any of the from_*_edgelist() methods, pagerank assumes that the vertices in the edgelist are contiguous and start from 0. If the actual set of vertices in the edgelist is not contiguous (has gaps) or does not start from zero, pagerank will assume the "missing" vertices are isolated vertices in the graph, and will compute and return pagerank values for each. If this is not the desired behavior, ensure the input cugraph.Graph is created from the from_*_edgelist() functions with the renumber=True option (the default) ddf['vertex'] : dask_cudf.Series Contains the vertex identifiers ddf['pagerank'] : dask_cudf.Series Contains the PageRank score If fail_on_nonconvergence is False: (PageRank, converged) : tuple of (dask_cudf.DataFrame, bool) PageRank is the GPU dataframe described above, converged is a bool indicating if the solver converged (True) or not (False). Examples -------- >>> import cugraph.dask as dcg >>> import dask_cudf >>> # ... Init a DASK Cluster >>> # see https://docs.rapids.ai/api/cugraph/stable/dask-cugraph.html >>> # Download dataset from https://github.com/rapidsai/cugraph/datasets/.. >>> chunksize = dcg.get_chunksize(datasets_path / "karate.csv") >>> ddf = dask_cudf.read_csv(datasets_path / "karate.csv", ... chunksize=chunksize, delimiter=" ", ... names=["src", "dst", "value"], ... dtype=["int32", "int32", "float32"]) >>> dg = cugraph.Graph(directed=True) >>> dg.from_dask_cudf_edgelist(ddf, source='src', destination='dst') >>> pr = dcg.pagerank(dg) """ # Initialize dask client client = default_client() if input_graph.store_transposed is False: warning_msg = ( "Pagerank expects the 'store_transposed' flag " "to be set to 'True' for optimal performance during " "the graph creation" ) warnings.warn(warning_msg, UserWarning) initial_guess_vertices = None initial_guess_values = None precomputed_vertex_out_weight_vertices = None precomputed_vertex_out_weight_sums = None do_expensive_check = False # FIXME: Distribute the 'precomputed_vertex_out_weight' # across GPUs for performance optimization if precomputed_vertex_out_weight is not None: if input_graph.renumbered is True: precomputed_vertex_out_weight = renumber_vertices( input_graph, precomputed_vertex_out_weight ) precomputed_vertex_out_weight = ensure_valid_dtype( input_graph, precomputed_vertex_out_weight, "precomputed_vertex_out_weight" ) precomputed_vertex_out_weight_vertices = precomputed_vertex_out_weight["vertex"] precomputed_vertex_out_weight_sums = precomputed_vertex_out_weight["sums"] # FIXME: Distribute the 'nstart' across GPUs for performance optimization if nstart is not None: if input_graph.renumbered is True: nstart = renumber_vertices(input_graph, nstart) nstart = ensure_valid_dtype(input_graph, nstart, "nstart") initial_guess_vertices = nstart["vertex"] initial_guess_values = nstart["values"] if personalization is not None: if input_graph.renumbered is True: personalization = renumber_vertices(input_graph, personalization) personalization = ensure_valid_dtype( input_graph, personalization, "personalization" ) personalization_ddf = dask_cudf.from_cudf( personalization, npartitions=len(Comms.get_workers()) ) data_prsztn = get_distributed_data(personalization_ddf) result = [ client.submit( _call_plc_personalized_pagerank, Comms.get_session_id(), input_graph._plc_graph[w], precomputed_vertex_out_weight_vertices, precomputed_vertex_out_weight_sums, data_personalization[0], initial_guess_vertices, initial_guess_values, alpha, tol, max_iter, do_expensive_check, fail_on_nonconvergence, workers=[w], allow_other_workers=False, ) for w, data_personalization in data_prsztn.worker_to_parts.items() ] else: result = [ client.submit( _call_plc_pagerank, Comms.get_session_id(), input_graph._plc_graph[w], precomputed_vertex_out_weight_vertices, precomputed_vertex_out_weight_sums, initial_guess_vertices, initial_guess_values, alpha, tol, max_iter, do_expensive_check, fail_on_nonconvergence, workers=[w], allow_other_workers=False, ) for w in Comms.get_workers() ] wait(result) vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] # Have each worker convert tuple of arrays and bool from PLC to cudf # DataFrames and bools. This will be a list of futures. result_tuples = [ client.submit(convert_to_return_tuple, cp_arrays) for cp_arrays in result ] # Convert the futures to dask delayed objects so the tuples can be # split. nout=2 is passed since each tuple/iterable is a fixed length of 2. result_tuples = [dask.delayed(r, nout=2) for r in result_tuples] # Create the ddf and get the converged bool from the delayed objs. Use a # meta DataFrame to pass the expected dtypes for the DataFrame to prevent # another compute to determine them automatically. meta = cudf.DataFrame(columns=["vertex", "pagerank"]) meta = meta.astype({"pagerank": "float64", "vertex": vertex_dtype}) ddf = dask_cudf.from_delayed([t[0] for t in result_tuples], meta=meta).persist() converged = all(dask.compute(*[t[1] for t in result_tuples])) wait(ddf) # Wait until the inactive futures are released wait([(r.release(), c_r.release()) for r, c_r in zip(result, result_tuples)]) if input_graph.renumbered: ddf = input_graph.unrenumber(ddf, "vertex") if fail_on_nonconvergence: return ddf else: return (ddf, converged)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/common/mg_utils.py
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import os import numba.cuda # FIXME: this raft import breaks the library if ucx-py is # not available. They are necessary only when doing MG work. from cugraph.dask.common.read_utils import MissingUCXPy try: from raft_dask.common.utils import default_client except ImportError as err: # FIXME: Generalize since err.name is arr when # libnuma.so.1 is not available if err.name == "ucp" or err.name == "arr": default_client = MissingUCXPy() else: raise # FIXME: We currently look for the default client from dask, as such is the # if there is a dask client running without any GPU we will still try # to run MG using this client, it also implies that more work will be # required in order to run an MG Batch in Combination with mutli-GPU Graph def get_client(): try: client = default_client() except ValueError: client = None return client def prepare_worker_to_parts(data, client=None): if client is None: client = get_client() for placeholder, worker in enumerate(client.has_what().keys()): if worker not in data.worker_to_parts: data.worker_to_parts[worker] = [placeholder] return data def is_single_gpu(): ngpus = len(numba.cuda.gpus) if ngpus > 1: return False else: return True def get_visible_devices(): _visible_devices = os.environ.get("CUDA_VISIBLE_DEVICES") if _visible_devices is None: # FIXME: We assume that if the variable is unset there is only one GPU visible_devices = ["0"] else: visible_devices = _visible_devices.strip().split(",") return visible_devices
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/common/input_utils.py
# Copyright (c) 2020-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from collections.abc import Sequence from collections import OrderedDict from dask_cudf.core import DataFrame as dcDataFrame from dask_cudf.core import Series as daskSeries import cugraph.dask.comms.comms as Comms # FIXME: this raft import breaks the library if ucx-py is # not available. They are necessary only when doing MG work. from cugraph.dask.common.read_utils import MissingUCXPy try: from raft_dask.common.utils import get_client except ImportError as err: # FIXME: Generalize since err.name is arr when # libnuma.so.1 is not available if err.name == "ucp" or err.name == "arr": get_client = MissingUCXPy() else: raise from cugraph.dask.common.part_utils import _extract_partitions from dask.distributed import default_client from toolz import first from functools import reduce class DistributedDataHandler: """ Class to centralize distributed data management. Functionalities include: - Data colocation - Worker information extraction - GPU futures extraction, Additional functionality can be added as needed. This class **does not** contain the actual data, just the metadata necessary to handle it, including common pieces of code that need to be performed to call Dask functions. The constructor is not meant to be used directly, but through the factory method DistributedDataHandler.create """ def __init__( self, gpu_futures=None, workers=None, datatype=None, multiple=False, client=None ): self.client = get_client(client) self.gpu_futures = gpu_futures self.worker_to_parts = _workers_to_parts(gpu_futures) self.workers = workers self.datatype = datatype self.multiple = multiple self.worker_info = None self.total_rows = None self.max_vertex_id = None self.ranks = None self.parts_to_sizes = None self.local_data = None @classmethod def get_client(cls, client=None): return default_client() if client is None else client """ Class methods for initalization """ @classmethod def create(cls, data, client=None, batch_enabled=False): """ Creates a distributed data handler instance with the given distributed data set(s). Parameters ---------- data : dask.array, dask.dataframe, or unbounded Sequence of dask.array or dask.dataframe. client : dask.distributedClient """ client = cls.get_client(client) multiple = isinstance(data, Sequence) if isinstance(first(data) if multiple else data, (dcDataFrame, daskSeries)): datatype = "cudf" else: raise TypeError("Graph data must be dask-cudf dataframe") broadcast_worker = None if batch_enabled: worker_ranks = client.run(Comms.get_worker_id, Comms.get_session_id()) # The worker with 'rank = 0' must be the root of the broadcast. broadcast_worker = list(worker_ranks.keys())[ list(worker_ranks.values()).index(0) ] gpu_futures = client.sync( _extract_partitions, data, client, batch_enabled=batch_enabled, broadcast_worker=broadcast_worker, ) workers = tuple(OrderedDict.fromkeys(map(lambda x: x[0], gpu_futures))) return DistributedDataHandler( gpu_futures=gpu_futures, workers=workers, datatype=datatype, multiple=multiple, client=client, ) """ Methods to calculate further attributes """ def calculate_worker_and_rank_info(self, comms): self.worker_info = comms.worker_info(comms.worker_addresses) self.ranks = dict() for w, futures in self.worker_to_parts.items(): self.ranks[w] = self.worker_info[w]["rank"] def calculate_parts_to_sizes(self, comms=None, ranks=None): if self.worker_info is None and comms is not None: self.calculate_worker_and_rank_info(comms) self.total_rows = 0 self.parts_to_sizes = dict() parts = [ ( wf[0], self.client.submit( _get_rows, wf[1], self.multiple, workers=[wf[0]], pure=False ), ) for idx, wf in enumerate(self.worker_to_parts.items()) ] sizes = self.client.compute(parts, sync=True) for w, sizes_parts in sizes: sizes, total = sizes_parts self.parts_to_sizes[self.worker_info[w]["rank"]] = sizes self.total_rows += total def calculate_local_data(self, comms, by): if self.worker_info is None and comms is not None: self.calculate_worker_and_rank_info(comms) local_data = dict( [ ( self.worker_info[wf[0]]["rank"], self.client.submit(_get_local_data, wf[1], by, workers=[wf[0]]), ) for idx, wf in enumerate(self.worker_to_parts.items()) ] ) _local_data_dict = self.client.compute(local_data, sync=True) local_data_dict = {"edges": [], "offsets": [], "verts": []} max_vid = 0 for rank in range(len(_local_data_dict)): data = _local_data_dict[rank] local_data_dict["edges"].append(data[0]) if rank == 0: local_offset = 0 else: prev_data = _local_data_dict[rank - 1] local_offset = prev_data[1] + 1 local_data_dict["offsets"].append(local_offset) local_data_dict["verts"].append(data[1] - local_offset + 1) if data[2] > max_vid: max_vid = data[2] import numpy as np local_data_dict["edges"] = np.array(local_data_dict["edges"], dtype=np.int32) local_data_dict["offsets"] = np.array( local_data_dict["offsets"], dtype=np.int32 ) local_data_dict["verts"] = np.array(local_data_dict["verts"], dtype=np.int32) self.local_data = local_data_dict self.max_vertex_id = max_vid def _get_local_data(df, by): df = df[0] num_local_edges = len(df) local_by_max = df[by].iloc[-1] local_max = df[["src", "dst"]].max().max() return num_local_edges, local_by_max, local_max """ Internal methods, API subject to change """ def _workers_to_parts(futures): """ Builds an ordered dict mapping each worker to their list of parts :param futures: list of (worker, part) tuples :return: """ w_to_p_map = OrderedDict.fromkeys(Comms.get_workers()) for w, p in futures: if w_to_p_map[w] is None: w_to_p_map[w] = [] w_to_p_map[w].append(p) keys_to_delete = [w for (w, p) in w_to_p_map.items() if p is None] for k in keys_to_delete: del w_to_p_map[k] return w_to_p_map def _get_rows(objs, multiple): def get_obj(x): return x[0] if multiple else x total = list(map(lambda x: get_obj(x).shape[0], objs)) return total, reduce(lambda a, b: a + b, total) def get_mg_batch_data(dask_cudf_data, batch_enabled=False): data = DistributedDataHandler.create( data=dask_cudf_data, batch_enabled=batch_enabled ) return data def get_distributed_data(input_ddf): ddf = input_ddf comms = Comms.get_comms() data = DistributedDataHandler.create(data=ddf) if data.worker_info is None and comms is not None: data.calculate_worker_and_rank_info(comms) return data def get_vertex_partition_offsets(input_graph): import cudf renumber_vertex_count = input_graph.renumber_map.implementation.ddf.map_partitions( len ).compute() renumber_vertex_cumsum = renumber_vertex_count.cumsum() # Assume the input_graph edgelist was renumbered src_col_name = input_graph.renumber_map.renumbered_src_col_name vertex_dtype = input_graph.edgelist.edgelist_df[src_col_name].dtype vertex_partition_offsets = cudf.Series([0], dtype=vertex_dtype) vertex_partition_offsets = vertex_partition_offsets.append( cudf.Series(renumber_vertex_cumsum, dtype=vertex_dtype) ) return vertex_partition_offsets
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/common/part_utils.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import numpy as np from dask.distributed import futures_of, default_client, wait from toolz import first import collections import dask_cudf from dask.array.core import Array as daskArray from dask_cudf.core import DataFrame as daskDataFrame from dask_cudf.core import Series as daskSeries from functools import reduce import cugraph.dask.comms.comms as Comms from dask.delayed import delayed import cudf def workers_to_parts(futures): """ Builds an ordered dict mapping each worker to their list of parts :param futures: list of (worker, part) tuples :return: """ w_to_p_map = collections.OrderedDict() for w, p in futures: if w not in w_to_p_map: w_to_p_map[w] = [] w_to_p_map[w].append(p) return w_to_p_map def _func_get_rows(df): return df.shape[0] def parts_to_ranks(client, worker_info, part_futures): """ Builds a list of (rank, size) tuples of partitions :param worker_info: dict of {worker, {"rank": rank }}. Note: \ This usually comes from the underlying communicator :param part_futures: list of (worker, future) tuples :return: [(part, size)] in the same order of part_futures """ futures = [ ( worker_info[wf[0]]["rank"], client.submit(_func_get_rows, wf[1], workers=[wf[0]], pure=False), ) for idx, wf in enumerate(part_futures) ] sizes = client.compute(list(map(lambda x: x[1], futures)), sync=True) total = reduce(lambda a, b: a + b, sizes) return [(futures[idx][0], size) for idx, size in enumerate(sizes)], total def persist_distributed_data(dask_df, client): client = default_client() if client is None else client worker_addresses = Comms.get_workers() _keys = dask_df.__dask_keys__() worker_dict = {} for i, key in enumerate(_keys): worker_dict[key] = tuple([worker_addresses[i]]) persisted = client.persist(dask_df, workers=worker_dict) parts = futures_of(persisted) return parts def _create_empty_dask_df_future(meta_df, client, worker): df_future = client.scatter(meta_df.head(0), workers=[worker]) wait(df_future) return [df_future] def get_persisted_df_worker_map(dask_df, client): ddf_keys = futures_of(dask_df) output_map = {} for w, w_keys in client.has_what().items(): output_map[w] = [ddf_k for ddf_k in ddf_keys if ddf_k.key in w_keys] if len(output_map[w]) == 0: output_map[w] = _create_empty_dask_df_future(dask_df._meta, client, w) return output_map def _chunk_lst(ls, num_parts): return [ls[i::num_parts] for i in range(num_parts)] def persist_dask_df_equal_parts_per_worker( dask_df, client, return_type="dask_cudf.DataFrame" ): """ Persist dask_df with equal parts per worker Args: dask_df: dask_cudf.DataFrame client: dask.distributed.Client return_type: str, "dask_cudf.DataFrame" or "dict" Returns: persisted_keys: dict of {worker: [persisted_keys]} """ if return_type not in ["dask_cudf.DataFrame", "dict"]: raise ValueError("return_type must be either 'dask_cudf.DataFrame' or 'dict'") ddf_keys = dask_df.to_delayed() workers = client.scheduler_info()["workers"].keys() ddf_keys_ls = _chunk_lst(ddf_keys, len(workers)) persisted_keys_d = {} for w, ddf_k in zip(workers, ddf_keys_ls): persisted_keys_d[w] = client.compute( ddf_k, workers=w, allow_other_workers=False, pure=False ) persisted_keys_ls = [ item for sublist in persisted_keys_d.values() for item in sublist ] wait(persisted_keys_ls) if return_type == "dask_cudf.DataFrame": dask_df = dask_cudf.from_delayed( persisted_keys_ls, meta=dask_df._meta ).persist() wait(dask_df) return dask_df return persisted_keys_d def get_length_of_parts(persisted_keys_d, client): """ Get the length of each partition Args: persisted_keys_d: dict of {worker: [persisted_keys]} client: dask.distributed.Client Returns: length_of_parts: dict of {worker: [length_of_parts]} """ length_of_parts = {} for w, p_keys in persisted_keys_d.items(): length_of_parts[w] = [ client.submit( len, p_key, pure=False, workers=[w], allow_other_workers=False ) for p_key in p_keys ] for w, len_futures in length_of_parts.items(): length_of_parts[w] = client.gather(len_futures) return length_of_parts async def _extract_partitions( dask_obj, client=None, batch_enabled=False, broadcast_worker=None ): client = default_client() if client is None else client worker_list = Comms.get_workers() # dask.dataframe or dask.array if isinstance(dask_obj, (daskDataFrame, daskArray, daskSeries)): if batch_enabled: persisted = client.persist(dask_obj, workers=broadcast_worker) else: # repartition the 'dask_obj' to get as many partitions as there # are workers dask_obj = dask_obj.repartition(npartitions=len(worker_list)) # Have the first n workers persisting the n partitions # Ideally, there would be as many partitions as there are workers persisted = [ client.persist(dask_obj.get_partition(p), workers=w) for p, w in enumerate(worker_list[: dask_obj.npartitions]) ] # Persist empty dataframe/series with the remaining workers if # there are less partitions than workers if dask_obj.npartitions < len(worker_list): # The empty df should have the same column names and dtypes as # dask_obj if isinstance(dask_obj, dask_cudf.DataFrame): empty_df = cudf.DataFrame(columns=list(dask_obj.columns)) empty_df = empty_df.astype( dict(zip(dask_obj.columns, dask_obj.dtypes)) ) else: empty_df = cudf.Series(dtype=dask_obj.dtype) for p, w in enumerate(worker_list[dask_obj.npartitions :]): empty_ddf = dask_cudf.from_cudf(empty_df, npartitions=1) persisted.append(client.persist(empty_ddf, workers=w)) parts = futures_of(persisted) # iterable of dask collections (need to colocate them) elif isinstance(dask_obj, collections.abc.Sequence): # NOTE: We colocate (X, y) here by zipping delayed # n partitions of them as (X1, y1), (X2, y2)... # and asking client to compute a single future for # each tuple in the list. dela = [np.asarray(d.to_delayed()) for d in dask_obj] # TODO: ravel() is causing strange behavior w/ delayed Arrays which are # not yet backed by futures. Need to investigate this behavior. # ref: https://github.com/rapidsai/cuml/issues/2045 raveled = [d.flatten() for d in dela] parts = client.compute([p for p in zip(*raveled)]) await wait(parts) key_to_part = [(part.key, part) for part in parts] who_has = await client.who_has(parts) return [(first(who_has[key]), part) for key, part in key_to_part] def create_dict(futures): w_to_p_map = collections.OrderedDict() for w, k, p in futures: if w not in w_to_p_map: w_to_p_map[w] = [] w_to_p_map[w].append([p, k]) return w_to_p_map def set_global_index(df, cumsum): df.index = df.index + cumsum df.index = df.index.astype("int64") return df def get_cumsum(df, by): return df[by].value_counts(sort=False).cumsum() def repartition(ddf, cumsum): # Calculate new optimal divisions and repartition the data # for load balancing. import math npartitions = ddf.npartitions count = math.ceil(len(ddf) / npartitions) new_divisions = [0] move_count = 0 i = npartitions - 2 for i in range(npartitions - 1): search_val = count - move_count index = cumsum[i].searchsorted(search_val) if index == len(cumsum[i]): index = -1 elif index > 0: left = cumsum[i].iloc[index - 1] right = cumsum[i].iloc[index] index -= search_val - left < right - search_val new_divisions.append(new_divisions[i] + cumsum[i].iloc[index] + move_count) move_count = cumsum[i].iloc[-1] - cumsum[i].iloc[index] new_divisions.append(new_divisions[i + 1] + cumsum[-1].iloc[-1] + move_count - 1) return ddf.repartition(divisions=tuple(new_divisions)) def load_balance_func(ddf_, by, client=None): # Load balances the sorted dask_cudf DataFrame. # Input is a dask_cudf dataframe ddf_ which is sorted by # the column name passed as the 'by' argument. client = default_client() if client is None else client parts = persist_distributed_data(ddf_, client) wait(parts) who_has = client.who_has(parts) key_to_part = [(part.key, part) for part in parts] gpu_fututres = [ (first(who_has[key]), part.key[1], part) for key, part in key_to_part ] worker_to_data = create_dict(gpu_fututres) # Calculate cumulative sum in each dataframe partition cumsum_parts = [ client.submit(get_cumsum, wf[1][0][0], by, workers=[wf[0]]).result() for idx, wf in enumerate(worker_to_data.items()) ] num_rows = [] for cumsum in cumsum_parts: num_rows.append(cumsum.iloc[-1]) # Calculate current partition divisions. divisions = [sum(num_rows[0:x:1]) for x in range(0, len(num_rows) + 1)] divisions[-1] = divisions[-1] - 1 divisions = tuple(divisions) # Set global index from 0 to len(dask_cudf_dataframe) so that global # indexing of divisions can be used for repartitioning. futures = [ client.submit( set_global_index, wf[1][0][0], divisions[wf[1][0][1]], workers=[wf[0]] ) for idx, wf in enumerate(worker_to_data.items()) ] wait(futures) ddf = dask_cudf.from_delayed(futures) ddf.divisions = divisions # Repartition the data ddf = repartition(ddf, cumsum_parts) return ddf def concat_dfs(df_list): """ Concat a list of cudf dataframes. """ return cudf.concat(df_list) def get_delayed_dict(ddf): """ Returns a dicitionary with the dataframe tasks as keys and the dataframe delayed objects as values. """ df_delayed = {} for delayed_obj in ddf.to_delayed(): df_delayed[delayed_obj.key] = delayed_obj return df_delayed def concat_within_workers(client, ddf): """ Concats all partitions within workers without transfers. """ df_delayed = get_delayed_dict(ddf) result = [] for worker, tasks in client.has_what().items(): worker_task_list = [] for task in list(tasks): if task in df_delayed: worker_task_list.append(df_delayed[task]) concat_tasks = delayed(concat_dfs)(worker_task_list) result.append(client.persist(collections=concat_tasks, workers=worker)) return dask_cudf.from_delayed(result)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/common/read_utils.py
# Copyright (c) 2019-2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. def get_n_workers(): from dask.distributed import default_client client = default_client() return len(client.scheduler_info()["workers"]) def get_chunksize(input_path): """ Calculate the appropriate chunksize for dask_cudf.read_csv to get a number of partitions equal to the number of GPUs. Examples -------- >>> import cugraph.dask as dcg >>> chunksize = dcg.get_chunksize(datasets_path / 'netscience.csv') """ import os from glob import glob import math input_files = sorted(glob(str(input_path))) if len(input_files) == 1: size = os.path.getsize(input_files[0]) chunksize = math.ceil(size / get_n_workers()) else: size = [os.path.getsize(_file) for _file in input_files] chunksize = max(size) return chunksize class MissingUCXPy: def __call__(self, *args, **kwargs): raise ModuleNotFoundError( "ucx-py could not be imported but is required for MG operations" )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/common/__init__.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/structure/replication.pyx
# Copyright (c) 2020-2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # cython: profile=False # distutils: language = c++ # cython: embedsignature = True # cython: language_level = 3 from libc.stdint cimport uintptr_t from cugraph.structure cimport utils as c_utils from cugraph.structure.graph_primtypes cimport * from libc.stdint cimport uintptr_t import cudf import dask.distributed as dd from cugraph.dask.common.input_utils import get_mg_batch_data import dask_cudf import cugraph.dask.comms.comms as Comms import cugraph.dask.common.mg_utils as mg_utils import numpy as np def replicate_cudf_dataframe(cudf_dataframe, client=None, comms=None): if type(cudf_dataframe) is not cudf.DataFrame: raise TypeError("Expected a cudf.Series to replicate") client = mg_utils.get_client() if client is None else client comms = Comms.get_comms() if comms is None else comms dask_cudf_df = dask_cudf.from_cudf(cudf_dataframe, npartitions=1) df_length = len(dask_cudf_df) _df_data = get_mg_batch_data(dask_cudf_df, batch_enabled=True) df_data = mg_utils.prepare_worker_to_parts(_df_data, client) workers_to_futures = {worker: client.submit(_replicate_cudf_dataframe, (data, cudf_dataframe.columns.values, cudf_dataframe.dtypes, df_length), comms.sessionId, workers=[worker]) for (worker, data) in df_data.worker_to_parts.items()} dd.wait(workers_to_futures) return workers_to_futures def _replicate_cudf_dataframe(input_data, session_id): cdef uintptr_t c_handle = <uintptr_t> NULL cdef uintptr_t c_series = <uintptr_t> NULL result = None handle = Comms.get_handle(session_id) c_handle = <uintptr_t>handle.getHandle() _data, columns, dtypes, df_length = input_data data = _data[0] has_data = type(data) is cudf.DataFrame series = None df_data = {} for idx, column in enumerate(columns): if has_data: series = data[column] else: dtype = dtypes[idx] series = cudf.Series(np.zeros(df_length), dtype=dtype) df_data[column] = series c_series = series.__cuda_array_interface__['data'][0] comms_bcast(c_handle, c_series, df_length, series.dtype) if has_data: result = data else: result = cudf.DataFrame(data=df_data) return result def replicate_cudf_series(cudf_series, client=None, comms=None): if type(cudf_series) is not cudf.Series: raise TypeError("Expected a cudf.Series to replicate") client = mg_utils.get_client() if client is None else client comms = Comms.get_comms() if comms is None else comms dask_cudf_series = dask_cudf.from_cudf(cudf_series, npartitions=1) series_length = len(dask_cudf_series) _series_data = get_mg_batch_data(dask_cudf_series, batch_enabled=True) series_data = mg_utils.prepare_worker_to_parts(_series_data) dtype = cudf_series.dtype workers_to_futures = {worker: client.submit(_replicate_cudf_series, (data, series_length, dtype), comms.sessionId, workers=[worker]) for (worker, data) in series_data.worker_to_parts.items()} dd.wait(workers_to_futures) return workers_to_futures def _replicate_cudf_series(input_data, session_id): cdef uintptr_t c_handle = <uintptr_t> NULL cdef uintptr_t c_result = <uintptr_t> NULL result = None handle = Comms.get_handle(session_id) c_handle = <uintptr_t>handle.getHandle() (_data, size, dtype) = input_data data = _data[0] has_data = type(data) is cudf.Series if has_data: result = data else: result = cudf.Series(np.zeros(size), dtype=dtype) c_result = result.__cuda_array_interface__['data'][0] comms_bcast(c_handle, c_result, size, dtype) return result cdef comms_bcast(uintptr_t handle, uintptr_t value_ptr, size_t count, dtype): if dtype == np.int32: c_utils.comms_bcast((<handle_t*> handle)[0], <int*> value_ptr, count) elif dtype == np.float32: c_utils.comms_bcast((<handle_t*> handle)[0], <float*> value_ptr, count) elif dtype == np.float64: c_utils.comms_bcast((<handle_t*> handle)[0], <double*> value_ptr, count) else: raise TypeError("Unsupported broadcast type")
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/structure/CMakeLists.txt
# ============================================================================= # Copyright (c) 2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except # in compliance with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software distributed under the License # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express # or implied. See the License for the specific language governing permissions and limitations under # the License. # ============================================================================= set(cython_sources replication.pyx) set(linked_libraries cugraph::cugraph) rapids_cython_create_modules( CXX SOURCE_FILES "${cython_sources}" LINKED_LIBRARIES "${linked_libraries}" MODULE_PREFIX structure_ ASSOCIATED_TARGETS cugraph )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/structure/mg_property_graph.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import cudf import cupy import cugraph import dask_cudf import cugraph.dask as dcg from cugraph.utilities.utils import import_optional, create_list_series_from_2d_ar from typing import Union pd = import_optional("pandas") class EXPERIMENTAL__MGPropertySelection: """ Instances of this class are returned from the PropertyGraph.select_*() methods and can be used by the PropertyGraph.extract_subgraph() method to extract a Graph containing vertices and edges with only the selected properties. """ def __init__(self, vertex_selection_series=None, edge_selection_series=None): self.vertex_selections = vertex_selection_series self.edge_selections = edge_selection_series def __add__(self, other): """ Add either the vertex_selections, edge_selections, or both to this instance from "other" if either are not already set. """ vs = self.vertex_selections if vs is None: vs = other.vertex_selections es = self.edge_selections if es is None: es = other.edge_selections return EXPERIMENTAL__MGPropertySelection(vs, es) # FIXME: remove leading __ when no longer experimental class EXPERIMENTAL__MGPropertyGraph: """ Class which stores vertex and edge properties that can be used to construct Graphs from individual property selections and used later to annotate graph algorithm results with corresponding properties. """ # column name constants used in internal DataFrames vertex_col_name = "_VERTEX_" src_col_name = "_SRC_" dst_col_name = "_DST_" type_col_name = "_TYPE_" edge_id_col_name = "_EDGE_ID_" weight_col_name = "_WEIGHT_" _default_type_name = "" def __init__(self, num_workers=None): # The dataframe containing the properties for each vertex. # Each vertex occupies a row, and individual properties are maintained # in individual columns. The table contains a column for each property # of each vertex. If a vertex does not contain a property, it will have # a NaN value in that property column. Each vertex will also have a # "type_name" that can be assigned by the caller to describe the type # of the vertex for a given application domain. If no type_name is # provided, the default type_name is "". # Example: # vertex | type_name | propA | propB | propC # ------------------------------------------ # 3 | "user" | 22 | NaN | 11 # 88 | "service" | NaN | 3.14 | 21 # 9 | "" | NaN | NaN | 2 self.__vertex_prop_dataframe = None # The dataframe containing the properties for each edge. # The description is identical to the vertex property dataframe, except # edges are identified by ordered pairs of vertices (src and dst). # Example: # src | dst | type_name | propA | propB | propC # --------------------------------------------- # 3 | 88 | "started" | 22 | NaN | 11 # 88 | 9 | "called" | NaN | 3.14 | 21 # 9 | 88 | "" | NaN | NaN | 2 self.__edge_prop_dataframe = None # The var:value dictionaries used during evaluation of filter/query # expressions for vertices and edges. These dictionaries contain # entries for each column name in their respective dataframes which # are mapped to instances of PropertyColumn objects. # # When filter/query expressions are evaluated, PropertyColumn objects # are used in place of DataFrame columns in order to support string # comparisons when cuDF DataFrames are used. This approach also allows # expressions to contain var names that can be used in expressions that # are different than those in the actual internal tables, allowing for # the tables to contain additional or different column names than what # can be used in expressions. # # Example: "type_name == 'user' & propC > 10" # # The above would be evaluated and "type_name" and "propC" would be # PropertyColumn instances which support specific operators used in # queries. self.__vertex_prop_eval_dict = {} self.__edge_prop_eval_dict = {} self.__dataframe_type = dask_cudf.DataFrame self.__series_type = dask_cudf.Series # The dtypes for each column in each DataFrame. This is required since # merge operations can often change the dtypes to accommodate NaN # values (eg. int64 to float64, since NaN is a float). self.__vertex_prop_dtypes = {} self.__edge_prop_dtypes = {} # Lengths of the properties that are vectors self.__vertex_vector_property_lengths = {} self.__edge_vector_property_lengths = {} # Add unique edge IDs to the __edge_prop_dataframe by simply # incrementing this counter. Remains None if user provides edge IDs. self.__last_edge_id = None # Are edge IDs automatically generated sequentially by PG (True), # provided by the user (False), or no edges added yet (None). self.__is_edge_id_autogenerated = None # Cached property values self.__num_vertices = None self.__vertex_type_value_counts = None self.__edge_type_value_counts = None # number of gpu's to use if num_workers is None: self.__num_workers = dcg.get_n_workers() else: self.__num_workers = num_workers def _build_from_components( self, *, vertex_prop_dataframe, edge_prop_dataframe, dataframe_type, series_type, vertex_prop_dtypes, edge_prop_dtypes, vertex_vector_property_lengths, edge_vector_property_lengths, last_edge_id, is_edge_id_autogenerated, # Computable vertex_prop_eval_dict=None, edge_prop_eval_dict=None, # Cached properties num_vertices=None, vertex_type_value_counts=None, edge_type_value_counts=None, # MG-specific num_workers=None, ): """Backdoor to populate a PropertyGraph from existing data. Use only if you know what you're doing. """ self.__vertex_prop_dataframe = vertex_prop_dataframe self.__edge_prop_dataframe = edge_prop_dataframe if vertex_prop_eval_dict is None: vertex_prop_eval_dict = {} if vertex_prop_dataframe is not None: self._update_eval_dict( vertex_prop_eval_dict, vertex_prop_dataframe, self.vertex_col_name ) self.__vertex_prop_eval_dict = vertex_prop_eval_dict if edge_prop_eval_dict is None: edge_prop_eval_dict = {} if edge_prop_dataframe is not None: self._update_eval_dict( edge_prop_eval_dict, edge_prop_dataframe, self.edge_id_col_name ) self.__edge_prop_eval_dict = edge_prop_eval_dict self.__dataframe_type = dataframe_type self.__series_type = series_type self.__vertex_prop_dtypes = vertex_prop_dtypes self.__edge_prop_dtypes = edge_prop_dtypes self.__vertex_vector_property_lengths = vertex_vector_property_lengths self.__edge_vector_property_lengths = edge_vector_property_lengths self.__last_edge_id = last_edge_id self.__is_edge_id_autogenerated = is_edge_id_autogenerated self.__num_vertices = num_vertices self.__vertex_type_value_counts = vertex_type_value_counts self.__edge_type_value_counts = edge_type_value_counts if num_workers is None: self.__num_workers = dcg.get_n_workers() else: self.__num_workers = num_workers # PropertyGraph read-only attributes @property def edges(self): if self.__edge_prop_dataframe is not None: return self.__edge_prop_dataframe[ [self.src_col_name, self.dst_col_name] ].reset_index() return None @property def vertex_property_names(self): if self.__vertex_prop_dataframe is not None: props = list(self.__vertex_prop_dataframe.columns) props.remove(self.type_col_name) # should "type" be removed? return props return [] @property def edge_property_names(self): if self.__edge_prop_dataframe is not None: props = list(self.__edge_prop_dataframe.columns) props.remove(self.src_col_name) props.remove(self.dst_col_name) props.remove(self.type_col_name) # should "type" be removed? if self.weight_col_name in props: props.remove(self.weight_col_name) return props return [] @property def vertex_types(self): """The set of vertex type names""" value_counts = self._vertex_type_value_counts if value_counts is None: names = set() elif self.__series_type is dask_cudf.Series: names = set(value_counts.index.to_arrow().to_pylist()) else: names = set(value_counts.index) default = self._default_type_name if default not in names and self.get_num_vertices(default) > 0: # include "" from vertices that only exist in edge data names.add(default) return names @property def edge_types(self): """The set of edge type names""" value_counts = self._edge_type_value_counts if value_counts is None: return set() elif self.__series_type is dask_cudf.Series: return set(value_counts.index.to_arrow().to_pylist()) else: return set(value_counts.index) # PropertyGraph read-only attributes for debugging @property def _vertex_prop_dataframe(self): return self.__vertex_prop_dataframe @property def _edge_prop_dataframe(self): return self.__edge_prop_dataframe @property def _vertex_type_value_counts(self): """A Series of the counts of types in __vertex_prop_dataframe""" if self.__vertex_prop_dataframe is None: return if self.__vertex_type_value_counts is None: # Types should all be strings; what should we do if we see NaN? self.__vertex_type_value_counts = ( self.__vertex_prop_dataframe[self.type_col_name] .value_counts(sort=False, dropna=False) .compute() ) return self.__vertex_type_value_counts @property def _edge_type_value_counts(self): """A Series of the counts of types in __edge_prop_dataframe""" if self.__edge_prop_dataframe is None: return if self.__edge_type_value_counts is None: # Types should all be strings; what should we do if we see NaN? self.__edge_type_value_counts = ( self.__edge_prop_dataframe[self.type_col_name] .value_counts(sort=False, dropna=False) .compute() ) return self.__edge_type_value_counts def get_num_vertices(self, type=None, *, include_edge_data=True): """Return the number of all vertices or vertices of a given type. Parameters ---------- type : string, optional If type is None (the default), return the total number of vertices, otherwise return the number of vertices of the specified type. include_edge_data : bool (default True) If True, include vertices that were added in vertex and edge data. If False, only include vertices that were added in vertex data. Note that vertices that only exist in edge data are assumed to have the default type. See Also -------- PropertyGraph.get_num_edges """ if type is None: if not include_edge_data: if self.__vertex_prop_dataframe is None: return 0 return len(self.__vertex_prop_dataframe) if self.__num_vertices is not None: return self.__num_vertices self.__num_vertices = 0 vert_sers = self.__get_all_vertices_series() if vert_sers: if self.__series_type is dask_cudf.Series: vert_count = dask_cudf.concat( vert_sers, ignore_index=True ).nunique() self.__num_vertices = vert_count.compute() return self.__num_vertices value_counts = self._vertex_type_value_counts if type == self._default_type_name and include_edge_data: # The default type, "", can refer to both vertex and edge data if self.__vertex_prop_dataframe is None: return self.get_num_vertices() return ( self.get_num_vertices() - len(self.__vertex_prop_dataframe) + (value_counts[type] if type in value_counts else 0) ) if self.__vertex_prop_dataframe is None: return 0 return value_counts[type] if type in value_counts else 0 def get_num_edges(self, type=None): """Return the number of all edges or edges of a given type. Parameters ---------- type : string, optional If type is None (the default), return the total number of edges, otherwise return the number of edges of the specified type. See Also -------- PropertyGraph.get_num_vertices """ if type is None: if self.__edge_prop_dataframe is not None: return len(self.__edge_prop_dataframe) else: return 0 if self.__edge_prop_dataframe is None: return 0 value_counts = self._edge_type_value_counts return value_counts[type] if type in value_counts else 0 def get_vertices(self, selection=None): """ Return a Series containing the unique vertex IDs contained in both the vertex and edge property data. """ vert_sers = self.__get_all_vertices_series() if vert_sers: if self.__series_type is dask_cudf.Series: return ( dask_cudf.concat(vert_sers, ignore_index=True) .unique() .sort_values() ) else: raise TypeError("dataframe must be a CUDF Dask dataframe.") return self.__series_type() def vertices_ids(self): """ Alias for get_vertices() """ return self.get_vertices() def vertex_types_from_numerals( self, nums: Union[cudf.Series, pd.Series] ) -> Union[cudf.Series, pd.Series]: """ Returns the string vertex type names given the numeric category labels. Note: Does not accept or return dask_cudf Series. Parameters ---------- nums: Union[cudf.Series, pandas.Series] (Required) The list of numeric category labels to convert. Returns ------- Union[cudf.Series, pd.Series] The string type names converted from the input numerals. """ return ( self.__vertex_prop_dataframe[self.type_col_name] .dtype.categories.to_series() .iloc[nums] .reset_index(drop=True) ) def edge_types_from_numerals( self, nums: Union[cudf.Series, pd.Series] ) -> Union[cudf.Series, pd.Series]: """ Returns the string edge type names given the numeric category labels. Note: Does not accept or return dask_cudf Series. Parameters ---------- nums: Union[cudf.Series, pandas.Series] (Required) The list of numeric category labels to convert. Returns ------- Union[cudf.Series, pd.Series] The string type names converted from the input numerals. """ return ( self.__edge_prop_dataframe[self.type_col_name] .dtype.categories.to_series() .iloc[nums] .reset_index(drop=True) ) def add_vertex_data( self, dataframe, vertex_col_name, type_name=None, property_columns=None, vector_properties=None, vector_property=None, ): """ Add a dataframe describing vertex properties to the PropertyGraph. Parameters ---------- dataframe : DataFrame-compatible instance A DataFrame instance with a compatible Pandas-like DataFrame interface. vertex_col_name : string The column name that contains the values to be used as vertex IDs, or the name of the index if the index is vertex IDs. Specifying the index may be more efficient, and will be the most efficient if the index of vertex IDs is already sorted. type_name : string The name to be assigned to the type of property being added. For example, if dataframe contains data about users, type_name might be "users". If not specified, the type of properties will be added as the empty string, "". property_columns : list of strings List of column names in dataframe to be added as properties. All other columns in dataframe will be ignored. If not specified, all columns in dataframe are added. vector_properties : dict of string to list of strings, optional A dict of vector properties to create from columns in the dataframe. Each vector property stores an array for each vertex. The dict keys are the new vector property names, and the dict values should be Python lists of column names from which to create the vector property. Columns used to create vector properties won't be added to the property graph by default, but may be included as properties by including them in the property_columns argument. Use ``MGPropertyGraph.vertex_vector_property_to_array`` to convert a vertex vector property to an array. vector_property : string, optional If provided, all columns not included in other arguments will be used to create a vector property with the given name. This is often used for convenience instead of ``vector_properties`` when all input properties should be converted to a vector property. Returns ------- None Examples -------- >>> """ if type(dataframe) is not dask_cudf.DataFrame: raise TypeError("dataframe must be a Dask dataframe.") if vertex_col_name not in dataframe.columns: if vertex_col_name != dataframe.index.name: raise ValueError( f"{vertex_col_name} is not a column in or the index name of " f"dataframe: {dataframe.columns}" ) index_is_set = True else: index_is_set = False if type_name is not None and not isinstance(type_name, str): raise TypeError(f"type_name must be a string, got: {type(type_name)}") if type_name is None: type_name = self._default_type_name if property_columns: if type(property_columns) is not list: raise TypeError( f"property_columns must be a list, got: {type(property_columns)}" ) invalid_columns = set(property_columns).difference(dataframe.columns) if invalid_columns: raise ValueError( "property_columns contains column(s) not found in dataframe: " f"{list(invalid_columns)}" ) existing_vectors = ( set(property_columns) & self.__vertex_vector_property_lengths.keys() ) if existing_vectors: raise ValueError( "Non-vector property columns cannot be added to existing " f"vector properties: {', '.join(sorted(existing_vectors))}" ) TCN = self.type_col_name if vector_properties is not None: invalid_keys = {self.vertex_col_name, TCN} if property_columns: invalid_keys.update(property_columns) self._check_vector_properties( dataframe, vector_properties, self.__vertex_vector_property_lengths, invalid_keys, ) if vector_property is not None: invalid_keys = {self.vertex_col_name, TCN, vertex_col_name} if property_columns: invalid_keys.update(property_columns) if vector_properties: invalid_keys.update(*vector_properties.values()) d = { vector_property: [ col for col in dataframe.columns if col not in invalid_keys ] } invalid_keys.remove(vertex_col_name) self._check_vector_properties( dataframe, d, self.__vertex_vector_property_lengths, invalid_keys, ) # Update vector_properties, but don't mutate the original if vector_properties is not None: d.update(vector_properties) vector_properties = d # Clear the cached values related to the number of vertices since more # could be added in this method. self.__num_vertices = None self.__vertex_type_value_counts = None # Could update instead # Add `type_name` to the TYPE categorical dtype if necessary is_first_data = self.__vertex_prop_dataframe is None if is_first_data: # Initialize the __vertex_prop_dataframe using the same # type as the incoming dataframe. temp_dataframe = cudf.DataFrame(columns=[self.vertex_col_name, TCN]) self.__vertex_prop_dataframe = dask_cudf.from_cudf( temp_dataframe, npartitions=self.__num_workers ) # Initialize the new columns to the same dtype as the appropriate # column in the incoming dataframe, since the initial merge may not # result in the same dtype. (see # https://github.com/rapidsai/cudf/issues/9981) if not index_is_set: self.__update_dataframe_dtypes( self.__vertex_prop_dataframe, {self.vertex_col_name: dataframe[vertex_col_name].dtype}, ) self.__vertex_prop_dataframe = self.__vertex_prop_dataframe.set_index( self.vertex_col_name ) # Use categorical dtype for the type column if self.__series_type is dask_cudf.Series: cat_class = cudf.CategoricalDtype else: cat_class = pd.CategoricalDtype cat_dtype = cat_class([type_name], ordered=False) else: cat_dtype = self.__update_categorical_dtype( self.__vertex_prop_dataframe, TCN, type_name ) # NOTE: This copies the incoming DataFrame in order to add the new # columns. The copied DataFrame is then merged (another copy) and then # deleted when out-of-scope. # Ensure that both the predetermined vertex ID column name and vertex # type column name are present for proper merging. tmp_df = dataframe.copy() if not index_is_set: tmp_df[self.vertex_col_name] = tmp_df[vertex_col_name] elif tmp_df.index.name != self.vertex_col_name: tmp_df.index = tmp_df.index.rename(self.vertex_col_name) # FIXME: handle case of a type_name column already being in tmp_df # FIXME: We should do categorization first # Related issue: https://github.com/rapidsai/cugraph/issues/2903 tmp_df[TCN] = type_name tmp_df[TCN] = tmp_df[TCN].astype(cat_dtype) if property_columns: # all columns column_names_to_drop = set(tmp_df.columns) # remove the ones to keep column_names_to_drop.difference_update( property_columns + [self.vertex_col_name, TCN] ) else: column_names_to_drop = {vertex_col_name} if vector_properties: # Drop vector property source columns by default more_to_drop = set().union(*vector_properties.values()) if property_columns is not None: more_to_drop.difference_update(property_columns) column_names_to_drop |= more_to_drop column_names_to_drop -= vector_properties.keys() tmp_df = self._create_vector_properties(tmp_df, vector_properties) if index_is_set: column_names_to_drop -= {self.vertex_col_name, vertex_col_name} tmp_df = tmp_df.drop(labels=column_names_to_drop, axis=1) # Save the original dtypes for each new column so they can be restored # prior to constructing subgraphs (since column dtypes may get altered # during merge to accommodate NaN values). if is_first_data: new_col_info = tmp_df.dtypes.items() else: new_col_info = self.__get_new_column_dtypes( tmp_df, self.__vertex_prop_dataframe ) self.__vertex_prop_dtypes.update(new_col_info) tmp_df = tmp_df.persist() if not index_is_set: tmp_df = tmp_df.set_index(self.vertex_col_name).persist() self.__update_dataframe_dtypes(tmp_df, self.__vertex_prop_dtypes) if is_first_data: self.__vertex_prop_dataframe = tmp_df else: # Join on vertex ids (the index) # TODO: can we automagically determine when we to use concat? df = self.__vertex_prop_dataframe.join( tmp_df, how="outer", rsuffix="_NEW_", # npartitions=self.__num_workers # TODO: see how this behaves ).persist() cols = self.__vertex_prop_dataframe.columns.intersection( tmp_df.columns ).to_list() rename_cols = {f"{col}_NEW_": col for col in cols} new_cols = list(rename_cols) sub_df = df[new_cols].rename(columns=rename_cols) # This only adds data--it doesn't replace existing data df = df.drop(columns=new_cols).fillna(sub_df).persist() if df.npartitions > 4 * self.__num_workers: # TODO: better understand behavior of npartitions argument in join df = df.repartition(npartitions=2 * self.__num_workers).persist() self.__vertex_prop_dataframe = df # Update the vertex eval dict with the latest column instances self._update_eval_dict( self.__vertex_prop_eval_dict, self.__vertex_prop_dataframe, self.vertex_col_name, ) def _update_eval_dict(self, eval_dict, df, index_name): # Update the vertex eval dict with the latest column instances latest = {n: df[n] for n in df.columns} eval_dict.update(latest) eval_dict[index_name] = df.index def get_vertex_data(self, vertex_ids=None, types=None, columns=None): """ Return a dataframe containing vertex properties for only the specified vertex_ids, columns, and/or types, or all vertex IDs if not specified. """ if self.__vertex_prop_dataframe is not None: df = self.__vertex_prop_dataframe if vertex_ids is not None: if isinstance(vertex_ids, int): vertex_ids = [vertex_ids] try: df = df.loc[vertex_ids] except TypeError: raise TypeError( "vertex_ids needs to be a list-like type " f"compatible with DataFrame.loc[], got {type(vertex_ids)}" ) if types is not None: if isinstance(types, str): df_mask = df[self.type_col_name] == types else: df_mask = df[self.type_col_name].isin(types) df = df.loc[df_mask] # The "internal" pG.vertex_col_name and pG.type_col_name columns # are also included/added since they are assumed to be needed by # the caller. if columns is not None: # FIXME: invalid columns will result in a KeyError, should a # check be done here and a more PG-specific error raised? df = df[[self.type_col_name] + columns] df_out = df.reset_index() # Preserve the dtype (vertex id type) to avoid cugraph algorithms # throwing errors due to a dtype mismatch index_dtype = self.__vertex_prop_dataframe.index.dtype df_out.index = df_out.index.astype(index_dtype) return df_out return None def add_edge_data( self, dataframe, vertex_col_names, edge_id_col_name=None, type_name=None, property_columns=None, vector_properties=None, vector_property=None, ): """ Add a dataframe describing edge properties to the PropertyGraph. Parameters ---------- dataframe : DataFrame-compatible instance A DataFrame instance with a compatible Pandas-like DataFrame interface. vertex_col_names : list of strings The column names that contain the values to be used as the source and destination vertex IDs for the edges. edge_id_col_name : string, optional The column name that contains the values to be used as edge IDs, or the name of the index if the index is edge IDs. Specifying the index may be more efficient, and will be the most efficient if the index of edge IDs is already sorted. If unspecified, edge IDs will be automatically assigned. Currently, all edge data must be added with the same method: either with automatically generated IDs, or from user-provided edge IDs. type_name : string The name to be assigned to the type of property being added. For example, if dataframe contains data about transactions, type_name might be "transactions". If not specified, the type of properties will be added as the empty string "". property_columns : list of strings List of column names in dataframe to be added as properties. All other columns in dataframe will be ignored. If not specified, all columns in dataframe are added. vector_properties : dict of string to list of strings, optional A dict of vector properties to create from columns in the dataframe. Each vector property stores an array for each edge. The dict keys are the new vector property names, and the dict values should be Python lists of column names from which to create the vector property. Columns used to create vector properties won't be added to the property graph by default, but may be included as properties by including them in the property_columns argument. Use ``MGPropertyGraph.edge_vector_property_to_array`` to convert an edge vector property to an array. vector_property : string, optional If provided, all columns not included in other arguments will be used to create a vector property with the given name. This is often used for convenience instead of ``vector_properties`` when all input properties should be converted to a vector property. Returns ------- None Examples -------- >>> """ if type(dataframe) is not dask_cudf.DataFrame: raise TypeError("dataframe must be a Dask dataframe.") if type(vertex_col_names) not in [list, tuple]: raise TypeError( "vertex_col_names must be a list or tuple, got: " f"{type(vertex_col_names)}" ) if edge_id_col_name is not None: if not isinstance(edge_id_col_name, str): raise TypeError( "edge_id_col_name must be a string, got: " f"{type(edge_id_col_name)}" ) if edge_id_col_name not in dataframe.columns: if edge_id_col_name != dataframe.index.name: raise ValueError( "edge_id_col_name argument not in columns, " f"got {edge_id_col_name!r}" ) index_is_set = True else: index_is_set = False invalid_columns = set(vertex_col_names).difference(dataframe.columns) if invalid_columns: raise ValueError( "vertex_col_names contains column(s) not found " f"in dataframe: {list(invalid_columns)}" ) if type_name is not None and not isinstance(type_name, str): raise TypeError(f"type_name must be a string, got: {type(type_name)}") if type_name is None: type_name = self._default_type_name if property_columns: if type(property_columns) is not list: raise TypeError( f"property_columns must be a list, got: {type(property_columns)}" ) invalid_columns = set(property_columns).difference(dataframe.columns) if invalid_columns: raise ValueError( "property_columns contains column(s) not found in dataframe: " f"{list(invalid_columns)}" ) existing_vectors = ( set(property_columns) & self.__vertex_vector_property_lengths.keys() ) if existing_vectors: raise ValueError( "Non-vector property columns cannot be added to existing " f"vector properties: {', '.join(sorted(existing_vectors))}" ) if self.__is_edge_id_autogenerated is False and edge_id_col_name is None: raise NotImplementedError( "Unable to automatically generate edge IDs. " "`edge_id_col_name` must be specified if edge data has been " "previously added with edge_id_col_name." ) if self.__is_edge_id_autogenerated is True and edge_id_col_name is not None: raise NotImplementedError( "Invalid use of `edge_id_col_name`. Edge data has already " "been added with automatically generated IDs, so now all " "edge data must be added using automatically generated IDs." ) TCN = self.type_col_name if vector_properties is not None: invalid_keys = {self.src_col_name, self.dst_col_name, TCN} if property_columns: invalid_keys.update(property_columns) self._check_vector_properties( dataframe, vector_properties, self.__edge_vector_property_lengths, invalid_keys, ) if vector_property is not None: invalid_keys = { self.src_col_name, self.dst_col_name, TCN, vertex_col_names[0], vertex_col_names[1], } if property_columns: invalid_keys.update(property_columns) if vector_properties: invalid_keys.update(*vector_properties.values()) d = { vector_property: [ col for col in dataframe.columns if col not in invalid_keys ] } invalid_keys.difference_update(vertex_col_names) self._check_vector_properties( dataframe, d, self.__edge_vector_property_lengths, invalid_keys, ) # Update vector_properties, but don't mutate the original if vector_properties is not None: d.update(vector_properties) vector_properties = d # Clear the cached value for num_vertices since more could be added in # this method. This method cannot affect __node_type_value_counts self.__num_vertices = None self.__edge_type_value_counts = None # Could update instead # Add `type_name` to the categorical dtype if necessary is_first_data = self.__edge_prop_dataframe is None if is_first_data: temp_dataframe = cudf.DataFrame( columns=[self.src_col_name, self.dst_col_name, TCN] ) self.__update_dataframe_dtypes( temp_dataframe, { self.src_col_name: dataframe[vertex_col_names[0]].dtype, self.dst_col_name: dataframe[vertex_col_names[1]].dtype, }, ) temp_dataframe.index = temp_dataframe.index.rename(self.edge_id_col_name) if edge_id_col_name is not None and not index_is_set: temp_dataframe.index = temp_dataframe.index.astype( dataframe[edge_id_col_name].dtype ) # Use categorical dtype for the type column if self.__series_type is dask_cudf.Series: cat_class = cudf.CategoricalDtype else: cat_class = pd.CategoricalDtype cat_dtype = cat_class([type_name], ordered=False) self.__is_edge_id_autogenerated = edge_id_col_name is None self.__edge_prop_dataframe = temp_dataframe else: cat_dtype = self.__update_categorical_dtype( self.__edge_prop_dataframe, TCN, type_name ) # NOTE: This copies the incoming DataFrame in order to add the new # columns. The copied DataFrame is then merged (another copy) and then # deleted when out-of-scope. tmp_df = dataframe.copy() tmp_df[self.src_col_name] = tmp_df[vertex_col_names[0]] tmp_df[self.dst_col_name] = tmp_df[vertex_col_names[1]] # FIXME: We should do categorization first # Related issue: https://github.com/rapidsai/cugraph/issues/2903 tmp_df[TCN] = type_name tmp_df[TCN] = tmp_df[TCN].astype(cat_dtype) # Add unique edge IDs to the new rows. This is just a count for each # row starting from the last edge ID value, with initial edge ID 0. if edge_id_col_name is None: # FIXME: can we assign index instead of column? starting_eid = -1 if self.__last_edge_id is None else self.__last_edge_id tmp_df[self.edge_id_col_name] = 1 tmp_df[self.edge_id_col_name] = ( tmp_df[self.edge_id_col_name].cumsum() + starting_eid ) tmp_df = tmp_df.persist().set_index(self.edge_id_col_name).persist() self.__last_edge_id = starting_eid + len(tmp_df) else: if not index_is_set: tmp_df = tmp_df.rename( columns={edge_id_col_name: self.edge_id_col_name} ).persist() tmp_df = tmp_df.set_index(self.edge_id_col_name) tmp_df.index = tmp_df.index.astype(dataframe[edge_id_col_name].dtype) elif tmp_df.index.name != self.edge_id_col_name: tmp_df.index = tmp_df.index.rename(self.edge_id_col_name) tmp_df = tmp_df.persist() if property_columns: # all columns column_names_to_drop = set(tmp_df.columns) # remove the ones to keep column_names_to_drop.difference_update( property_columns + [self.src_col_name, self.dst_col_name, TCN] ) else: column_names_to_drop = {vertex_col_names[0], vertex_col_names[1]} if vector_properties: # Drop vector property source columns by default more_to_drop = set().union(*vector_properties.values()) if property_columns is not None: more_to_drop.difference_update(property_columns) column_names_to_drop |= more_to_drop column_names_to_drop -= vector_properties.keys() tmp_df = self._create_vector_properties(tmp_df, vector_properties) tmp_df = tmp_df.drop(labels=column_names_to_drop, axis=1) # Save the original dtypes for each new column so they can be restored # prior to constructing subgraphs (since column dtypes may get altered # during merge to accommodate NaN values). if is_first_data: new_col_info = tmp_df.dtypes.items() else: new_col_info = self.__get_new_column_dtypes( tmp_df, self.__edge_prop_dataframe ) self.__edge_prop_dtypes.update(new_col_info) self.__update_dataframe_dtypes(tmp_df, self.__edge_prop_dtypes) if is_first_data: self.__edge_prop_dataframe = tmp_df else: # Join on edge ids (the index) # TODO: can we automagically determine when we to use concat? df = self.__edge_prop_dataframe.join( tmp_df, how="outer", rsuffix="_NEW_", # npartitions=self.__num_workers # TODO: see how this behaves ).persist() cols = self.__edge_prop_dataframe.columns.intersection( tmp_df.columns ).to_list() rename_cols = {f"{col}_NEW_": col for col in cols} new_cols = list(rename_cols) sub_df = df[new_cols].rename(columns=rename_cols) # This only adds data--it doesn't replace existing data df = df.drop(columns=new_cols).fillna(sub_df).persist() if df.npartitions > 4 * self.__num_workers: # TODO: better understand behavior of npartitions argument in join df = df.repartition(npartitions=2 * self.__num_workers).persist() self.__edge_prop_dataframe = df # Update the edge eval dict with the latest column instances self._update_eval_dict( self.__edge_prop_eval_dict, self.__edge_prop_dataframe, self.edge_id_col_name, ) def get_edge_data(self, edge_ids=None, types=None, columns=None): """ Return a dataframe containing edge properties for only the specified edge_ids, columns, and/or edge type, or all edge IDs if not specified. """ if self.__edge_prop_dataframe is not None: df = self.__edge_prop_dataframe if edge_ids is not None: if isinstance(edge_ids, int): edge_ids = [edge_ids] try: df = df.loc[edge_ids] except TypeError: raise TypeError( "edge_ids needs to be a list-like type " f"compatible with DataFrame.loc[], got {type(edge_ids)}" ) if types is not None: if isinstance(types, str): df_mask = df[self.type_col_name] == types else: df_mask = df[self.type_col_name].isin(types) df = df.loc[df_mask] # The "internal" src, dst, edge_id, and type columns are also # included/added since they are assumed to be needed by the caller. if columns is None: # remove the "internal" weight column if one was added all_columns = list(self.__edge_prop_dataframe.columns) if self.weight_col_name in all_columns: all_columns.remove(self.weight_col_name) df = df[all_columns] else: # FIXME: invalid columns will result in a KeyError, should a # check be done here and a more PG-specific error raised? df = df[ [self.src_col_name, self.dst_col_name, self.type_col_name] + columns ] df_out = df.reset_index() # Preserve the dtype (edge id type) to avoid cugraph algorithms # throwing errors due to a dtype mismatch index_dtype = self.__edge_prop_dataframe.index.dtype df_out.index = df_out.index.astype(index_dtype) return df_out return None def fillna_vertices(self, val=0): """ Fills empty vertex property values with the given value, zero by default. Fills in-place. Parameters ---------- val : object, Series, or dict The object that will replace "na". Default = 0. If a dict or Series is passed, the index or keys are the columns to fill and the values are the fill value for the corresponding column. """ self.__vertex_prop_dataframe = self.__vertex_prop_dataframe.fillna( val ).persist() def fillna_edges(self, val=0): """ Fills empty edge property values with the given value, zero by default. Fills in-place. Parameters ---------- val : object, Series, or dict The object that will replace "na". Default = 0. If a dict or Series is passed, the index or keys are the columns to fill and the values are the fill value for the corresponding column. """ self.__edge_prop_dataframe = self.__edge_prop_dataframe.fillna(val).persist() def select_vertices(self, expr, from_previous_selection=None): raise NotImplementedError def select_edges(self, expr): """ Evaluate expr and return a PropertySelection object representing the edges that match the expression. Parameters ---------- expr : string A python expression using property names and operators to select specific edges. Returns ------- PropertySelection instance to be used for calls to extract_subgraph() in order to construct a Graph containing only specific edges. Examples -------- >>> """ # FIXME: check types globals = {} locals = self.__edge_prop_eval_dict selected_col = eval(expr, globals, locals) return EXPERIMENTAL__MGPropertySelection(edge_selection_series=selected_col) def extract_subgraph( self, create_using=None, selection=None, edge_weight_property=None, default_edge_weight=None, check_multi_edges=True, renumber_graph=True, add_edge_data=True, ): """ Return a subgraph of the overall PropertyGraph containing vertices and edges that match a selection. Parameters ---------- create_using : type or instance of cugraph.Graph or PropertyGraph, optional Creates a Graph to return using the type specified. If an instance is specified, the type of the instance is used to construct the return Graph, and all relevant attributes set on the instance are copied to the return Graph (eg. directed). If not specified the returned Graph will be a directed cugraph.MultiGraph instance. selection : PropertySelection, optional A PropertySelection returned from one or more calls to select_vertices() and/or select_edges(), used for creating a Graph with only the selected properties. If not specified the returned Graph will have all properties. Note, this could result in a Graph with multiple edges, which may not be supported based on the value of create_using. edge_weight_property : string, optional The name of the property whose values will be used as weights on the returned Graph. If not specified, the returned Graph will be unweighted. Ignored for PropertyGraph return type. default_edge_weight : float64, optional Value that replaces empty weight property fields. Ignored for PropertyGraph return type. check_multi_edges : bool (default True) When True and create_using argument is given and not a MultiGraph, this will perform a check to verify that the edges in the edge dataframe do not form a multigraph with duplicate edges. Ignored for PropertyGraph return type. renumber_graph : bool (default True) If True, return a Graph that has been renumbered for use by graph algorithms. If False, the returned graph will need to be manually renumbered prior to calling graph algos. Ignored for PropertyGraph return type. add_edge_data : bool (default True) If True, add meta data about the edges contained in the extracted graph which are required for future calls to annotate_dataframe(). Ignored for PropertyGraph return type. Returns ------- A Graph instance of the same type as create_using containing only the vertices and edges resulting from applying the selection to the set of vertex and edge property data. Examples -------- >>> """ if selection is not None and not isinstance( selection, EXPERIMENTAL__MGPropertySelection ): raise TypeError( "selection must be an instance of " f"PropertySelection, got {type(selection)}" ) # NOTE: the expressions passed in to extract specific edges and # vertices assume the original dtypes in the user input have been # preserved. However, merge operations on the DataFrames can change # dtypes (eg. int64 to float64 in order to add NaN entries). This # should not be a problem since the conversions do not change the # values. if selection is not None and selection.vertex_selections is not None: selected_vertex_dataframe = self.__vertex_prop_dataframe[ selection.vertex_selections ] else: selected_vertex_dataframe = None if selection is not None and selection.edge_selections is not None: selected_edge_dataframe = self.__edge_prop_dataframe[ selection.edge_selections ] else: selected_edge_dataframe = self.__edge_prop_dataframe # FIXME: check that self.__edge_prop_dataframe is set! # If vertices were specified, select only the edges that contain the # selected verts in both src and dst if ( selected_vertex_dataframe is not None and not selected_vertex_dataframe.empty ): has_srcs = selected_edge_dataframe[self.src_col_name].isin( selected_vertex_dataframe.index ) has_dsts = selected_edge_dataframe[self.dst_col_name].isin( selected_vertex_dataframe.index ) edges = selected_edge_dataframe[has_srcs & has_dsts] # Alternative to benchmark # edges = selected_edge_dataframe.merge( # selected_vertex_dataframe[[]], # left_on=self.src_col_name, # right_index=True, # ).merge( # selected_vertex_dataframe[[]], # left_on=self.dst_col_name, # right_index=True, # ) else: edges = selected_edge_dataframe # Default create_using set here instead of function signature to # prevent cugraph from running on import. This may help diagnose errors create_kind = "cugraph" if create_using is None: create_using = cugraph.MultiGraph(directed=True) elif isinstance(create_using, type(self)): rv = type(create_using)() create_kind = "propertygraph" elif type(create_using) is type and issubclass(create_using, type(self)): rv = create_using() create_kind = "propertygraph" if create_kind == "cugraph": # The __*_prop_dataframes have likely been merged several times and # possibly had their dtypes converted in order to accommodate NaN # values. Restore the original dtypes in the resulting edges df prior # to creating a Graph. self.__update_dataframe_dtypes(edges, self.__edge_prop_dtypes) return self.edge_props_to_graph( edges, create_using=create_using, edge_weight_property=edge_weight_property, default_edge_weight=default_edge_weight, check_multi_edges=check_multi_edges, renumber_graph=renumber_graph, add_edge_data=add_edge_data, ) # Return a subgraph as PropertyGraph if ( selected_vertex_dataframe is None and self.__vertex_prop_dataframe is not None ): selected_vertex_dataframe = self.__vertex_prop_dataframe.copy() num_vertices = self.__num_vertices vertex_type_value_counts = self.__vertex_type_value_counts else: num_vertices = None vertex_type_value_counts = None if edges is not None and edges is self.__edge_prop_dataframe: edges = edges.copy() edge_type_value_counts = self.__edge_type_value_counts else: edge_type_value_counts = None rv._build_from_components( vertex_prop_dataframe=selected_vertex_dataframe, edge_prop_dataframe=edges, dataframe_type=self.__dataframe_type, series_type=self.__series_type, vertex_prop_dtypes=dict(self.__vertex_prop_dtypes), edge_prop_dtypes=dict(self.__edge_prop_dtypes), vertex_vector_property_lengths=dict(self.__vertex_vector_property_lengths), edge_vector_property_lengths=dict(self.__edge_vector_property_lengths), last_edge_id=self.__last_edge_id, is_edge_id_autogenerated=self.__is_edge_id_autogenerated, # Cached properties num_vertices=num_vertices, vertex_type_value_counts=vertex_type_value_counts, edge_type_value_counts=edge_type_value_counts, ) return rv def annotate_dataframe(self, df, G, edge_vertex_col_names): raise NotImplementedError() def edge_props_to_graph( self, edge_prop_df, create_using, edge_weight_property=None, default_edge_weight=None, check_multi_edges=True, renumber_graph=True, add_edge_data=True, ): """ Create and return a Graph from the edges in edge_prop_df. """ # Don't mutate input data edge_prop_df = edge_prop_df.copy() # FIXME: check default_edge_weight is valid if edge_weight_property: if ( edge_weight_property not in edge_prop_df.columns and edge_prop_df.index.name != edge_weight_property ): raise ValueError( "edge_weight_property " f'"{edge_weight_property}" was not found in ' "edge_prop_df" ) # Ensure a valid edge_weight_property can be used for applying # weights to the subgraph, and if a default_edge_weight was # specified, apply it to all NAs in the weight column. # Also allow the type column to be specified as the edge weight # property so that uniform_neighbor_sample can be called with # the weights interpreted as types. if edge_weight_property == self.type_col_name: prop_col = edge_prop_df[self.type_col_name].cat.codes.astype("float32") edge_prop_df["_temp_type_col"] = prop_col edge_weight_property = "_temp_type_col" elif edge_weight_property in edge_prop_df.columns: prop_col = edge_prop_df[edge_weight_property] else: prop_col = edge_prop_df.index.to_series() edge_prop_df[edge_weight_property] = prop_col if prop_col.count().compute() != prop_col.size: if default_edge_weight is None: raise ValueError( f'edge_weight_property "{edge_weight_property}" ' "contains NA values in the subgraph and " "default_edge_weight is not set" ) else: prop_col.fillna(default_edge_weight, inplace=True) edge_attr = edge_weight_property # If a default_edge_weight was specified but an edge_weight_property # was not, a new edge weight column must be added. elif default_edge_weight: edge_attr = self.weight_col_name edge_prop_df[edge_attr] = default_edge_weight else: edge_attr = None # Set up the new Graph to return if isinstance(create_using, cugraph.Graph): # FIXME: extract more attrs from the create_using instance attrs = {"directed": create_using.is_directed()} G = type(create_using)(**attrs) elif type(create_using) is type and issubclass(create_using, cugraph.Graph): G = create_using() else: raise TypeError( "create_using must be a cugraph.Graph " "(or subclass) type or instance, got: " f"{type(create_using)}" ) # Prevent duplicate edges (if not allowed) since applying them to # non-MultiGraphs would result in ambiguous edge properties. if ( check_multi_edges and not G.is_multigraph() and self.is_multigraph(edge_prop_df).compute() ): if create_using: if type(create_using) is type: t = create_using.__name__ else: t = type(create_using).__name__ msg = f"'{t}' graph type specified by create_using" else: msg = "default Graph graph type" raise RuntimeError( "query resulted in duplicate edges which " f"cannot be represented with the {msg}" ) col_names = [self.src_col_name, self.dst_col_name] if edge_attr is not None: col_names.append(edge_attr) edge_prop_df = edge_prop_df.reset_index().drop( [col for col in edge_prop_df if col not in col_names], axis=1 ) edge_prop_df = edge_prop_df.repartition( npartitions=self.__num_workers * 4 ).persist() G.from_dask_cudf_edgelist( edge_prop_df, source=self.src_col_name, destination=self.dst_col_name, edge_attr=edge_attr, renumber=renumber_graph, ) if add_edge_data: # Set the edge_data on the resulting Graph to a DataFrame # containing the edges and the edge ID for each. Edge IDs are # needed for future calls to annotate_dataframe() in order to # associate edges with their properties, since the PG can contain # multiple edges between vertrices with different properties. # FIXME: also add vertex_data G.edge_data = self.__create_property_lookup_table(edge_prop_df) del edge_prop_df return G def renumber_vertices_by_type(self, prev_id_column=None): """Renumber vertex IDs to be contiguous by type. Parameters ---------- prev_id_column : str, optional Column name to save the vertex ID before renumbering. Returns a DataFrame with the start and stop IDs for each vertex type. Stop is *inclusive*. """ # Check if some vertex IDs exist only in edge data TCN = self.type_col_name default = self._default_type_name if self.__edge_prop_dataframe is not None and self.get_num_vertices( default, include_edge_data=True ) != self.get_num_vertices(default, include_edge_data=False): raise NotImplementedError( "Currently unable to renumber vertices when some vertex " "IDs only exist in edge data" ) if self.__vertex_prop_dataframe is None: return None if ( prev_id_column is not None and prev_id_column in self.__vertex_prop_dataframe ): raise ValueError( f"Can't save previous IDs to existing column {prev_id_column!r}" ) # Use categorical dtype for the type column if self.__series_type is dask_cudf.Series: cat_class = cudf.CategoricalDtype else: cat_class = pd.CategoricalDtype is_cat = isinstance(self.__vertex_prop_dataframe.dtypes[TCN], cat_class) if not is_cat: cat_dtype = cat_class([TCN], ordered=False) self.__vertex_prop_dataframe[TCN] = self.__vertex_prop_dataframe[ TCN ].astype(cat_dtype) df = self.__vertex_prop_dataframe index_dtype = df.index.dtype # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 cat_dtype = df.dtypes[TCN] df[TCN] = df[TCN].astype(str) # Include self.vertex_col_name when sorting by values to ensure we can # evenly distribute the data across workers. df = df.reset_index().persist() if len(cat_dtype.categories) > 1 and len(self.vertex_types) > 1: # `self.vertex_types` is currently not cheap, b/c it looks at edge df df = df.sort_values( by=[TCN, self.vertex_col_name], ignore_index=True ).persist() if self.__edge_prop_dataframe is not None: new_name = f"new_{self.vertex_col_name}" df[new_name] = 1 df[new_name] = df[new_name].cumsum() - 1 mapper = df[[self.vertex_col_name, new_name]] edge_index_dtype = self.__edge_prop_dataframe.index.dtype self.__edge_prop_dataframe = ( self.__edge_prop_dataframe # map src_col_name IDs .merge(mapper, left_on=self.src_col_name, right_on=self.vertex_col_name) .drop(columns=[self.src_col_name, self.vertex_col_name]) .rename(columns={new_name: self.src_col_name}) # map dst_col_name IDs .merge(mapper, left_on=self.dst_col_name, right_on=self.vertex_col_name) .drop(columns=[self.dst_col_name, self.vertex_col_name]) .rename(columns={new_name: self.dst_col_name}) ) self.__edge_prop_dataframe.index = self.__edge_prop_dataframe.index.astype( edge_index_dtype ) self.__edge_prop_dataframe.index = self.__edge_prop_dataframe.index.rename( self.edge_id_col_name ) if prev_id_column is None: df[self.vertex_col_name] = df[new_name] del df[new_name] else: df = df.rename( columns={ new_name: self.vertex_col_name, self.vertex_col_name: prev_id_column, } ) else: if prev_id_column is not None: df[prev_id_column] = df[self.vertex_col_name] df[self.vertex_col_name] = 1 df[self.vertex_col_name] = df[self.vertex_col_name].cumsum() - 1 # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 df[TCN] = df[TCN].astype(cat_dtype) df[self.vertex_col_name] = df[self.vertex_col_name].astype(index_dtype) self.__vertex_prop_dataframe = ( df.persist().set_index(self.vertex_col_name, sorted=True).persist() ) # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 df = self._vertex_type_value_counts cat_dtype = df.index.dtype df.index = df.index.astype(str) # self._vertex_type_value_counts rv = df.sort_index().cumsum().to_frame("stop") # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 df.index = df.index.astype(cat_dtype) rv["start"] = rv["stop"].shift(1, fill_value=0) rv["stop"] -= 1 # Make inclusive return rv[["start", "stop"]] def renumber_edges_by_type(self, prev_id_column=None): """Renumber edge IDs to be contiguous by type. Parameters ---------- prev_id_column : str, optional Column name to save the edge ID before renumbering. Returns a DataFrame with the start and stop IDs for each edge type. Stop is *inclusive*. """ # TODO: keep track if edges are already numbered correctly. if self.__edge_prop_dataframe is None: return None if prev_id_column is not None and prev_id_column in self.__edge_prop_dataframe: raise ValueError( f"Can't save previous IDs to existing column {prev_id_column!r}" ) df = self.__edge_prop_dataframe index_dtype = df.index.dtype # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 cat_dtype = df.dtypes[self.type_col_name] df[self.type_col_name] = df[self.type_col_name].astype(str) # Include self.edge_id_col_name when sorting by values to ensure we can # evenly distribute the data across workers. df = df.reset_index().persist() if len(cat_dtype.categories) > 1 and len(self.edge_types) > 1: df = df.sort_values( by=[self.type_col_name, self.edge_id_col_name], ignore_index=True ).persist() if prev_id_column is not None: df[prev_id_column] = df[self.edge_id_col_name] # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 df[self.type_col_name] = df[self.type_col_name].astype(cat_dtype) df[self.edge_id_col_name] = 1 df[self.edge_id_col_name] = df[self.edge_id_col_name].cumsum() - 1 df[self.edge_id_col_name] = df[self.edge_id_col_name].astype(index_dtype) self.__edge_prop_dataframe = ( df.persist().set_index(self.edge_id_col_name, sorted=True).persist() ) # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 df = self._edge_type_value_counts if df.index.dtype == cat_dtype: df.index = df.index.astype(str) # self._edge_type_value_counts rv = df.sort_index().cumsum().to_frame("stop") # FIXME DASK_CUDF: https://github.com/rapidsai/cudf/issues/11795 df.index = df.index.astype(cat_dtype) rv["start"] = rv["stop"].shift(1, fill_value=0) rv["stop"] -= 1 # Make inclusive return rv[["start", "stop"]] def vertex_vector_property_to_array( self, df, col_name, fillvalue=None, *, missing="ignore" ): """Convert a known vertex vector property in a DataFrame to an array. Parameters ---------- df : dask_cudf.DataFrame col_name : str The column name in the DataFrame to convert to an array. This vector property should have been created by MGPropertyGraph. fillvalue : scalar or list, optional (default None) Fill value for rows with missing vector data. If it is a list, it must be the correct size of the vector property. If fillvalue is None, then behavior if missing data is controlled by ``missing`` keyword. Leave this as None for better performance if all rows should have data. missing : {"ignore", "error"} If "ignore", empty or null rows without vector data will be skipped when creating the array, so output array shape will be [# of non-empty rows] by [size of vector property]. When "error", RuntimeError will be raised if there are any empty rows. Ignored if fillvalue is given. Returns ------- dask.array (of cupy.ndarray) """ if col_name not in self.__vertex_vector_property_lengths: raise ValueError(f"{col_name!r} is not a known vertex vector property") length = self.__vertex_vector_property_lengths[col_name] return self._get_vector_property(df, col_name, length, fillvalue, missing) def edge_vector_property_to_array( self, df, col_name, fillvalue=None, *, missing="ignore" ): """Convert a known edge vector property in a DataFrame to an array. Parameters ---------- df : dask_cudf.DataFrame col_name : str The column name in the DataFrame to convert to an array. This vector property should have been created by MGPropertyGraph. fillvalue : scalar or list, optional (default None) Fill value for rows with missing vector data. If it is a list, it must be the correct size of the vector property. If fillvalue is None, then behavior if missing data is controlled by ``missing`` keyword. Leave this as None for better performance if all rows should have data. missing : {"ignore", "error"} If "ignore", empty or null rows without vector data will be skipped when creating the array, so output array shape will be [# of non-empty rows] by [size of vector property]. When "error", RuntimeError will be raised if there are any empty rows. Ignored if fillvalue is given. Returns ------- dask.array (of cupy.ndarray) """ if col_name not in self.__edge_vector_property_lengths: raise ValueError(f"{col_name!r} is not a known edge vector property") length = self.__edge_vector_property_lengths[col_name] return self._get_vector_property(df, col_name, length, fillvalue, missing) def _check_vector_properties( self, df, vector_properties, vector_property_lengths, invalid_keys ): """Check if vector_properties is valid and update vector_property_lengths""" df_cols = set(df.columns) for key, columns in vector_properties.items(): if key in invalid_keys: raise ValueError( "Cannot assign new vector property to existing " f"non-vector property: {key}" ) if isinstance(columns, str): # If df[columns] is a ListDtype column, should we allow it? raise TypeError( f"vector property columns for {key!r} should be a list; " f"got a str ({columns!r})" ) if not df_cols.issuperset(columns): missing = ", ".join(set(columns) - df_cols) raise ValueError( f"Dataframe does not have columns for vector property {key!r}:" f"{missing}" ) if not columns: raise ValueError("Empty vector property columns for {key!r}!") if vector_property_lengths.get(key, len(columns)) != len(columns): prev_length = vector_property_lengths[key] new_length = len(columns) raise ValueError( f"Wrong size for vector property {key}; got {new_length}, but " f"this vector property already exists with size {prev_length}" ) for key, columns in vector_properties.items(): vector_property_lengths[key] = len(columns) def _create_vector_properties(self, df, vector_properties): return df.map_partitions( self._create_vector_properties_partition, vector_properties ) def _get_vector_property(self, df, col_name, length, fillvalue, missing): if type(df) is not self.__dataframe_type: raise TypeError( f"Expected type {self.__dataframe_type}; got type {type(df)}" ) if col_name not in df.columns: raise ValueError(f"Column name {col_name} is not in the columns of df") if missing not in {"error", "ignore"}: raise ValueError( f'missing keyword must be one of "error" or "ignore"; got {missing!r}' ) if fillvalue is not None: try: fillvalue = list(fillvalue) except Exception: fillvalue = [fillvalue] * length else: if len(fillvalue) != length: raise ValueError( f"Wrong size of list as fill value; got {len(fillvalue)}, " f"expected {length}" ) if df.dtypes[col_name] != "list": raise TypeError( "Wrong dtype for vector property; expected 'list', " f"got {df.dtypes[col_name]}" ) s = df[col_name] meta = self._vector_series_to_array_partition( s._meta, length, fillvalue, "ignore" ) return s.map_partitions( self._vector_series_to_array_partition, length, fillvalue, missing, meta=meta, ) def is_multi_gpu(self): """ Return True if this is a multi-gpu graph. Always returns True for MGPropertyGraph. """ return True @classmethod def is_multigraph(cls, df): """ Return True if df has >1 of the same src, dst pair """ return cls._has_duplicates(df, [cls.src_col_name, cls.dst_col_name]) @classmethod def has_duplicate_edges(cls, df, columns=None): """ Return True if df has rows with the same src, dst, type, and columns """ cols = [cls.src_col_name, cls.dst_col_name, cls.type_col_name] if columns: cols.extend(columns) return cls._has_duplicates(df, cols) @classmethod def _has_duplicates(cls, df, cols): # empty not supported by dask if len(df.columns) == 0: return False unique_pair_len = df.drop_duplicates( split_out=df.npartitions, ignore_index=True ).shape[0] # if unique_pairs == len(df) # then no duplicate edges return unique_pair_len != df.shape[0] def __create_property_lookup_table(self, edge_prop_df): """ Returns a DataFrame containing the src vertex, dst vertex, and edge_id values from edge_prop_df. """ return edge_prop_df[[self.src_col_name, self.dst_col_name]].reset_index() def __get_all_vertices_series(self): """ Return a list of all Series objects that contain vertices from all tables. """ vpd = self.__vertex_prop_dataframe epd = self.__edge_prop_dataframe vert_sers = [] if vpd is not None: vert_sers.append(vpd.index.to_series()) if epd is not None: vert_sers.append(epd[self.src_col_name]) vert_sers.append(epd[self.dst_col_name]) # `dask_cudf.concat` doesn't work when the index dtypes are different # See: https://github.com/rapidsai/cudf/issues/11741 if len(vert_sers) > 1 and not all( cudf.api.types.is_dtype_equal(vert_sers[0].index.dtype, s.index.dtype) for s in vert_sers ): vert_sers = [s.reset_index(drop=True) for s in vert_sers] return vert_sers @staticmethod def __get_new_column_dtypes(from_df, to_df): """ Returns a list containing tuples of (column name, dtype) for each column in from_df that is not present in to_df. """ new_cols = set(from_df.columns) - set(to_df.columns) return [(col, from_df.dtypes[col]) for col in new_cols] @staticmethod def __update_dataframe_dtypes(df, column_dtype_dict): """ Set the dtype for columns in df using the dtypes in column_dtype_dict. This also handles converting standard integer dtypes to nullable integer dtypes, needed to accommodate NA values in columns. """ for (col, dtype) in column_dtype_dict.items(): if col not in df.columns: continue # If the DataFrame is Pandas and the dtype is an integer type, # ensure a nullable integer array is used by specifying the correct # dtype. The alias for these dtypes is simply a capitalized string # (eg. "Int64") # https://pandas.pydata.org/pandas-docs/stable/user_guide/missing_data.html#integer-dtypes-and-missing-data dtype_str = str(dtype) if dtype_str in ["int32", "int64"]: dtype_str = dtype_str.title() if str(df.dtypes[col]) != dtype_str: df[col] = df[col].astype(dtype_str) def __update_categorical_dtype(self, df, column, val): """Add a new category to a categorical dtype column of a dataframe. Returns the new categorical dtype. """ # Add `val` to the categorical dtype if necessary if val not in df.dtypes[column].categories: df[column] = df[column].cat.add_categories([val]) return df.dtypes[column] @staticmethod def _create_vector_properties_partition(df, vector_properties): # Make each vector contigous and 1-d new_cols = {} for key, columns in vector_properties.items(): values = df[columns].values new_cols[key] = create_list_series_from_2d_ar(values, index=df.index) return df.assign(**new_cols) @staticmethod def _vector_series_to_array_partition(s, length, fillvalue, missing): # This returns a writable view (i.e., no copies!) if len(s) == 0: # TODO: fix bug in cudf; operating on dask_cudf dataframes nests list dtype dtype = s.dtype while dtype == "list": dtype = dtype.element_type return cupy.empty((0, length), dtype=dtype) if fillvalue is not None: s = s.copy() # copy b/c we mutate below s[s.isnull()] = fillvalue rv = s._data.columns[0].children[-1].values.reshape(-1, length) if fillvalue is None and missing == "error" and rv.shape[0] != len(s): raise RuntimeError( f"Vector property {s.name!r} has empty rows! " 'Provide a fill value or use `missing="ignore"` to ignore empty rows.' ) return rv
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask
rapidsai_public_repos/cugraph/python/cugraph/cugraph/dask/structure/__init__.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/triangle_count.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.utilities import ensure_cugraph_obj_for_nx import cudf from pylibcugraph import triangle_count as pylibcugraph_triangle_count from pylibcugraph import ResourceHandle import warnings # FIXME: Move this function to the utility module so that it can be # shared by other algos def ensure_valid_dtype(input_graph, start_list): vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] if isinstance(start_list, cudf.Series): start_list_dtypes = start_list.dtype else: start_list_dtypes = start_list.dtypes[0] if start_list_dtypes != vertex_dtype: warning_msg = ( "Triangle_count requires 'start_list' to match the graph's 'vertex' type. " f"input graph's vertex type is: {vertex_dtype} and got " f"'start_list' of type: {start_list_dtypes}." ) warnings.warn(warning_msg, UserWarning) start_list = start_list.astype(vertex_dtype) return start_list def triangle_count(G, start_list=None): """ Compute the number of triangles (cycles of length three) in the input graph. Parameters ---------- G : cugraph.graph or networkx.Graph cuGraph graph descriptor, should contain the connectivity information, (edge weights are not used in this algorithm). The current implementation only supports undirected graphs. start_list : list or cudf.Series list of vertices for triangle count. if None the entire set of vertices in the graph is processed Returns ------- result : cudf.DataFrame GPU data frame containing 2 cudf.Series ddf['vertex']: cudf.Series Contains the triangle counting vertices ddf['counts']: cudf.Series Contains the triangle counting counts Examples -------- >>> gdf = cudf.read_csv(datasets_path / 'karate.csv', ... delimiter = ' ', ... dtype=['int32', 'int32', 'float32'], ... header=None) >>> G = cugraph.Graph() >>> G.from_cudf_edgelist(gdf, source='0', destination='1', edge_attr='2') >>> count = cugraph.triangle_count(G) """ G, _ = ensure_cugraph_obj_for_nx(G) if G.is_directed(): raise ValueError("input graph must be undirected") if start_list is not None: if isinstance(start_list, int): start_list = [start_list] if isinstance(start_list, list): start_list = cudf.Series(start_list) if not isinstance(start_list, cudf.Series): raise TypeError( f"'start_list' must be either a list or a cudf.Series," f"got: {start_list.dtype}" ) start_list = ensure_valid_dtype(G, start_list) if G.renumbered is True: if isinstance(start_list, cudf.DataFrame): start_list = G.lookup_internal_vertex_id(start_list, start_list.columns) else: start_list = G.lookup_internal_vertex_id(start_list) do_expensive_check = False vertex, counts = pylibcugraph_triangle_count( resource_handle=ResourceHandle(), graph=G._plc_graph, start_list=start_list, do_expensive_check=do_expensive_check, ) df = cudf.DataFrame() df["vertex"] = vertex df["counts"] = counts if G.renumbered: df = G.unrenumber(df, "vertex") return df
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/induced_subgraph.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import warnings from typing import Union, Tuple import cudf from pylibcugraph import ResourceHandle from pylibcugraph import induced_subgraph as pylibcugraph_induced_subgraph from cugraph.structure import Graph from cugraph.utilities import ( ensure_cugraph_obj_for_nx, cugraph_to_nx, ) from cugraph.utilities.utils import import_optional # FIXME: the networkx.Graph type used in type annotations is specified # using a string literal to avoid depending on and importing networkx. # Instead, networkx is imported optionally, which may cause a problem # for a type checker if run in an environment where networkx is not installed. networkx = import_optional("networkx") # FIXME: Move this function to the utility module so that it can be # shared by other algos def ensure_valid_dtype(input_graph: Graph, input: cudf.Series, input_name: str): vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] input_dtype = input.dtype if input_dtype != vertex_dtype: warning_msg = ( f"Subgraph requires '{input_name}' " "to match the graph's 'vertex' type. " f"input graph's vertex type is: {vertex_dtype} and got " f"'{input_name}' of type: " f"{input_dtype}." ) warnings.warn(warning_msg, UserWarning) input = input.astype(vertex_dtype) return input def induced_subgraph( G: Union[Graph, "networkx.Graph"], vertices: Union[cudf.Series, cudf.DataFrame], offsets: Union[list, cudf.Series] = None, ) -> Tuple[Union[Graph, "networkx.Graph"], cudf.Series]: """ Compute a subgraph of the existing graph including only the specified vertices. This algorithm works with both directed and undirected graphs and does not actually traverse the edges, but instead simply pulls out any edges that are incident on vertices that are both contained in the vertices list. If no subgraph can be extracted from the vertices provided, a 'None' value will be returned. Parameters ---------- G : cugraph.Graph or networkx.Graph The current implementation only supports weighted graphs. vertices : cudf.Series or cudf.DataFrame Specifies the vertices of the induced subgraph. For multi-column vertices, vertices should be provided as a cudf.DataFrame offsets : list or cudf.Series, optional Specifies the subgraph offsets into the subgraph vertices. If no offsets array is provided, a default array [0, len(vertices)] will be used. Returns ------- Sg : cugraph.Graph or networkx.Graph A graph object containing the subgraph induced by the given vertex set. seeds_offsets: cudf.Series A cudf Series containing the starting offset in the returned edge list for each seed. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> verts = np.zeros(3, dtype=np.int32) >>> verts[0] = 0 >>> verts[1] = 1 >>> verts[2] = 2 >>> sverts = cudf.Series(verts) >>> Sg, seeds_offsets = cugraph.induced_subgraph(G, sverts) """ G, isNx = ensure_cugraph_obj_for_nx(G) directed = G.is_directed() # FIXME: Hardcoded for now offsets = None if G.renumbered: if isinstance(vertices, cudf.DataFrame): vertices = G.lookup_internal_vertex_id(vertices, vertices.columns) else: vertices = G.lookup_internal_vertex_id(vertices) vertices = ensure_valid_dtype(G, vertices, "subgraph_vertices") if not isinstance(offsets, cudf.Series): if isinstance(offsets, list): offsets = cudf.Series(offsets) elif offsets is None: # FIXME: Does the offsets always start from zero? offsets = cudf.Series([0, len(vertices)]) result_graph = Graph(directed=directed) do_expensive_check = False source, destination, weight, offsets = pylibcugraph_induced_subgraph( resource_handle=ResourceHandle(), graph=G._plc_graph, subgraph_vertices=vertices, subgraph_offsets=offsets, do_expensive_check=do_expensive_check, ) df = cudf.DataFrame() df["src"] = source df["dst"] = destination df["weight"] = weight if len(df) == 0: return None, None seeds_offsets = cudf.Series(offsets) if G.renumbered: df, src_names = G.unrenumber(df, "src", get_column_names=True) df, dst_names = G.unrenumber(df, "dst", get_column_names=True) else: # FIXME: THe original 'src' and 'dst' are not stored in 'simpleGraph' src_names = "src" dst_names = "dst" if G.edgelist.weights: result_graph.from_cudf_edgelist( df, source=src_names, destination=dst_names, edge_attr="weight" ) else: result_graph.from_cudf_edgelist(df, source=src_names, destination=dst_names) if isNx is True: result_graph = cugraph_to_nx(result_graph) return result_graph, seeds_offsets
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/ecg.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.utilities import ( ensure_cugraph_obj_for_nx, df_score_to_dictionary, ) import cudf from pylibcugraph import ecg as pylibcugraph_ecg from pylibcugraph import ResourceHandle def ecg(input_graph, min_weight=0.05, ensemble_size=16, weight=None): """ Compute the Ensemble Clustering for Graphs (ECG) partition of the input graph. ECG runs truncated Louvain on an ensemble of permutations of the input graph, then uses the ensemble partitions to determine weights for the input graph. The final result is found by running full Louvain on the input graph using the determined weights. See https://arxiv.org/abs/1809.05578 for further information. Parameters ---------- input_graph : cugraph.Graph or NetworkX Graph The graph descriptor should contain the connectivity information and weights. The adjacency list will be computed if not already present. min_weight : float, optional (default=0.5) The minimum value to assign as an edgeweight in the ECG algorithm. It should be a value in the range [0,1] usually left as the default value of .05 ensemble_size : integer, optional (default=16) The number of graph permutations to use for the ensemble. The default value is 16, larger values may produce higher quality partitions for some graphs. weight : str, optional (default=None) This parameter is here for NetworkX compatibility and represents which NetworkX data column represents Edge weights. Returns ------- parts : cudf.DataFrame or python dictionary GPU data frame of size V containing two columns, the vertex id and the partition id it is assigned to. df[vertex] : cudf.Series Contains the vertex identifiers df[partition] : cudf.Series Contains the partition assigned to the vertices Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> parts = cugraph.ecg(G) """ input_graph, isNx = ensure_cugraph_obj_for_nx(input_graph) vertex, partition = pylibcugraph_ecg( resource_handle=ResourceHandle(), graph=input_graph._plc_graph, min_weight=min_weight, ensemble_size=ensemble_size, do_expensive_check=False, ) df = cudf.DataFrame() df["vertex"] = vertex df["partition"] = partition if input_graph.renumbered: df = input_graph.unrenumber(df, "vertex") if isNx is True: df = df_score_to_dictionary(df, "partition") return df
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/subgraph_extraction.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from typing import Union import warnings import cudf import cugraph from cugraph.structure import Graph from cugraph.utilities.utils import import_optional # FIXME: the networkx.Graph type used in the type annotation for subgraph() is # specified using a string literal to avoid depending on and importing # networkx. Instead, networkx is imported optionally, which may cause a problem # for a type checker if run in an environment where networkx is not installed. networkx = import_optional("networkx") def subgraph( G: Union[Graph, "networkx.Graph"], vertices: Union[cudf.Series, cudf.DataFrame], ) -> Union[Graph, "networkx.Graph"]: """ Compute a subgraph of the existing graph including only the specified vertices. This algorithm works with both directed and undirected graphs and does not actually traverse the edges, but instead simply pulls out any edges that are incident on vertices that are both contained in the vertices list. If no subgraph can be extracted from the vertices provided, a 'None' value will be returned. Parameters ---------- G : cugraph.Graph or networkx.Graph The current implementation only supports weighted graphs. vertices : cudf.Series or cudf.DataFrame Specifies the vertices of the induced subgraph. For multi-column vertices, vertices should be provided as a cudf.DataFrame Returns ------- Sg : cugraph.Graph or networkx.Graph A graph object containing the subgraph induced by the given vertex set. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> verts = np.zeros(3, dtype=np.int32) >>> verts[0] = 0 >>> verts[1] = 1 >>> verts[2] = 2 >>> sverts = cudf.Series(verts) >>> Sg = cugraph.subgraph(G, sverts) """ warning_msg = ( "This call is deprecated. Please call 'cugraph.induced_subgraph()' instead." ) warnings.warn(warning_msg, DeprecationWarning) result_graph, _ = cugraph.induced_subgraph(G, vertices) return result_graph
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/spectral_clustering.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.utilities import ( ensure_cugraph_obj_for_nx, df_score_to_dictionary, ) from pylibcugraph import ( balanced_cut_clustering as pylibcugraph_balanced_cut_clustering, spectral_modularity_maximization as pylibcugraph_spectral_modularity_maximization, analyze_clustering_modularity as pylibcugraph_analyze_clustering_modularity, analyze_clustering_edge_cut as pylibcugraph_analyze_clustering_edge_cut, analyze_clustering_ratio_cut as pylibcugraph_analyze_clustering_ratio_cut, ) from pylibcugraph import ResourceHandle import cudf import numpy as np def spectralBalancedCutClustering( G, num_clusters, num_eigen_vects=2, evs_tolerance=0.00001, evs_max_iter=100, kmean_tolerance=0.00001, kmean_max_iter=100, ): """ Compute a clustering/partitioning of the given graph using the spectral balanced cut method. Parameters ---------- G : cugraph.Graph or networkx.Graph Graph descriptor num_clusters : integer Specifies the number of clusters to find, must be greater than 1 num_eigen_vects : integer, optional Specifies the number of eigenvectors to use. Must be lower or equal to num_clusters. Default is 2 evs_tolerance: float, optional Specifies the tolerance to use in the eigensolver. Default is 0.00001 evs_max_iter: integer, optional Specifies the maximum number of iterations for the eigensolver. Default is 100 kmean_tolerance: float, optional Specifies the tolerance to use in the k-means solver. Default is 0.00001 kmean_max_iter: integer, optional Specifies the maximum number of iterations for the k-means solver. Default is 100 Returns ------- df : cudf.DataFrame GPU data frame containing two cudf.Series of size V: the vertex identifiers and the corresponding cluster assignments. df['vertex'] : cudf.Series contains the vertex identifiers df['cluster'] : cudf.Series contains the cluster assignments Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.spectralBalancedCutClustering(G, 5) """ # Error checking in C++ code G, isNx = ensure_cugraph_obj_for_nx(G) # Check if vertex type is "int32" if ( G.edgelist.edgelist_df.dtypes[0] != np.int32 or G.edgelist.edgelist_df.dtypes[1] != np.int32 ): raise ValueError( "'spectralBalancedCutClustering' requires the input graph's vertex to be " "of type 'int32'" ) vertex, partition = pylibcugraph_balanced_cut_clustering( ResourceHandle(), G._plc_graph, num_clusters, num_eigen_vects, evs_tolerance, evs_max_iter, kmean_tolerance, kmean_max_iter, do_expensive_check=False, ) df = cudf.DataFrame() df["vertex"] = vertex df["cluster"] = partition if G.renumbered: df = G.unrenumber(df, "vertex") if isNx is True: df = df_score_to_dictionary(df, "cluster") return df def spectralModularityMaximizationClustering( G, num_clusters, num_eigen_vects=2, evs_tolerance=0.00001, evs_max_iter=100, kmean_tolerance=0.00001, kmean_max_iter=100, ): """ Compute a clustering/partitioning of the given graph using the spectral modularity maximization method. Parameters ---------- G : cugraph.Graph or networkx.Graph cuGraph graph descriptor. This graph should have edge weights. num_clusters : integer Specifies the number of clusters to find num_eigen_vects : integer, optional Specifies the number of eigenvectors to use. Must be lower or equal to num_clusters. Default is 2 evs_tolerance: float, optional Specifies the tolerance to use in the eigensolver. Default is 0.00001 evs_max_iter: integer, optional Specifies the maximum number of iterations for the eigensolver. Default is 100 kmean_tolerance: float, optional Specifies the tolerance to use in the k-means solver. Default is 0.00001 kmean_max_iter: integer, optional Specifies the maximum number of iterations for the k-means solver. Default is 100 Returns ------- df : cudf.DataFrame GPU data frame containing two cudf.Series of size V: the vertex identifiers and the corresponding cluster assignments. df['vertex'] : cudf.Series contains the vertex identifiers df['cluster'] : cudf.Series contains the cluster assignments Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.spectralModularityMaximizationClustering(G, 5) """ G, isNx = ensure_cugraph_obj_for_nx(G) if ( G.edgelist.edgelist_df.dtypes[0] != np.int32 or G.edgelist.edgelist_df.dtypes[1] != np.int32 ): raise ValueError( "'spectralModularityMaximizationClustering' requires the input graph's " "vertex to be of type 'int32'" ) vertex, partition = pylibcugraph_spectral_modularity_maximization( ResourceHandle(), G._plc_graph, num_clusters, num_eigen_vects, evs_tolerance, evs_max_iter, kmean_tolerance, kmean_max_iter, do_expensive_check=False, ) df = cudf.DataFrame() df["vertex"] = vertex df["cluster"] = partition if G.renumbered: df = G.unrenumber(df, "vertex") if isNx is True: df = df_score_to_dictionary(df, "cluster") return df def analyzeClustering_modularity( G, n_clusters, clustering, vertex_col_name="vertex", cluster_col_name="cluster" ): """ Compute the modularity score for a given partitioning/clustering. The assumption is that “clustering” is the results from a call from a special clustering algorithm and contains columns named “vertex” and “cluster”. Parameters ---------- G : cugraph.Graph or networkx.Graph graph descriptor. This graph should have edge weights. n_clusters : integer Specifies the number of clusters in the given clustering clustering : cudf.DataFrame The cluster assignment to analyze. vertex_col_name : str or list of str, optional (default='vertex') The names of the column in the clustering dataframe identifying the external vertex id cluster_col_name : str, optional (default='cluster') The name of the column in the clustering dataframe identifying the cluster id Returns ------- score : float The computed modularity score Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.spectralBalancedCutClustering(G, 5) >>> score = cugraph.analyzeClustering_modularity(G, 5, df) """ if type(vertex_col_name) is list: if not all(isinstance(name, str) for name in vertex_col_name): raise Exception("vertex_col_name must be list of string") elif type(vertex_col_name) is not str: raise Exception("vertex_col_name must be a string") if type(cluster_col_name) is not str: raise Exception("cluster_col_name must be a string") G, isNx = ensure_cugraph_obj_for_nx(G) if ( G.edgelist.edgelist_df.dtypes[0] != np.int32 or G.edgelist.edgelist_df.dtypes[1] != np.int32 ): raise ValueError( "'analyzeClustering_modularity' requires the input graph's " "vertex to be of type 'int32'" ) if G.renumbered: clustering = G.add_internal_vertex_id( clustering, "vertex", vertex_col_name, drop=True ) if clustering.dtypes[0] != np.int32 or clustering.dtypes[1] != np.int32: raise ValueError( "'analyzeClustering_modularity' requires both the clustering 'vertex' " "and 'cluster' to be of type 'int32'" ) score = pylibcugraph_analyze_clustering_modularity( ResourceHandle(), G._plc_graph, n_clusters, clustering["vertex"], clustering[cluster_col_name], ) return score def analyzeClustering_edge_cut( G, n_clusters, clustering, vertex_col_name="vertex", cluster_col_name="cluster" ): """ Compute the edge cut score for a partitioning/clustering The assumption is that “clustering” is the results from a call from a special clustering algorithm and contains columns named “vertex” and “cluster”. Parameters ---------- G : cugraph.Graph cuGraph graph descriptor n_clusters : integer Specifies the number of clusters in the given clustering clustering : cudf.DataFrame The cluster assignment to analyze. vertex_col_name : str, optional (default='vertex') The name of the column in the clustering dataframe identifying the external vertex id cluster_col_name : str, optional (default='cluster') The name of the column in the clustering dataframe identifying the cluster id Returns ------- score : float The computed edge cut score Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.spectralBalancedCutClustering(G, 5) >>> score = cugraph.analyzeClustering_edge_cut(G, 5, df) """ if type(vertex_col_name) is list: if not all(isinstance(name, str) for name in vertex_col_name): raise Exception("vertex_col_name must be list of string") elif type(vertex_col_name) is not str: raise Exception("vertex_col_name must be a string") if type(cluster_col_name) is not str: raise Exception("cluster_col_name must be a string") G, isNx = ensure_cugraph_obj_for_nx(G) if ( G.edgelist.edgelist_df.dtypes[0] != np.int32 or G.edgelist.edgelist_df.dtypes[1] != np.int32 ): raise ValueError( "'analyzeClustering_edge_cut' requires the input graph's vertex to be " "of type 'int32'" ) if G.renumbered: clustering = G.add_internal_vertex_id( clustering, "vertex", vertex_col_name, drop=True ) if clustering.dtypes[0] != np.int32 or clustering.dtypes[1] != np.int32: raise ValueError( "'analyzeClustering_edge_cut' requires both the clustering 'vertex' " "and 'cluster' to be of type 'int32'" ) score = pylibcugraph_analyze_clustering_edge_cut( ResourceHandle(), G._plc_graph, n_clusters, clustering["vertex"], clustering[cluster_col_name], ) return score def analyzeClustering_ratio_cut( G, n_clusters, clustering, vertex_col_name="vertex", cluster_col_name="cluster" ): """ Compute the ratio cut score for a partitioning/clustering Parameters ---------- G : cugraph.Graph cuGraph graph descriptor. This graph should have edge weights. n_clusters : integer Specifies the number of clusters in the given clustering clustering : cudf.DataFrame The cluster assignment to analyze. vertex_col_name : str, optional (default='vertex') The name of the column in the clustering dataframe identifying the external vertex id cluster_col_name : str, optional (default='cluster') The name of the column in the clustering dataframe identifying the cluster id Returns ------- score : float The computed ratio cut score Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> df = cugraph.spectralBalancedCutClustering(G, 5) >>> score = cugraph.analyzeClustering_ratio_cut(G, 5, df, 'vertex', ... 'cluster') """ if type(vertex_col_name) is list: if not all(isinstance(name, str) for name in vertex_col_name): raise Exception("vertex_col_name must be list of string") elif type(vertex_col_name) is not str: raise Exception("vertex_col_name must be a string") if type(cluster_col_name) is not str: raise Exception("cluster_col_name must be a string") if G.renumbered: clustering = G.add_internal_vertex_id( clustering, "vertex", vertex_col_name, drop=True ) if clustering.dtypes[0] != np.int32 or clustering.dtypes[1] != np.int32: raise ValueError( "'analyzeClustering_ratio_cut' requires both the clustering 'vertex' " "and 'cluster' to be of type 'int32'" ) score = pylibcugraph_analyze_clustering_ratio_cut( ResourceHandle(), G._plc_graph, n_clusters, clustering["vertex"], clustering[cluster_col_name], ) return score
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/louvain.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from typing import Union, Tuple from cugraph.structure import Graph from cugraph.utilities import ( is_nx_graph_type, ensure_cugraph_obj_for_nx, df_score_to_dictionary, ) import cudf import warnings from pylibcugraph import louvain as pylibcugraph_louvain from pylibcugraph import ResourceHandle from cugraph.utilities.utils import import_optional # FIXME: the networkx.Graph type used in type annotations is specified # using a string literal to avoid depending on and importing networkx. # Instead, networkx is imported optionally, which may cause a problem # for a type checker if run in an environment where networkx is not installed. networkx = import_optional("networkx") VERTEX_COL_NAME = "vertex" CLUSTER_ID_COL_NAME = "partition" # FIXME: max_level should default to 100 once max_iter is removed def louvain( G: Union[Graph, "networkx.Graph"], max_level: Union[int, None] = None, max_iter: Union[int, None] = None, resolution: float = 1.0, threshold: float = 1e-7, ) -> Tuple[Union[cudf.DataFrame, dict], float]: """ Compute the modularity optimizing partition of the input graph using the Louvain method It uses the Louvain method described in: VD Blondel, J-L Guillaume, R Lambiotte and E Lefebvre: Fast unfolding of community hierarchies in large networks, J Stat Mech P10008 (2008), http://arxiv.org/abs/0803.0476 Parameters ---------- G : cugraph.Graph or NetworkX Graph The graph descriptor should contain the connectivity information and weights. The adjacency list will be computed if not already present. The current implementation only supports undirected graphs. max_level : integer, optional (default=100) This controls the maximum number of levels of the Louvain algorithm. When specified the algorithm will terminate after no more than the specified number of levels. No error occurs when the algorithm terminates early in this manner. If max_level > 500, it will be set to 500 and a warning is emitted in order to prevent excessive runtime. max_iter : integer, optional (default=None) This parameter is deprecated in favor of max_level. Previously it was used to control the maximum number of levels of the Louvain algorithm. resolution: float, optional (default=1.0) Called gamma in the modularity formula, this changes the size of the communities. Higher resolutions lead to more smaller communities, lower resolutions lead to fewer larger communities. Defaults to 1. threshold: float Modularity gain threshold for each level. If the gain of modularity between 2 levels of the algorithm is less than the given threshold then the algorithm stops and returns the resulting communities. Defaults to 1e-7. Returns ------- result: cudf.DataFrame or dict If input graph G is of type cugraph.Graph, a GPU dataframe with two columns. result[VERTEX_COL_NAME] : cudf.Series Contains the vertex identifiers result[CLUSTER_ID_COL_NAME] : cudf.Series Contains the partition assigned to the vertices If input graph G is of type networkx.Graph, a dict Dictionary of vertices and their partition ids. modularity_score : float A floating point number containing the global modularity score of the partitioning. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> parts = cugraph.louvain(G) """ # FIXME: Onece the graph construction calls support isolated vertices through # the C API (the C++ interface already supports this) then there will be # no need to compute isolated vertices here. isolated_vertices = list() if is_nx_graph_type(type(G)): isolated_vertices = [v for v in range(G.number_of_nodes()) if G.degree[v] == 0] else: # FIXME: Gather isolated vertices of G pass G, isNx = ensure_cugraph_obj_for_nx(G) if G.is_directed(): raise ValueError("input graph must be undirected") # FIXME: This max_iter logic and the max_level defaulting can be deleted # in favor of defaulting max_level in call once max_iter is deleted if max_iter: if max_level: raise ValueError( "max_iter is deprecated. Cannot specify both max_iter and max_level" ) warning_msg = ( "max_iter has been renamed max_level. Use of max_iter is " "deprecated and will no longer be supported in the next releases." ) warnings.warn(warning_msg, FutureWarning) max_level = max_iter if max_level is None: max_level = 100 if max_level > 500: w_msg = "max_level is set too high, clamping it down to 500." warnings.warn(w_msg) max_level = 500 vertex, partition, modularity_score = pylibcugraph_louvain( resource_handle=ResourceHandle(), graph=G._plc_graph, max_level=max_level, threshold=threshold, resolution=resolution, do_expensive_check=False, ) result = cudf.DataFrame() result[VERTEX_COL_NAME] = vertex result[CLUSTER_ID_COL_NAME] = partition if len(isolated_vertices) > 0: unique_cids = result[CLUSTER_ID_COL_NAME].unique() max_cluster_id = -1 if len(result) == 0 else unique_cids.max() isolated_vtx_and_cids = cudf.DataFrame() isolated_vtx_and_cids[VERTEX_COL_NAME] = isolated_vertices isolated_vtx_and_cids[CLUSTER_ID_COL_NAME] = [ (max_cluster_id + i + 1) for i in range(len(isolated_vertices)) ] result = cudf.concat( [result, isolated_vtx_and_cids], ignore_index=True, sort=False ) if G.renumbered and len(G.input_df) > 0: result = G.unrenumber(result, VERTEX_COL_NAME) if isNx is True: result = df_score_to_dictionary(result, CLUSTER_ID_COL_NAME) return result, modularity_score
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/ktruss_subgraph.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.structure.graph_classes import Graph from typing import Union from cugraph.utilities import ( ensure_cugraph_obj_for_nx, cugraph_to_nx, ) from pylibcugraph import k_truss_subgraph as pylibcugraph_k_truss_subgraph from pylibcugraph import ResourceHandle import warnings from numba import cuda import cudf from cugraph.utilities.utils import import_optional # FIXME: the networkx.Graph type used in the type annotation for # ktruss_subgraph() is specified using a string literal to avoid depending on # and importing networkx. Instead, networkx is imported optionally, which may # cause a problem for a type checker if run in an environment where networkx is # not installed. networkx = import_optional("networkx") # FIXME: special case for ktruss on CUDA 11.4: an 11.4 bug causes ktruss to # crash in that environment. Allow ktruss to import on non-11.4 systems, but # raise an exception if ktruss is directly imported on 11.4. def _ensure_compatible_cuda_version(): try: cuda_version = cuda.runtime.get_version() except cuda.cudadrv.runtime.CudaRuntimeAPIError: cuda_version = "n/a" unsupported_cuda_version = (11, 4) if cuda_version == unsupported_cuda_version: ver_string = ".".join([str(n) for n in unsupported_cuda_version]) raise NotImplementedError( "k_truss is not currently supported in CUDA" f" {ver_string} environments." ) def k_truss( G: Union[Graph, "networkx.Graph"], k: int ) -> Union[Graph, "networkx.Graph"]: """ Returns the K-Truss subgraph of a graph for a specific k. NOTE: this function is currently not available on CUDA 11.4 systems. The k-truss of a graph is a subgraph where each edge is part of at least (k−2) triangles. K-trusses are used for finding tighlty knit groups of vertices in a graph. A k-truss is a relaxation of a k-clique in the graph and was define in [1]. Finding cliques is computationally demanding and finding the maximal k-clique is known to be NP-Hard. Parameters ---------- G : cuGraph.Graph or networkx.Graph cuGraph graph descriptor with connectivity information. k-Trusses are defined for only undirected graphs as they are defined for undirected triangle in a graph. k : int The desired k to be used for extracting the k-truss subgraph. Returns ------- G_truss : cuGraph.Graph or networkx.Graph A cugraph graph descriptor with the k-truss subgraph for the given k. The networkx graph will NOT have all attributes copied over Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> k_subgraph = cugraph.k_truss(G, 3) """ _ensure_compatible_cuda_version() G, isNx = ensure_cugraph_obj_for_nx(G) if isNx is True: k_sub = ktruss_subgraph(G, k) S = cugraph_to_nx(k_sub) return S else: return ktruss_subgraph(G, k) # FIXME: merge this function with k_truss def ktruss_subgraph( G: Union[Graph, "networkx.Graph"], k: int, use_weights=True, # deprecated ) -> Graph: """ Returns the K-Truss subgraph of a graph for a specific k. NOTE: this function is currently not available on CUDA 11.4 systems. The k-truss of a graph is a subgraph where each edge is part of at least (k−2) triangles. K-trusses are used for finding tighlty knit groups of vertices in a graph. A k-truss is a relaxation of a k-clique in the graph and was define in [1]. Finding cliques is computationally demanding and finding the maximal k-clique is known to be NP-Hard. In contrast, finding a k-truss is computationally tractable as its key building block, namely triangle counting, can be executed in polnymomial time.Typically, it takes many iterations of triangle counting to find the k-truss of a graph. Yet these iterations operate on a weakly monotonically shrinking graph. Therefore, finding the k-truss of a graph can be done in a fairly reasonable amount of time. The solution in cuGraph is based on a GPU algorithm first shown in [2] and uses the triangle counting algorithm from [3]. References ---------- [1] Cohen, J., "Trusses: Cohesive subgraphs for social network analysis" National security agency technical report, 2008 [2] O. Green, J. Fox, E. Kim, F. Busato, et al. “Quickly Finding a Truss in a Haystack” IEEE High Performance Extreme Computing Conference (HPEC), 2017 https://doi.org/10.1109/HPEC.2017.8091038 [3] O. Green, P. Yalamanchili, L.M. Munguia, “Fast Triangle Counting on GPU” Irregular Applications: Architectures and Algorithms (IA3), 2014 Parameters ---------- G : cuGraph.Graph cuGraph graph descriptor with connectivity information. k-Trusses are defined for only undirected graphs as they are defined for undirected triangle in a graph. The current implementation only supports undirected graphs. k : int The desired k to be used for extracting the k-truss subgraph. use_weights : bool, optional (default=True) Whether the output should contain the edge weights if G has them. Deprecated: If 'weights' were passed at the graph creation, they will be used. Returns ------- G_truss : cuGraph.Graph A cugraph graph descriptor with the k-truss subgraph for the given k. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> k_subgraph = cugraph.ktruss_subgraph(G, 3) """ _ensure_compatible_cuda_version() KTrussSubgraph = Graph() if G.is_directed(): raise ValueError("input graph must be undirected") if use_weights: warning_msg = ( "The use_weights flag is deprecated " "and will be removed in the next release. if weights " "were passed at the graph creation, they will be used." ) warnings.warn(warning_msg, FutureWarning) sources, destinations, edge_weights, _ = pylibcugraph_k_truss_subgraph( resource_handle=ResourceHandle(), graph=G._plc_graph, k=k, do_expensive_check=True, ) subgraph_df = cudf.DataFrame() subgraph_df["src"] = sources subgraph_df["dst"] = destinations if edge_weights is not None: subgraph_df["weight"] = edge_weights if G.renumbered: subgraph_df = G.unrenumber(subgraph_df, "src") subgraph_df = G.unrenumber(subgraph_df, "dst") if G.edgelist.weights: KTrussSubgraph.from_cudf_edgelist( subgraph_df, source="src", destination="dst", edge_attr="weight" ) else: KTrussSubgraph.from_cudf_edgelist(subgraph_df, source="src", destination="dst") return KTrussSubgraph
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/__init__.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.community.louvain import louvain from cugraph.community.leiden import leiden from cugraph.community.ecg import ecg from cugraph.community.spectral_clustering import ( spectralBalancedCutClustering, spectralModularityMaximizationClustering, analyzeClustering_modularity, analyzeClustering_edge_cut, analyzeClustering_ratio_cut, ) from cugraph.community.subgraph_extraction import subgraph from cugraph.community.induced_subgraph import induced_subgraph from cugraph.community.triangle_count import triangle_count from cugraph.community.ktruss_subgraph import ktruss_subgraph from cugraph.community.ktruss_subgraph import k_truss from cugraph.community.egonet import ego_graph from cugraph.community.egonet import batched_ego_graphs
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/leiden.py
# Copyright (c) 2019-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from pylibcugraph import leiden as pylibcugraph_leiden from pylibcugraph import ResourceHandle from cugraph.structure import Graph import cudf from typing import Union, Tuple from cugraph.utilities import ( ensure_cugraph_obj_for_nx, df_score_to_dictionary, ) from cugraph.utilities.utils import import_optional # FIXME: the networkx.Graph type used in the type annotation for # leiden() is specified using a string literal to avoid depending on # and importing networkx. Instead, networkx is imported optionally, which may # cause a problem for a type checker if run in an environment where networkx is # not installed. networkx = import_optional("networkx") def leiden( G: Union[Graph, "networkx.Graph"], max_iter: int = 100, resolution: float = 1.0, random_state: int = None, theta: int = 1.0, ) -> Tuple[cudf.DataFrame, float]: """ Compute the modularity optimizing partition of the input graph using the Leiden algorithm It uses the Leiden method described in: Traag, V. A., Waltman, L., & van Eck, N. J. (2019). From Louvain to Leiden: guaranteeing well-connected communities. Scientific reports, 9(1), 5233. doi: 10.1038/s41598-019-41695-z Parameters ---------- G : cugraph.Graph cuGraph graph descriptor of type Graph The current implementation only supports undirected weighted graphs. The adjacency list will be computed if not already present. max_iter : integer, optional (default=100) This controls the maximum number of levels/iterations of the Leiden algorithm. When specified the algorithm will terminate after no more than the specified number of iterations. No error occurs when the algorithm terminates early in this manner. resolution: float, optional (default=1.0) Called gamma in the modularity formula, this changes the size of the communities. Higher resolutions lead to more smaller communities, lower resolutions lead to fewer larger communities. Defaults to 1. random_state: int, optional(default=None) Random state to use when generating samples. Optional argument, defaults to a hash of process id, time, and hostname. theta: float, optional (default=1.0) Called theta in the Leiden algorithm, this is used to scale modularity gain in Leiden refinement phase, to compute the probability of joining a random leiden community. Returns ------- parts : cudf.DataFrame GPU data frame of size V containing two columns the vertex id and the partition id it is assigned to. df['vertex'] : cudf.Series Contains the vertex identifiers df['partition'] : cudf.Series Contains the partition assigned to the vertices modularity_score : float a floating point number containing the global modularity score of the partitioning. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> parts, modularity_score = cugraph.leiden(G) """ G, isNx = ensure_cugraph_obj_for_nx(G) if G.is_directed(): raise ValueError("input graph must be undirected") vertex, partition, modularity_score = pylibcugraph_leiden( resource_handle=ResourceHandle(), random_state=random_state, graph=G._plc_graph, max_level=max_iter, resolution=resolution, theta=theta, do_expensive_check=False, ) df = cudf.DataFrame() df["vertex"] = vertex df["partition"] = partition if G.renumbered: parts = G.unrenumber(df, "vertex") else: parts = df if isNx is True: parts = df_score_to_dictionary(df, "partition") return parts, modularity_score
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/community/egonet.py
# Copyright (c) 2021-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.utilities import ( ensure_cugraph_obj, is_nx_graph_type, ) from cugraph.utilities import cugraph_to_nx import cudf from pylibcugraph import ego_graph as pylibcugraph_ego_graph from pylibcugraph import ResourceHandle import warnings def _convert_graph_to_output_type(G, input_type): """ Given a cugraph.Graph, convert it to a new type appropriate for the graph algos in this module, based on input_type. """ if is_nx_graph_type(input_type): return cugraph_to_nx(G) else: return G def _convert_df_series_to_output_type(df, offsets, input_type): """ Given a cudf.DataFrame df, convert it to a new type appropriate for the graph algos in this module, based on input_type. """ if is_nx_graph_type(input_type): return df.to_pandas(), offsets.values_host.tolist() else: return df, offsets def ego_graph(G, n, radius=1, center=True, undirected=None, distance=None): """ Compute the induced subgraph of neighbors centered at node n, within a given radius. Parameters ---------- G : cugraph.Graph, networkx.Graph, CuPy or SciPy sparse matrix Graph or matrix object, which should contain the connectivity information. Edge weights, if present, should be single or double precision floating point values. n : integer or list, cudf.Series, cudf.DataFrame A single node as integer or a cudf.DataFrame if nodes are represented with multiple columns. If a cudf.DataFrame is provided, only the first row is taken as the node input. radius: integer, optional (default=1) Include all neighbors of distance<=radius from n. center: bool, optional Defaults to True. False is not supported undirected: bool, optional This parameter is here for NetworkX compatibility and is ignored distance: key, optional (default=None) This parameter is here for NetworkX compatibility and is ignored Returns ------- G_ego : cuGraph.Graph or networkx.Graph A graph descriptor with a minimum spanning tree or forest. The networkx graph will not have all attributes copied over Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> ego_graph = cugraph.ego_graph(G, 1, radius=2) """ (G, input_type) = ensure_cugraph_obj(G, nx_weight_attr="weight") result_graph = type(G)(directed=G.is_directed()) if undirected is not None: warning_msg = ( "The parameter 'undirected' is deprecated and " "will be removed in the next release" ) warnings.warn(warning_msg, PendingDeprecationWarning) if isinstance(n, (int, list)): n = cudf.Series(n) if isinstance(n, cudf.Series): if G.renumbered is True: n = G.lookup_internal_vertex_id(n) elif isinstance(n, cudf.DataFrame): if G.renumbered is True: n = G.lookup_internal_vertex_id(n, n.columns) else: raise TypeError( f"'n' must be either an integer or a list or a cudf.Series" f" or a cudf.DataFrame, got: {type(n)}" ) # Match the seed to the vertex dtype n_type = G.edgelist.edgelist_df["src"].dtype n = n.astype(n_type) do_expensive_check = False source, destination, weight, _ = pylibcugraph_ego_graph( resource_handle=ResourceHandle(), graph=G._plc_graph, source_vertices=n, radius=radius, do_expensive_check=do_expensive_check, ) df = cudf.DataFrame() df["src"] = source df["dst"] = destination if weight is not None: df["weight"] = weight if G.renumbered: df, src_names = G.unrenumber(df, "src", get_column_names=True) df, dst_names = G.unrenumber(df, "dst", get_column_names=True) else: # FIXME: The original 'src' and 'dst' are not stored in 'simpleGraph' src_names = "src" dst_names = "dst" if G.edgelist.weights: result_graph.from_cudf_edgelist( df, source=src_names, destination=dst_names, edge_attr="weight" ) else: result_graph.from_cudf_edgelist(df, source=src_names, destination=dst_names) return _convert_graph_to_output_type(result_graph, input_type) def batched_ego_graphs(G, seeds, radius=1, center=True, undirected=None, distance=None): """ Compute the induced subgraph of neighbors for each node in seeds within a given radius. Parameters ---------- G : cugraph.Graph, networkx.Graph, CuPy or SciPy sparse matrix Graph or matrix object, which should contain the connectivity information. Edge weights, if present, should be single or double precision floating point values. seeds : cudf.Series or list or cudf.DataFrame Specifies the seeds of the induced egonet subgraphs. radius: integer, optional (default=1) Include all neighbors of distance<=radius from n. center: bool, optional Defaults to True. False is not supported undirected: bool, optional Defaults to False. True is not supported distance: key, optional (default=None) Distances are counted in hops from n. Other cases are not supported. Returns ------- ego_edge_lists : cudf.DataFrame or pandas.DataFrame GPU data frame containing all induced sources identifiers, destination identifiers, edge weights seeds_offsets: cudf.Series Series containing the starting offset in the returned edge list for each seed. Examples -------- >>> from cugraph.datasets import karate >>> G = karate.get_graph(download=True) >>> b_ego_graph, offsets = cugraph.batched_ego_graphs(G, seeds=[1,5], ... radius=2) """ (G, input_type) = ensure_cugraph_obj(G, nx_weight_attr="weight") if seeds is not None: if isinstance(seeds, int): seeds = [seeds] if isinstance(seeds, list): seeds = cudf.Series(seeds) if G.renumbered is True: if isinstance(seeds, cudf.DataFrame): seeds = G.lookup_internal_vertex_id(seeds, seeds.columns) else: seeds = G.lookup_internal_vertex_id(seeds) # Match the seed to the vertex dtype seeds_type = G.edgelist.edgelist_df["src"].dtype seeds = seeds.astype(seeds_type) do_expensive_check = False source, destination, weight, offset = pylibcugraph_ego_graph( resource_handle=ResourceHandle(), graph=G._plc_graph, source_vertices=seeds, radius=radius, do_expensive_check=do_expensive_check, ) offsets = cudf.Series(offset) df = cudf.DataFrame() df["src"] = source df["dst"] = destination df["weight"] = weight if G.renumbered: df = G.unrenumber(df, "src", preserve_order=True) df = G.unrenumber(df, "dst", preserve_order=True) return _convert_df_series_to_output_type(df, offsets, input_type)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/internals/CMakeLists.txt
# ============================================================================= # Copyright (c) 2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except # in compliance with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software distributed under the License # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express # or implied. See the License for the specific language governing permissions and limitations under # the License. # ============================================================================= set(cython_sources internals.pyx) set(linked_libraries cugraph::cugraph) rapids_cython_create_modules( CXX SOURCE_FILES "${cython_sources}" LINKED_LIBRARIES "${linked_libraries}" MODULE_PREFIX internals_ ASSOCIATED_TARGETS cugraph ) target_include_directories(internals_internals PRIVATE "${CMAKE_CURRENT_LIST_DIR}")
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/internals/internals.pyx
# Copyright (c) 2020-2021, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # cython: profile=False # distutils: language = c++ # cython: embedsignature = True # cython: language_level = 3 from libc.stdint cimport uintptr_t from numba.cuda.api import from_cuda_array_interface import numpy as np cdef extern from "Python.h": cdef cppclass PyObject cdef extern from "callbacks_implems.hpp" namespace "cugraph::internals": cdef cppclass Callback: pass cdef cppclass DefaultGraphBasedDimRedCallback(Callback): void setup(int n, int d) except + void on_preprocess_end(void *positions) except + void on_epoch_end(void *positions) except + void on_train_end(void *positions) except + PyObject* pyCallbackClass cdef class PyCallback: def get_numba_matrix(self, positions, shape, typestr): sizeofType = 4 if typestr == "float32" else 8 desc = { 'shape': shape, 'strides': (sizeofType, shape[0]*sizeofType), 'typestr': typestr, 'data': [positions], 'order': 'C', 'version': 1 } return from_cuda_array_interface(desc) cdef class GraphBasedDimRedCallback(PyCallback): """ Usage ----- class CustomCallback(GraphBasedDimRedCallback): def on_preprocess_end(self, positions): print(positions.copy_to_host()) def on_epoch_end(self, positions): print(positions.copy_to_host()) def on_train_end(self, positions): print(positions.copy_to_host()) """ cdef DefaultGraphBasedDimRedCallback native_callback def __init__(self): self.native_callback.pyCallbackClass = <PyObject *><void*>self def get_native_callback(self): return <uintptr_t>&(self.native_callback)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/internals/__init__.py
# Copyright (c) 2020-2021, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.internals.internals import GraphBasedDimRedCallback
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/internals/callbacks_implems.hpp
/* * Copyright (c) 2020-2023, NVIDIA CORPORATION. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ #pragma once #include <Python.h> #include <cugraph/legacy/internals.hpp> #include <iostream> namespace cugraph { namespace internals { class DefaultGraphBasedDimRedCallback : public GraphBasedDimRedCallback { public: PyObject* get_numba_matrix(void* positions) { PyObject* pycl = (PyObject*)this->pyCallbackClass; if (isFloat) { return PyObject_CallMethod( pycl, "get_numba_matrix", "(l(ll)s)", positions, n, n_components, "float32"); } else { return PyObject_CallMethod( pycl, "get_numba_matrix", "(l(ll)s)", positions, n, n_components, "float64"); } } void on_preprocess_end(void* positions) override { PyObject* numba_matrix = get_numba_matrix(positions); PyObject* res = PyObject_CallMethod(this->pyCallbackClass, "on_preprocess_end", "(O)", numba_matrix); Py_DECREF(numba_matrix); Py_DECREF(res); } void on_epoch_end(void* positions) override { PyObject* numba_matrix = get_numba_matrix(positions); PyObject* res = PyObject_CallMethod(this->pyCallbackClass, "on_epoch_end", "(O)", numba_matrix); Py_DECREF(numba_matrix); Py_DECREF(res); } void on_train_end(void* positions) override { PyObject* numba_matrix = get_numba_matrix(positions); PyObject* res = PyObject_CallMethod(this->pyCallbackClass, "on_train_end", "(O)", numba_matrix); Py_DECREF(numba_matrix); Py_DECREF(res); } public: PyObject* pyCallbackClass; }; } // namespace internals } // namespace cugraph
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/__init__.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.utilities.api_tools import experimental_warning_wrapper from cugraph.utilities.api_tools import deprecated_warning_wrapper from cugraph.utilities.api_tools import promoted_experimental_warning_wrapper from cugraph.structure.property_graph import EXPERIMENTAL__PropertyGraph PropertyGraph = experimental_warning_wrapper(EXPERIMENTAL__PropertyGraph) from cugraph.structure.property_graph import EXPERIMENTAL__PropertySelection PropertySelection = experimental_warning_wrapper(EXPERIMENTAL__PropertySelection) from cugraph.dask.structure.mg_property_graph import EXPERIMENTAL__MGPropertyGraph MGPropertyGraph = experimental_warning_wrapper(EXPERIMENTAL__MGPropertyGraph) from cugraph.dask.structure.mg_property_graph import EXPERIMENTAL__MGPropertySelection MGPropertySelection = experimental_warning_wrapper(EXPERIMENTAL__MGPropertySelection) # FIXME: Remove experimental.triangle_count next release from cugraph.community.triangle_count import triangle_count triangle_count = promoted_experimental_warning_wrapper(triangle_count) from cugraph.experimental.components.scc import EXPERIMENTAL__strong_connected_component strong_connected_component = experimental_warning_wrapper( EXPERIMENTAL__strong_connected_component ) from cugraph.experimental.structure.bicliques import EXPERIMENTAL__find_bicliques find_bicliques = deprecated_warning_wrapper( experimental_warning_wrapper(EXPERIMENTAL__find_bicliques) ) from cugraph.gnn.data_loading import EXPERIMENTAL__BulkSampler BulkSampler = experimental_warning_wrapper(EXPERIMENTAL__BulkSampler) from cugraph.link_prediction.jaccard import jaccard, jaccard_coefficient jaccard = promoted_experimental_warning_wrapper(jaccard) jaccard_coefficient = promoted_experimental_warning_wrapper(jaccard_coefficient) from cugraph.link_prediction.sorensen import sorensen, sorensen_coefficient sorensen = promoted_experimental_warning_wrapper(sorensen) sorensen_coefficient = promoted_experimental_warning_wrapper(sorensen_coefficient) from cugraph.link_prediction.overlap import overlap, overlap_coefficient overlap = promoted_experimental_warning_wrapper(overlap) overlap_coefficient = promoted_experimental_warning_wrapper(overlap_coefficient)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/gnn/__init__.py
# Copyright (c) 2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.gnn.data_loading import EXPERIMENTAL__BulkSampler from cugraph.utilities.api_tools import experimental_warning_wrapper BulkSampler = experimental_warning_wrapper(EXPERIMENTAL__BulkSampler)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/components/scc.py
# Copyright (c) 2019-2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import cudf import cugraph import numpy as np # TRIM Process: # - removed single vertex componenets # - select vertex with highest out degree # - forwards BFS # - backward BFS # - compute intersection = components # - remove component # - repeat def EXPERIMENTAL__strong_connected_component(source, destination): """ Generate the strongly connected components using the FW-BW-TRIM approach, but skipping the trimming) Parameters ---------- source : cudf.Series A cudf series that contains the source side of an edge list destination : cudf.Series A cudf series that contains the destination side of an edge list Returns ------- cdf : cudf.DataFrame - a dataframe for components df['vertex'] - the vertex ID df['id'] - the component ID sdf : cudf.DataFrame - a dataframe with single vertex components df['vertex'] - the vertex ID count - int - the number of components found Examples -------- >>> # M = read_mtx_file(graph_file) >>> # sources = cudf.Series(M.row) >>> # destinations = cudf.Series(M.col) >>> # components, single_components, count = >>> # cugraph.strong_connected_component(source, destination) """ # FIXME: Uncomment out the above example max_value = np.iinfo(np.int32).max # NOQA # create the FW and BW graphs - this version dopes nopt modify the graphs G_fw = cugraph.Graph() G_bw = cugraph.Graph() G_fw.add_edge_list(source, destination) G_bw.add_edge_list(destination, source) # get a list of vertices and sort the list on out_degree d = G_fw.degrees() d = d.sort_values(by="out_degree", ascending=False) num_verts = len(d) # create space for the answers components = [None] * num_verts single_components = [None] * num_verts # Counts - aka array indexies count = 0 single_count = 0 # remove vertices that cannot be in a component bad = d.query("in_degree == 0 or out_degree == 0") if len(bad): bad = bad.drop(["in_degree", "out_degree"]) single_components[single_count] = bad single_count = single_count + 1 d = _filter_list(d, bad) # ----- Start processing ----- while len(d) > 0: v = d["vertex"][0] # compute the forward BFS bfs_fw = cugraph.bfs(G_fw, v) bfs_fw = bfs_fw.query("distance != @max_value") # Now backwards bfs_bw = cugraph.bfs(G_bw, v) bfs_bw = bfs_bw.query("distance != @max_value") # intersection common = bfs_fw.merge(bfs_bw, on="vertex", how="inner") if len(common) > 1: common["id"] = v components[count] = common d = _filter_list(d, common) count = count + 1 else: # v is an isolated vertex vdf = cudf.DataFrame() vdf["vertex"] = v single_components[single_count] = vdf single_count = single_count + 1 d = d.iloc[1:] # end of loop until vertex queue is empty comp = _compress_array(components, count) sing = _compress_array(single_components, single_count) return comp, sing, count # --------- def _filter_list(vert_list, drop_list): t = cudf.DataFrame() t["vertex"] = drop_list["vertex"] t["d"] = 0 df = vert_list.merge(t, on="vertex", how="left") df["d"] = df["d"].fillna(1) df = df.query("d == 1") df.drop("d", inplace=True) return df def _compress_array(a, length): tmp = cudf.DataFrame() if length > 0: tmp_a = [None] * length for i in range(length): tmp_a[i] = a[i] tmp = cudf.concat(tmp_a) return tmp
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/components/__init__.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/dask/__init__.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/__init__.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/Graph.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import networkx as nx class Graph(nx.Graph): """ Class which extends NetworkX Graph class. It provides original NetworkX functionality and will be overridden as this compatibility layer moves functionality to gpus in future releases. """ pass
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/DiGraph.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import networkx as nx class DiGraph(nx.DiGraph): """ Class which extends NetworkX DiGraph class. It provides original NetworkX functionality and will be overridden as this compatibility layer moves functionality to gpus in future releases. """ pass
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/__init__.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import importlib from types import ModuleType import sys # FIXME: only perform the NetworkX imports below if NetworkX is installed. If # it's determined that NetworkX is required to use nx compat, then the contents # of this entire namespace may have to be optional, or packaged separately with # a hard dependency on NetworkX. # Start by populating this namespace with the same contents as # networkx/__init__.py from networkx import * # Override the individual NetworkX objects loaded above with the cugraph.nx # compat equivalents. This means if an equivalent compat obj is not available, # the standard NetworkX obj will be used. # # Each cugraph obj should have the same module path as the # NetworkX obj it isoverriding, and the submodules along the hierarchy should # each import the same sub objects/modules as NetworkX does. For example, # in NetworkX, "pagerank" is a function in # "networkx/algorithms/link_analysis/pagerank_alg.py", and is # directly imported in the namespaces "networkx.algorithms.link_analysis", # "networkx.algorithms", and "networkx". Therefore, the cugraph # compat pagerank should be defined in a module of the same name and # also be present in the same namespaces. # Refer to the networkx __init__.py files when adding new overriding # modules to ensure the same paths and used and namespaces are populated. from cugraph.experimental.compat.nx import algorithms from cugraph.experimental.compat.nx.algorithms import * from cugraph.experimental.compat.nx.algorithms import link_analysis from cugraph.experimental.compat.nx.algorithms.link_analysis import * # Recursively import all of the NetworkX modules into equivalent submodules # under this package. The above "from networkx import *" handles names in this # namespace, but it will not create the equivalent networkx submodule # hierarchy. For example, a user could expect to "import cugraph.nx.drawing", # which should simply redirect to "networkx.drawing". # # This can be accomplished by updating sys.modules with the import path and # module object of each NetworkX submodule in the NetworkX package hierarchy, # but only for module paths that have not been added yet (otherwise this would # overwrite the overides above). _visited = set() def _import_submodules_recursively(obj, mod_path): # Since modules can freely import any other modules, immediately mark this # obj as visited so submodules that import it are not re-examined # infinitely. _visited.add(obj) for name in dir(obj): sub_obj = getattr(obj, name) if type(sub_obj) is ModuleType: sub_mod_path = f"{mod_path}.{name}" # Do not overwrite modules that are already present, such as those # intended to override which were imported separately above. if sub_mod_path not in sys.modules: sys.modules[sub_mod_path] = sub_obj if sub_obj not in _visited: _import_submodules_recursively(sub_obj, sub_mod_path) _import_submodules_recursively(importlib.import_module("networkx"), __name__) del _visited del _import_submodules_recursively # At this point, individual types that cugraph.nx are overriding # could be used to override the corresponding types *inside* the # networkx modules imported above. For example, the networkx graph generators # will still return networkx.Graph objects instead of cugraph.nx.Graph # objects (unless the user knows to pass a "create_using" arg, if available). # For specific overrides, assignments could be made in the imported # a networkx modules so cugraph.nx types are used by default. # NOTE: this has the side-effect of causing all networkx # imports in this python process/interpreter to use the override (ie. the user # won't be able to use the original networkx types, # even from a networkx import)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/algorithms/__init__.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from networkx.algorithms import * from cugraph.experimental.compat.nx.algorithms.link_analysis import * from cugraph.experimental.compat.nx.algorithms import link_analysis
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/algorithms
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/algorithms/link_analysis/__init__.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from networkx.algorithms.link_analysis import * from cugraph.experimental.compat.nx.algorithms.link_analysis.pagerank_alg import *
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/algorithms
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/compat/nx/algorithms/link_analysis/pagerank_alg.py
# Copyright (c) 2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import cugraph import cugraph.utilities import cudf import numpy as np def create_cudf_from_dict(dict_in): """ converts python dictionary to a cudf.Dataframe as needed by this cugraph pagerank call. Parameters ---------- dictionary with node ids(key) and values Returns ------- a cudf DataFrame of (vertex)ids and values. """ if not (isinstance(dict_in, dict)): raise TypeError("type_name must be a dict, got: " f"{type(dict_in)}") # FIXME: Looking to replacing fromiter with rename and # compare performance k = np.fromiter(dict_in.keys(), dtype="int32") v = np.fromiter(dict_in.values(), dtype="float32") df = cudf.DataFrame({"vertex": k, "values": v}) return df def pagerank( G, alpha=0.85, personalization=None, max_iter=100, tol=1.0e-6, nstart=None, weight="weight", dangling=None, ): """ Calls the cugraph pagerank algorithm taking in a networkX object. In future releases it will maintain compatibility but will migrate more of the workflow to the GPU. Parameters ---------- G : networkx.Graph alpha : float, optional (default=0.85) The damping factor alpha represents the probability to follow an outgoing edge, standard value is 0.85. Thus, 1.0-alpha is the probability to “teleport” to a random vertex. Alpha should be greater than 0.0 and strictly lower than 1.0. personalization : dictionary, optional (default=None) dictionary comes from networkx is converted to a dataframe containing the personalization information. max_iter : int, optional (default=100) The maximum number of iterations before an answer is returned. This can be used to limit the execution time and do an early exit before the solver reaches the convergence tolerance. If this value is lower or equal to 0 cuGraph will use the default value, which is 100. tol : float, optional (default=1e-05) Set the tolerance the approximation, this parameter should be a small magnitude value. The lower the tolerance the better the approximation. If this value is 0.0f, cuGraph will use the default value which is 1.0E-5. Setting too small a tolerance can lead to non-convergence due to numerical roundoff. Usually values between 0.01 and 0.00001 are acceptable. nstart : dictionary, optional (default=None) dictionary containing the initial guess vertex and value for pagerank. Will be converted to a Dataframe before calling the cugraph algorithm nstart['vertex'] : cudf.Series Subset of vertices of graph for initial guess for pagerank values nstart['values'] : cudf.Series Pagerank values for vertices weight: str, optional (default=None) This parameter is here for NetworkX compatibility and not yet supported in this algorithm dangling : dict, optional (default=None) This parameter is here for NetworkX compatibility and ignored Returns ------- PageRank : dictionary A dictionary of nodes with the PageRank as value """ local_pers = None local_nstart = None if personalization is not None: local_pers = create_cudf_from_dict(personalization) if nstart is not None: local_nstart = create_cudf_from_dict(nstart) return cugraph.pagerank( G, alpha=alpha, personalization=local_pers, max_iter=max_iter, tol=tol, nstart=local_nstart, weight=weight, dangling=dangling, )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/structure/bicliques.py
# Copyright (c) 2019-2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # Import needed libraries import cudf import numpy as np from collections import OrderedDict def EXPERIMENTAL__find_bicliques( df, k, offset=0, max_iter=-1, support=1.0, min_features=1, min_machines=10 ): """ Find the top k maximal bicliques Parameters ---------- df : cudf:DataFrame A dataframe containing the bipartite graph edge list Columns must be called 'src', 'dst', and 'flag' k : int The max number of bicliques to return -1 mean all offset : int Returns ------- B : cudf.DataFrame A dataframe containing the list of machine and features. This is not the full edge list to save space. Since it is a biclique, it is ease to recreate the edges B['id'] - a cluster ID (this is a one up number - up to k) B['vert'] - the vertex ID B['type'] - 0 == machine, 1 == feature S : cudf.DataFrame A dataframe of statistics on the returned info. This dataframe is (relatively small) of size k. S['id'] - the cluster ID S['total'] - total vertex count S['machines'] - number of machine nodes S['features'] - number of feature vertices S['bad_ration'] - the ratio of bad machine / total machines """ # must be factor of 10 PART_SIZE = int(1000) x = [col for col in df.columns] if "src" not in x: raise NameError("src column not found") if "dst" not in x: raise NameError("dst column not found") if "flag" not in x: raise NameError("flag column not found") if support > 1.0 or support < 0.1: raise NameError("support must be between 0.1 and 1.0") # this removes a prep step that offset the values for CUDA process if offset > 0: df["dst"] = df["dst"] - offset # break the data into chunks to improve join/search performance src_by_dst, num_parts = _partition_data_by_feature(df, PART_SIZE) # Get a list of all the dst (features) sorted by degree f_list = _count_features(df, True) # create a dataframe for the answers bicliques = cudf.DataFrame() stats = cudf.DataFrame() # create a dataframe to help prevent duplication of work machine_old = cudf.DataFrame() # create a dataframe for stats stats = cudf.DataFrame() answer_id = 0 iter_max = len(f_list) if max_iter != -1: iter_max = max_iter # Loop over all the features (dst) or until K is reached for i in range(iter_max): # pop the next feature to process feature = f_list["dst"][i] degree = f_list["count"][i] # compute the index to this item (which dataframe chunk is in) idx = int(feature / PART_SIZE) # get all machines that have this feature machines = get_src_from_dst(src_by_dst[idx], feature) # if this set of machines is the same as the last, skip this feature if not is_same_as_last(machine_old, machines): # now from those machines, hop out to the list of all the features feature_list = get_all_feature(src_by_dst, machines, num_parts) # summarize occurances ic = _count_features(feature_list, True) goal = int(degree * support) # NOQA # only get dst nodes with the same degree c = ic.query("count >= @goal") # need more than X feature to make a biclique if len(c) > min_features: if len(machines) >= min_machines: bicliques, stats = update_results( machines, c, answer_id, bicliques, stats ) answer_id = answer_id + 1 # end - if same machine_old = machines if k > -1: if answer_id == k: break # end for loop # All done, reset data if offset > 0: df["dst"] = df["dst"] + offset return bicliques, stats def _partition_data_by_feature(_df, PART_SIZE): # compute the number of sets m = int((_df["dst"].max() / PART_SIZE) + 1) _ui = [None] * (m + 1) # Partition the data into a number of smaller DataFrame s = 0 e = s + PART_SIZE for i in range(m): _ui[i] = _df.query("dst >= @s and dst < @e") s = e e = e + PART_SIZE return _ui, m def _count_features(_gdf, sort=True): aggs = OrderedDict() aggs["dst"] = "count" c = _gdf.groupby(["dst"], as_index=False).agg(aggs) c = c.rename(columns={"count_dst": "count"}, copy=False) if sort: c = c.sort_values(by="count", ascending=False) return c # get all src vertices for a given dst def get_src_from_dst(_gdf, id): _src_list = _gdf.query("dst == @id") _src_list.drop("dst", inplace=True) return _src_list def is_same_as_last(_old, _new): status = False if len(_old) == len(_new): m = _old.merge(_new, on="src", how="left") if m["src"].null_count == 0: status = True return status # get all the items used by the specified users def get_all_feature(_gdf, src_list_df, N): c = [None] * N for i in range(N): c[i] = src_list_df.merge(_gdf[i], on="src", how="inner") return cudf.concat(c) def update_results(m, f, key, b, s): """ Input * m = machines * f = features * key = cluster ID * b = biclique answer * s = stats answer Returns ------- B : cudf.DataFrame A dataframe containing the list of machine and features. This is not the full edge list to save space. Since it is a biclique, it is ease to recreate the edges B['id'] - a cluster ID (this is a one up number - up to k) B['vert'] - the vertex ID B['type'] - 0 == machine, 1 == feature S : cudf.DataFrame A Pandas dataframe of statistics on the returned info. This dataframe is (relatively small) of size k. S['id'] - the cluster ID S['total'] - total vertex count S['machines'] - number of machine nodes S['features'] - number of feature vertices S['bad_ratio'] - the ratio of bad machine / total machines """ B = cudf.DataFrame() S = cudf.DataFrame() m_df = cudf.DataFrame() m_df["vert"] = m["src"] m_df["id"] = int(key) m_df["type"] = int(0) f_df = cudf.DataFrame() f_df["vert"] = f["dst"].astype(np.int32) f_df["id"] = int(key) f_df["type"] = int(1) if len(b) == 0: B = cudf.concat([m_df, f_df]) else: B = cudf.concat([b, m_df, f_df]) # now update the stats num_m = len(m_df) num_f = len(f_df) total = num_m + num_f num_bad = len(m.query("flag == 1")) ratio = num_bad / total # now stats s_tmp = cudf.DataFrame() s_tmp["id"] = key s_tmp["total"] = total s_tmp["machines"] = num_m s_tmp["features"] = num_f s_tmp["bad_ratio"] = ratio if len(s) == 0: S = s_tmp else: S = cudf.concat([s, s_tmp]) del m_df del f_df return B, S
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental
rapidsai_public_repos/cugraph/python/cugraph/cugraph/experimental/structure/__init__.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/sampling/CMakeLists.txt
# ============================================================================= # Copyright (c) 2022, NVIDIA CORPORATION. # # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except # in compliance with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software distributed under the License # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express # or implied. See the License for the specific language governing permissions and limitations under # the License. # ============================================================================= set(cython_sources random_walks_wrapper.pyx) set(linked_libraries cugraph::cugraph) rapids_cython_create_modules( CXX SOURCE_FILES "${cython_sources}" LINKED_LIBRARIES "${linked_libraries}" MODULE_PREFIX sampling_ ASSOCIATED_TARGETS cugraph )
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/sampling/random_walks.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import cudf import cupy as cp from pylibcugraph import ResourceHandle from pylibcugraph import ( uniform_random_walks as pylibcugraph_uniform_random_walks, ) from cugraph.utilities import ensure_cugraph_obj_for_nx from cugraph.structure import Graph import warnings from cugraph.utilities.utils import import_optional from typing import Union, Tuple # FIXME: the networkx.Graph type used in type annotations is specified # using a string literal to avoid depending on and importing networkx. # Instead, networkx is imported optionally, which may cause a problem # for a type checker if run in an environment where networkx is not installed. networkx = import_optional("networkx") def uniform_random_walks( G: Graph, start_vertices: Union[int, list, cudf.Series, cudf.DataFrame] = None, max_depth: int = None, ) -> Tuple[cp.ndarray, cp.ndarray, int]: return pylibcugraph_uniform_random_walks( resource_handle=ResourceHandle(), input_graph=G._plc_graph, start_vertices=start_vertices, max_length=max_depth, ) def random_walks( G: Union[Graph, "networkx.Graph"], random_walks_type: str = "uniform", start_vertices: Union[int, list, cudf.Series, cudf.DataFrame] = None, max_depth: int = None, use_padding: bool = False, legacy_result_type: bool = True, ) -> Tuple[cudf.Series, cudf.Series, Union[None, int, cudf.Series]]: """ Compute random walks for each nodes in 'start_vertices' and returns either a padded or a coalesced result. For the padded case, vertices with no outgoing edges will be padded with -1. When 'use_padding' is 'False', 'random_walks' returns a coalesced result which is a compressed version of the padded one. In the padded form, sources with no out_going edges are padded with -1s in the 'vertex_paths' array and their corresponding edges('edge_weight_paths') with 0.0s (when 'legacy_result_type' is 'True'). If 'legacy_result_type' is 'False', 'random_walks' returns padded results (vertex_paths, edge_weight_paths) but instead of 'sizes = None', returns the 'max_path_lengths'. When 'legacy_result_type' is 'False', the arhument 'use_padding' is ignored. parameters ---------- G : cuGraph.Graph or networkx.Graph The graph can be either directed or undirected. random_walks_type : str, optional (default='uniform') Type of random walks: 'uniform', 'biased', 'node2vec'. Only 'uniform' random walks is currently supported start_vertices : int or list or cudf.Series or cudf.DataFrame A single node or a list or a cudf.Series of nodes from which to run the random walks. In case of multi-column vertices it should be a cudf.DataFrame max_depth : int The maximum depth of the random walks When 'legacy_result_type' is set to False, 'max_depth' is relative to the number of edges otherwised, it is relative to the number of vertices. use_padding : bool, optional (default=False) If True, padded paths are returned else coalesced paths are returned. legacy_result_type : bool, optional (default=True) If True, will return a tuple of vertex_paths, edge_weight_paths and sizes. If False, will return a tuple of vertex_paths, vertex_paths and max_path_length Returns ------- vertex_paths : cudf.Series or cudf.DataFrame Series containing the vertices of edges/paths in the random walk. edge_weight_paths: cudf.Series Series containing the edge weights of edges represented by the returned vertex_paths and sizes: None or cudf.Series The path sizes in case of 'coalesced' paths or None if 'padded'. or max_path_length : int The maximum path length if 'legacy_result_type' is 'False' Examples -------- >>> from cugraph.datasets import karate >>> M = karate.get_edgelist(download=True) >>> G = karate.get_graph() >>> start_vertices = G.nodes()[:4] >>> _, _, _ = cugraph.random_walks(G, "uniform", start_vertices, 3) """ if legacy_result_type: warning_msg = ( "Coalesced path results, returned when setting legacy_result_type=True, " "is deprecated and will no longer be supported in the next releases. " "only padded paths will be returned instead" ) warnings.warn(warning_msg, PendingDeprecationWarning) if max_depth is None: raise TypeError("must specify a 'max_depth'") # FIXME: supporting Nx types should mean having a return type that better # matches Nx expectations (eg. data on the CPU, possibly using a different # data struct like a dictionary, etc.). The 2nd value is ignored here, # which is typically named isNx and used to convert the return type. # Consider a different return type if Nx types are passed in. G, _ = ensure_cugraph_obj_for_nx(G) if isinstance(start_vertices, int): start_vertices = [start_vertices] if isinstance(start_vertices, list): # Ensure the 'start_vertices' have the same dtype as the edge list. # Failing to do that may produce erroneous results. vertex_dtype = G.edgelist.edgelist_df.dtypes[0] start_vertices = cudf.Series(start_vertices, dtype=vertex_dtype) if G.renumbered is True: if isinstance(start_vertices, cudf.DataFrame): start_vertices = G.lookup_internal_vertex_id( start_vertices, start_vertices.columns ) else: start_vertices = G.lookup_internal_vertex_id(start_vertices) if random_walks_type == "uniform": vertex_paths, edge_wgt_paths, max_path_length = uniform_random_walks( G, start_vertices, max_depth ) else: raise ValueError("Only 'uniform' random walks is currently supported") vertex_paths = cudf.Series(vertex_paths) if G.renumbered: df_ = cudf.DataFrame() df_["vertex_paths"] = vertex_paths df_ = G.unrenumber(df_, "vertex_paths", preserve_order=True) vertex_paths = cudf.Series(df_["vertex_paths"]).fillna(-1) edge_wgt_paths = cudf.Series(edge_wgt_paths) # The PLC uniform random walks returns an extra vertex along with an extra # edge per path. In fact, the max depth is relative to the number of vertices # for the legacy implementation and edges for the PLC implementation if legacy_result_type: warning_msg = ( "The 'max_depth' is relative to the number of vertices and will be " "deprecated in the next release. For non legacy result type, it is " "relative to the number of edges which will only be supported." ) warnings.warn(warning_msg, PendingDeprecationWarning) # Drop the last vertex and and edge weight from each vertex and edge weight # paths. vertex_paths = vertex_paths.drop( index=vertex_paths[max_depth :: max_depth + 1].index ).reset_index(drop=True) edge_wgt_paths = edge_wgt_paths.drop( index=edge_wgt_paths[max_depth - 1 :: max_depth].index ).reset_index(drop=True) if use_padding: sizes = None # FIXME: Is it necessary to slice it with 'edge_wgt_paths_sz'? return vertex_paths, edge_wgt_paths, sizes # If 'use_padding' is False, compute the sizes of the unpadded results sizes = ( vertex_paths.apply(lambda x: 1 if x != -1 else 0) .groupby(vertex_paths.index // max_depth, sort=True) .sum() .reset_index(drop=True) ) # Drop the -1 values which are representative of no outgoing edges vertex_paths = vertex_paths.pipe(lambda x: x[x != -1]).reset_index(drop=True) # Drop the 0.0 values which are representative of no edges. edge_wgt_paths = edge_wgt_paths.pipe(lambda x: x[x != 0.0]).reset_index( drop=True ) return vertex_paths, edge_wgt_paths, sizes else: return ( vertex_paths, edge_wgt_paths, max_path_length, ) def rw_path( num_paths: int, sizes: cudf.Series ) -> Tuple[cudf.Series, cudf.Series, cudf.Series]: """ Retrieve more information on the obtained paths in case use_padding is False. parameters ---------- num_paths: int Number of paths in the random walk output. sizes: cudf.Series Path size returned in random walk output. Returns ------- path_data : cudf.DataFrame Dataframe containing vetex path offsets, edge weight offsets and edge weight sizes for each path. """ vertex_offsets = cudf.Series(0, dtype=sizes.dtype) vertex_offsets = cudf.concat( [vertex_offsets, sizes.cumsum()[:-1]], ignore_index=True ) weight_sizes = sizes - 1 weight_offsets = cudf.Series(0, dtype=sizes.dtype) num_edges = vertex_offsets.diff()[1:] - 1 weight_offsets = cudf.concat( [weight_offsets, num_edges.cumsum()], ignore_index=True ) # FIXME: CUDF bug. concatenating two series of type int32 but get a CUDF of # type 'int64' have to cast the results weight_offsets = weight_offsets.astype(sizes.dtype) path_data = cudf.DataFrame() path_data["vertex_offsets"] = vertex_offsets path_data["weight_sizes"] = weight_sizes path_data["weight_offsets"] = weight_offsets return path_data[:num_paths]
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/sampling/sampling_utilities.py
# Copyright (c) 2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import cupy import cudf import warnings def sampling_results_from_cupy_array_dict( cupy_array_dict, weight_t, num_hops, with_edge_properties=False, return_offsets=False, renumber=False, use_legacy_names=True, include_hop_column=True, ): """ Creates a cudf DataFrame from cupy arrays from pylibcugraph wrapper """ results_df = cudf.DataFrame() if use_legacy_names: major_col_name = "sources" minor_col_name = "destinations" warning_msg = ( "The legacy column names (sources, destinations)" " will no longer be supported for uniform_neighbor_sample" " in release 23.12. The use_legacy_names=False option will" " become the only option, and (majors, minors) will be the" " only supported column names." ) warnings.warn(warning_msg, FutureWarning) else: major_col_name = "majors" minor_col_name = "minors" if with_edge_properties: majors = cupy_array_dict["majors"] if majors is not None: results_df["majors"] = majors results_df_cols = [ "minors", "weight", "edge_id", "edge_type", ] for col in results_df_cols: array = cupy_array_dict[col] # The length of each of these arrays should be the same results_df[col] = array results_df.rename( columns={"majors": major_col_name, "minors": minor_col_name}, inplace=True ) label_hop_offsets = cupy_array_dict["label_hop_offsets"] batch_ids = cupy_array_dict["batch_id"] if renumber: renumber_df = cudf.DataFrame( { "map": cupy_array_dict["renumber_map"], } ) if not return_offsets: if len(batch_ids) > 0: batch_ids_r = cudf.Series(batch_ids).repeat( cupy.diff(cupy_array_dict["renumber_map_offsets"]) ) batch_ids_r.reset_index(drop=True, inplace=True) renumber_df["batch_id"] = batch_ids_r else: renumber_df["batch_id"] = None if return_offsets: batches_series = cudf.Series( batch_ids, name="batch_id", ) if include_hop_column: # TODO remove this logic in release 23.12 offsets_df = cudf.Series( label_hop_offsets[cupy.arange(len(batch_ids) + 1) * num_hops], name="offsets", ).to_frame() else: offsets_df = cudf.Series( label_hop_offsets, name="offsets", ).to_frame() if len(batches_series) > len(offsets_df): # this is extremely rare so the inefficiency is ok offsets_df = offsets_df.join(batches_series, how="outer").sort_index() else: offsets_df["batch_id"] = batches_series if renumber: renumber_offset_series = cudf.Series( cupy_array_dict["renumber_map_offsets"], name="renumber_map_offsets" ) if len(renumber_offset_series) > len(offsets_df): # this is extremely rare so the inefficiency is ok offsets_df = offsets_df.join( renumber_offset_series, how="outer" ).sort_index() else: offsets_df["renumber_map_offsets"] = renumber_offset_series else: if len(batch_ids) > 0: batch_ids_r = cudf.Series(cupy.repeat(batch_ids, num_hops)) batch_ids_r = cudf.Series(batch_ids_r).repeat( cupy.diff(label_hop_offsets) ) batch_ids_r.reset_index(drop=True, inplace=True) results_df["batch_id"] = batch_ids_r else: results_df["batch_id"] = None # TODO remove this logic in release 23.12, hops will always returned as offsets if include_hop_column: if len(batch_ids) > 0: hop_ids_r = cudf.Series(cupy.arange(num_hops)) hop_ids_r = cudf.concat([hop_ids_r] * len(batch_ids), ignore_index=True) # generate the hop column hop_ids_r = ( cudf.Series(hop_ids_r, name="hop_id") .repeat(cupy.diff(label_hop_offsets)) .reset_index(drop=True) ) else: hop_ids_r = cudf.Series(name="hop_id", dtype="int32") results_df = results_df.join(hop_ids_r, how="outer").sort_index() if major_col_name not in results_df: if use_legacy_names: raise ValueError("Can't use legacy names with major offsets") major_offsets_series = cudf.Series( cupy_array_dict["major_offsets"], name="major_offsets" ) if len(major_offsets_series) > len(results_df): # this is extremely rare so the inefficiency is ok results_df = results_df.join( major_offsets_series, how="outer" ).sort_index() else: results_df["major_offsets"] = major_offsets_series else: # TODO this is deprecated, remove it in 23.12 results_df[major_col_name] = cupy_array_dict["sources"] results_df[minor_col_name] = cupy_array_dict["destinations"] indices = cupy_array_dict["indices"] if indices is None: results_df["indices"] = None else: results_df["indices"] = indices if weight_t == "int32": results_df["indices"] = indices.astype("int32") elif weight_t == "int64": results_df["indices"] = indices.astype("int64") else: results_df["indices"] = indices if return_offsets: if renumber: return results_df, offsets_df, renumber_df else: return results_df, offsets_df if renumber: return results_df, renumber_df return (results_df,)
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/sampling/uniform_neighbor_sample.py
# Copyright (c) 2022-2023, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from __future__ import annotations from pylibcugraph import ResourceHandle from pylibcugraph import uniform_neighbor_sample as pylibcugraph_uniform_neighbor_sample from cugraph.sampling.sampling_utilities import sampling_results_from_cupy_array_dict import numpy import cudf import cupy as cp import warnings from typing import Union, Tuple, Sequence, List from typing import TYPE_CHECKING if TYPE_CHECKING: from cugraph import Graph start_col_name = "_START_" batch_col_name = "_BATCH_" # FIXME: Move this function to the utility module so that it can be # shared by other algos def ensure_valid_dtype(input_graph, start_list): vertex_dtype = input_graph.edgelist.edgelist_df.dtypes[0] if isinstance(start_list, cudf.Series): start_list_dtypes = start_list.dtype else: start_list_dtypes = start_list.dtypes[0] if start_list_dtypes != vertex_dtype: warning_msg = ( "Uniform neighbor sample requires 'start_list' to match the graph's " f"'vertex' type. input graph's vertex type is: {vertex_dtype} and got " f"'start_list' of type: {start_list_dtypes}." ) warnings.warn(warning_msg, UserWarning) start_list = start_list.astype(vertex_dtype) return start_list def uniform_neighbor_sample( G: Graph, start_list: Sequence, fanout_vals: List[int], *, with_replacement: bool = True, with_edge_properties: bool = False, # deprecated with_batch_ids: bool = False, random_state: int = None, return_offsets: bool = False, return_hops: bool = True, include_hop_column: bool = True, # deprecated prior_sources_behavior: str = None, deduplicate_sources: bool = False, renumber: bool = False, use_legacy_names: bool = True, # deprecated compress_per_hop: bool = False, compression: str = "COO", ) -> Union[cudf.DataFrame, Tuple[cudf.DataFrame, cudf.DataFrame]]: """ Does neighborhood sampling, which samples nodes from a graph based on the current node's neighbors, with a corresponding fanout value at each hop. Parameters ---------- G : cugraph.Graph cuGraph graph, which contains connectivity information as dask cudf edge list dataframe start_list : list or cudf.Series (int32) a list of starting vertices for sampling fanout_vals : list (int32) List of branching out (fan-out) degrees per starting vertex for each hop level. with_replacement: bool, optional (default=True) Flag to specify if the random sampling is done with replacement with_edge_properties: bool, optional (default=False) Deprecated. Flag to specify whether to return edge properties (weight, edge id, edge type, batch id, hop id) with the sampled edges. with_batch_ids: bool, optional (default=False) Flag to specify whether batch ids are present in the start_list Assumes they are the last column in the start_list dataframe random_state: int, optional Random seed to use when making sampling calls. return_offsets: bool, optional (default=False) Whether to return the sampling results with batch ids included as one dataframe, or to instead return two dataframes, one with sampling results and one with batch ids and their start offsets. return_hops: bool, optional (default=True) Whether to return the sampling results with hop ids corresponding to the hop where the edge appeared. Defaults to True. include_hop_column: bool, optional (default=True) Deprecated. Defaults to True. If True, will include the hop column even if return_offsets is True. This option will be removed in release 23.12. prior_sources_behavior: str, optional (default=None) Options are "carryover", and "exclude". Default will leave the source list as-is. Carryover will carry over sources from previous hops to the current hop. Exclude will exclude sources from previous hops from reappearing as sources in future hops. deduplicate_sources: bool, optional (default=False) Whether to first deduplicate the list of possible sources from the previous destinations before performing next hop. renumber: bool, optional (default=False) Whether to renumber on a per-batch basis. If True, will return the renumber map and renumber map offsets as an additional dataframe. use_legacy_names: bool, optional (default=True) Whether to use the legacy column names (sources, destinations). If True, will use "sources" and "destinations" as the column names. If False, will use "majors" and "minors" as the column names. Deprecated. Will be removed in release 23.12 in favor of always using the new names "majors" and "minors". compress_per_hop: bool, optional (default=False) Whether to compress globally (default), or to produce a separate compressed edgelist per hop. compression: str, optional (default=COO) Sets the compression type for the output minibatches. Valid options are COO (default), CSR, CSC, DCSR, and DCSC. Returns ------- result : cudf.DataFrame or Tuple[cudf.DataFrame, cudf.DataFrame] GPU data frame containing multiple cudf.Series If with_edge_properties=False: df['sources']: cudf.Series Contains the source vertices from the sampling result df['destinations']: cudf.Series Contains the destination vertices from the sampling result df['indices']: cudf.Series Contains the indices (edge weights) from the sampling result for path reconstruction If with_edge_properties=True: If return_offsets=False: df['sources']: cudf.Series Contains the source vertices from the sampling result df['destinations']: cudf.Series Contains the destination vertices from the sampling result df['edge_weight']: cudf.Series Contains the edge weights from the sampling result df['edge_id']: cudf.Series Contains the edge ids from the sampling result df['edge_type']: cudf.Series Contains the edge types from the sampling result df['batch_id']: cudf.Series Contains the batch ids from the sampling result df['hop_id']: cudf.Series Contains the hop ids from the sampling result If renumber=True: (adds the following dataframe) renumber_df['map']: cudf.Series Contains the renumber maps for each batch renumber_df['offsets']: cudf.Series Contains the batch offsets for the renumber maps If return_offsets=True: df['sources']: cudf.Series Contains the source vertices from the sampling result df['destinations']: cudf.Series Contains the destination vertices from the sampling result df['edge_weight']: cudf.Series Contains the edge weights from the sampling result df['edge_id']: cudf.Series Contains the edge ids from the sampling result df['edge_type']: cudf.Series Contains the edge types from the sampling result df['hop_id']: cudf.Series Contains the hop ids from the sampling result offsets_df['batch_id']: cudf.Series Contains the batch ids from the sampling result offsets_df['offsets']: cudf.Series Contains the offsets of each batch in the sampling result If renumber=True: (adds the following dataframe) renumber_df['map']: cudf.Series Contains the renumber maps for each batch renumber_df['offsets']: cudf.Series Contains the batch offsets for the renumber maps """ if use_legacy_names: major_col_name = "sources" minor_col_name = "destinations" warning_msg = ( "The legacy column names (sources, destinations)" " will no longer be supported for uniform_neighbor_sample" " in release 23.12. The use_legacy_names=False option will" " become the only option, and (majors, minors) will be the" " only supported column names." ) warnings.warn(warning_msg, FutureWarning) else: major_col_name = "majors" minor_col_name = "minors" if compression not in ["COO", "CSR", "CSC", "DCSR", "DCSC"]: raise ValueError("compression must be one of COO, CSR, CSC, DCSR, or DCSC") if ( (compression != "COO") and (not compress_per_hop) and prior_sources_behavior != "exclude" ): raise ValueError( "hop-agnostic compression is only supported with" " the exclude prior sources behavior due to limitations " "of the libcugraph C++ API" ) if compress_per_hop and prior_sources_behavior != "carryover": raise ValueError( "Compressing the edgelist per hop is only supported " "with the carryover prior sources behavior due to limitations" " of the libcugraph C++ API" ) if include_hop_column: warning_msg = ( "The include_hop_column flag is deprecated and will be" " removed in the next release in favor of always " "excluding the hop column when return_offsets is True" ) warnings.warn(warning_msg, FutureWarning) if compression != "COO": raise ValueError( "Including the hop id column is only supported with COO compression." ) if with_edge_properties: warning_msg = ( "The with_edge_properties flag is deprecated" " and will be removed in the next release in favor" " of returning all properties in the graph" ) warnings.warn(warning_msg, FutureWarning) if isinstance(start_list, int): start_list = [start_list] if isinstance(start_list, list): start_list = cudf.Series( start_list, dtype=G.edgelist.edgelist_df[G.srcCol].dtype ) if with_edge_properties and not with_batch_ids: if isinstance(start_list, cudf.Series): start_list = start_list.reset_index(drop=True).to_frame() start_list[batch_col_name] = cudf.Series( cp.zeros(len(start_list), dtype="int32") ) # fanout_vals must be passed to pylibcugraph as a host array if isinstance(fanout_vals, numpy.ndarray): fanout_vals = fanout_vals.astype("int32") elif isinstance(fanout_vals, list): fanout_vals = numpy.asarray(fanout_vals, dtype="int32") elif isinstance(fanout_vals, cp.ndarray): fanout_vals = fanout_vals.get().astype("int32") elif isinstance(fanout_vals, cudf.Series): fanout_vals = fanout_vals.values_host.astype("int32") else: raise TypeError("fanout_vals must be a sequence, " f"got: {type(fanout_vals)}") if "weights" in G.edgelist.edgelist_df: weight_t = G.edgelist.edgelist_df["weights"].dtype else: weight_t = "float32" start_list = ensure_valid_dtype(G, start_list) if isinstance(start_list, cudf.Series): start_list = start_list.rename(start_col_name) start_list = start_list.to_frame() if G.renumbered: start_list = G.lookup_internal_vertex_id(start_list, start_col_name) else: columns = start_list.columns if with_batch_ids: if G.renumbered: start_list = G.lookup_internal_vertex_id(start_list, columns[:-1]) start_list = start_list.rename( columns={columns[0]: start_col_name, columns[-1]: batch_col_name} ) else: if G.renumbered: start_list = G.lookup_internal_vertex_id(start_list, columns) start_list = start_list.rename(columns={columns[0]: start_col_name}) sampling_result_array_dict = pylibcugraph_uniform_neighbor_sample( resource_handle=ResourceHandle(), input_graph=G._plc_graph, start_list=start_list[start_col_name], batch_id_list=start_list[batch_col_name] if batch_col_name in start_list else None, h_fan_out=fanout_vals, with_replacement=with_replacement, do_expensive_check=False, with_edge_properties=with_edge_properties, random_state=random_state, prior_sources_behavior=prior_sources_behavior, deduplicate_sources=deduplicate_sources, return_hops=return_hops, renumber=renumber, compression=compression, compress_per_hop=compress_per_hop, return_dict=True, ) dfs = sampling_results_from_cupy_array_dict( sampling_result_array_dict, weight_t, len(fanout_vals), with_edge_properties=with_edge_properties, return_offsets=return_offsets, renumber=renumber, use_legacy_names=use_legacy_names, include_hop_column=include_hop_column, ) if G.renumbered and not renumber: dfs[0] = G.unrenumber(dfs[0], major_col_name, preserve_order=True) dfs[0] = G.unrenumber(dfs[0], minor_col_name, preserve_order=True) if len(dfs) > 1: return dfs return dfs[0]
0
rapidsai_public_repos/cugraph/python/cugraph/cugraph
rapidsai_public_repos/cugraph/python/cugraph/cugraph/sampling/__init__.py
# Copyright (c) 2021-2022, NVIDIA CORPORATION. # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from cugraph.sampling.random_walks import random_walks, rw_path from cugraph.sampling.node2vec import node2vec from cugraph.sampling.uniform_neighbor_sample import uniform_neighbor_sample
0