Refactor the python function symmetrizing the edgelist (#4649)

This PR updates the python API to symmetrize the edge list through the CAPI for PLC algorithms. This PR also deprecates the legacy python function symmetrizing the edge list closes #4588
rapidsai · Oct 3, 2024 · 5fad435 · 5fad435
1 parent e6be367
commit 5fad435
Show file tree

Hide file tree

Showing 21 changed files with 477 additions and 71 deletions.
diff --git a/cpp/include/cugraph_c/graph.h b/cpp/include/cugraph_c/graph.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021-2023, NVIDIA CORPORATION.
+ * Copyright (c) 2021-2024, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -105,6 +105,8 @@ cugraph_error_code_t cugraph_sg_graph_create(
  weights,
  *    or take the maximum weight), the caller should remove specific edges themselves and not rely
  *    on this flag.
+ * @param [in] symmetrize      If true, symmetrize the edgelist. The symmetrization of edges
+ * with edge_ids and/or edge_type_ids is currently not supported.
  * @param [in]  do_expensive_check    If true, do expensive checks to validate the input data
  *    is consistent with software assumptions.  If false bypass these checks.
  * @param [out] graph          A pointer to the graph object
@@ -126,6 +128,7 @@ cugraph_error_code_t cugraph_graph_create_sg(
   bool_t renumber,
   bool_t drop_self_loops,
   bool_t drop_multi_edges,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error);
@@ -150,6 +153,8 @@ cugraph_error_code_t cugraph_graph_create_sg(
  *    If false, do not renumber.  Renumbering enables some significant optimizations within
  *    the graph primitives library, so it is strongly encouraged.  Renumbering is required if
  *    the vertices are not sequential integer values from 0 to num_vertices.
+ * @param [in]  symmetrize     If true, symmetrize the edgelist. The symmetrization of edges
+ * with edge_ids and/or edge_type_ids is currently not supported.
  * @param [in]  do_expensive_check    If true, do expensive checks to validate the input data
  *    is consistent with software assumptions.  If false bypass these checks.
  * @param [out] graph          A pointer to the graph object
@@ -168,6 +173,7 @@ cugraph_error_code_t cugraph_sg_graph_create_from_csr(
   const cugraph_type_erased_device_array_view_t* edge_type_ids,
   bool_t store_transposed,
   bool_t renumber,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error);
@@ -190,6 +196,8 @@ cugraph_error_code_t cugraph_sg_graph_create_from_csr(
  *    If false, do not renumber.  Renumbering enables some significant optimizations within
  *    the graph primitives library, so it is strongly encouraged.  Renumbering is required if
  *    the vertices are not sequential integer values from 0 to num_vertices.
+ * @param [in]  symmetrize     If true, symmetrize the edgelist. The symmetrization of edges
+ * with edge_ids and/or edge_type_ids is currently not supported.
  * @param [in]  do_expensive_check    If true, do expensive checks to validate the input data
  *    is consistent with software assumptions.  If false bypass these checks.
  * @param [out] graph          A pointer to the graph object
@@ -208,6 +216,7 @@ cugraph_error_code_t cugraph_graph_create_sg_from_csr(
   const cugraph_type_erased_device_array_view_t* edge_type_ids,
   bool_t store_transposed,
   bool_t renumber,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error);
@@ -289,6 +298,8 @@ cugraph_error_code_t cugraph_mg_graph_create(
  *    Note that setting this flag will arbitrarily select one instance of a multi edge to be the
  *    edge that survives.  If the edges have properties that should be honored (e.g. sum the
  * weights, or take the maximum weight), the caller should do that on not rely on this flag.
+ * @param [in]  symmetrize      If true, symmetrize the edgelist. The symmetrization of edges
+ * with edge_ids and/or edge_type_ids is currently not supported.
  * @param [in]  do_expensive_check  If true, do expensive checks to validate the input data
  *    is consistent with software assumptions.  If false bypass these checks.
  * @param [out] graph           A pointer to the graph object
@@ -309,6 +320,7 @@ cugraph_error_code_t cugraph_graph_create_mg(
   size_t num_arrays,
   bool_t drop_self_loops,
   bool_t drop_multi_edges,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error);

diff --git a/cpp/src/c_api/graph_mg.cpp b/cpp/src/c_api/graph_mg.cpp
@@ -71,6 +71,7 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
   bool_t renumber_;
   bool_t drop_self_loops_;
   bool_t drop_multi_edges_;
+  bool_t symmetrize_;
   bool_t do_expensive_check_;
   cugraph::c_api::cugraph_graph_t* result_{};
 
@@ -91,6 +92,7 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
     bool_t renumber,
     bool_t drop_self_loops,
     bool_t drop_multi_edges,
+    bool_t symmetrize,
     bool_t do_expensive_check)
     : abstract_functor(),
       properties_(properties),
@@ -109,6 +111,7 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
       renumber_(renumber),
       drop_self_loops_(drop_self_loops),
       drop_multi_edges_(drop_multi_edges),
+      symmetrize_(symmetrize),
       do_expensive_check_(do_expensive_check)
   {
   }
@@ -224,6 +227,22 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
                                         : false);
       }
 
+      if (symmetrize_) {
+        if (edgelist_edge_ids || edgelist_edge_types) {
+          // Currently doesn't support the symmetrization of edgelist with edge_ids and edge_types
+          unsupported();
+        }
+
+        // Symmetrize the edgelist
+        std::tie(edgelist_srcs, edgelist_dsts, edgelist_weights) =
+          cugraph::symmetrize_edgelist<vertex_t, weight_t, store_transposed, multi_gpu>(
+            handle_,
+            std::move(edgelist_srcs),
+            std::move(edgelist_dsts),
+            std::move(edgelist_weights),
+            false);
+      }
+
       std::tie(*graph, new_edge_weights, new_edge_ids, new_edge_types, new_number_map) =
         cugraph::create_graph_from_edgelist<vertex_t,
                                             edge_t,
@@ -290,6 +309,7 @@ extern "C" cugraph_error_code_t cugraph_graph_create_mg(
   size_t num_arrays,
   bool_t drop_self_loops,
   bool_t drop_multi_edges,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error)
@@ -358,6 +378,14 @@ extern "C" cugraph_error_code_t cugraph_graph_create_mg(
       if (weight_type == cugraph_data_type_id_t::NTYPES) weight_type = p_weights[i]->type_;
     }
 
+    if (symmetrize == TRUE) {
+      CAPI_EXPECTS((properties->is_symmetric == TRUE),
+                   CUGRAPH_INVALID_INPUT,
+                   "Invalid input arguments: The graph property must be symmetric if 'symmetrize' "
+                   "is set to True.",
+                   *error);
+    }
+
     CAPI_EXPECTS(p_src[i]->type_ == vertex_type,
                  CUGRAPH_INVALID_INPUT,
                  "Invalid input arguments: all vertex types must match",
@@ -488,6 +516,7 @@ extern "C" cugraph_error_code_t cugraph_graph_create_mg(
                                bool_t::TRUE,
                                drop_self_loops,
                                drop_multi_edges,
+                               symmetrize,
                                do_expensive_check);
 
   try {
@@ -534,6 +563,7 @@ extern "C" cugraph_error_code_t cugraph_mg_graph_create(
                                  1,
                                  FALSE,
                                  FALSE,
+                                 FALSE,
                                  do_expensive_check,
                                  graph,
                                  error);

diff --git a/cpp/src/c_api/graph_sg.cpp b/cpp/src/c_api/graph_sg.cpp
@@ -43,6 +43,7 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
   bool_t renumber_;
   bool_t drop_self_loops_;
   bool_t drop_multi_edges_;
+  bool_t symmetrize_;
   bool_t do_expensive_check_;
   cugraph_data_type_id_t edge_type_;
   cugraph::c_api::cugraph_graph_t* result_{};
@@ -58,6 +59,7 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
                        bool_t renumber,
                        bool_t drop_self_loops,
                        bool_t drop_multi_edges,
+                       bool_t symmetrize,
                        bool_t do_expensive_check,
                        cugraph_data_type_id_t edge_type)
     : abstract_functor(),
@@ -72,6 +74,7 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
       renumber_(renumber),
       drop_self_loops_(drop_self_loops),
       drop_multi_edges_(drop_multi_edges),
+      symmetrize_(symmetrize),
       do_expensive_check_(do_expensive_check),
       edge_type_(edge_type)
   {
@@ -207,6 +210,22 @@ struct create_graph_functor : public cugraph::c_api::abstract_functor {
                                         : false);
       }
 
+      if (symmetrize_) {
+        if (edgelist_edge_ids || edgelist_edge_types) {
+          // Currently doesn't support the symmetrization with edge_ids and edge_types
+          unsupported();
+        }
+
+        // Symmetrize the edgelist
+        std::tie(edgelist_srcs, edgelist_dsts, edgelist_weights) =
+          cugraph::symmetrize_edgelist<vertex_t, weight_t, store_transposed, multi_gpu>(
+            handle_,
+            std::move(edgelist_srcs),
+            std::move(edgelist_dsts),
+            std::move(edgelist_weights),
+            false);
+      }
+
       std::tie(*graph, new_edge_weights, new_edge_ids, new_edge_types, new_number_map) =
         cugraph::create_graph_from_edgelist<vertex_t,
                                             edge_t,
@@ -268,6 +287,7 @@ struct create_graph_csr_functor : public cugraph::c_api::abstract_functor {
   cugraph::c_api::cugraph_type_erased_device_array_view_t const* edge_ids_;
   cugraph::c_api::cugraph_type_erased_device_array_view_t const* edge_type_ids_;
   bool_t renumber_;
+  bool_t symmetrize_;
   bool_t do_expensive_check_;
   cugraph::c_api::cugraph_graph_t* result_{};
 
@@ -280,6 +300,7 @@ struct create_graph_csr_functor : public cugraph::c_api::abstract_functor {
     cugraph::c_api::cugraph_type_erased_device_array_view_t const* edge_ids,
     cugraph::c_api::cugraph_type_erased_device_array_view_t const* edge_type_ids,
     bool_t renumber,
+    bool_t symmetrize,
     bool_t do_expensive_check)
     : abstract_functor(),
       properties_(properties),
@@ -290,6 +311,7 @@ struct create_graph_csr_functor : public cugraph::c_api::abstract_functor {
       edge_ids_(edge_ids),
       edge_type_ids_(edge_type_ids),
       renumber_(renumber),
+      symmetrize_(symmetrize),
       do_expensive_check_(do_expensive_check)
   {
   }
@@ -398,6 +420,22 @@ struct create_graph_csr_functor : public cugraph::c_api::abstract_functor {
         cugraph::graph_view_t<vertex_t, edge_t, store_transposed, multi_gpu>,
         edge_type_id_t>(handle_);
 
+      if (symmetrize_) {
+        if (edgelist_edge_ids || edgelist_edge_types) {
+          // Currently doesn't support the symmetrization with edge_ids and edge_types
+          unsupported();
+        }
+
+        // Symmetrize the edgelist
+        std::tie(edgelist_srcs, edgelist_dsts, edgelist_weights) =
+          cugraph::symmetrize_edgelist<vertex_t, weight_t, store_transposed, multi_gpu>(
+            handle_,
+            std::move(edgelist_srcs),
+            std::move(edgelist_dsts),
+            std::move(edgelist_weights),
+            false);
+      }
+
       std::tie(*graph, new_edge_weights, new_edge_ids, new_edge_types, new_number_map) =
         cugraph::create_graph_from_edgelist<vertex_t,
                                             edge_t,
@@ -518,6 +556,7 @@ extern "C" cugraph_error_code_t cugraph_graph_create_sg(
   bool_t renumber,
   bool_t drop_self_loops,
   bool_t drop_multi_edges,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error)
@@ -542,6 +581,14 @@ extern "C" cugraph_error_code_t cugraph_graph_create_sg(
   auto p_edge_type_ids =
     reinterpret_cast<cugraph::c_api::cugraph_type_erased_device_array_view_t const*>(edge_type_ids);
 
+  if (symmetrize == TRUE) {
+    CAPI_EXPECTS((properties->is_symmetric == TRUE),
+                 CUGRAPH_INVALID_INPUT,
+                 "Invalid input arguments: The graph property must be symmetric if 'symmetrize' is "
+                 "set to True.",
+                 *error);
+  }
+
   CAPI_EXPECTS(p_src->size_ == p_dst->size_,
                CUGRAPH_INVALID_INPUT,
                "Invalid input arguments: src size != dst size.",
@@ -606,6 +653,7 @@ extern "C" cugraph_error_code_t cugraph_graph_create_sg(
                                  renumber,
                                  drop_self_loops,
                                  drop_multi_edges,
+                                 symmetrize,
                                  do_expensive_check,
                                  edge_type);
 
@@ -658,6 +706,7 @@ extern "C" cugraph_error_code_t cugraph_sg_graph_create(
                                  renumber,
                                  FALSE,
                                  FALSE,
+                                 FALSE,
                                  do_expensive_check,
                                  graph,
                                  error);
@@ -673,6 +722,7 @@ cugraph_error_code_t cugraph_graph_create_sg_from_csr(
   const cugraph_type_erased_device_array_view_t* edge_type_ids,
   bool_t store_transposed,
   bool_t renumber,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error)
@@ -707,6 +757,14 @@ cugraph_error_code_t cugraph_graph_create_sg_from_csr(
     weight_type = cugraph_data_type_id_t::FLOAT32;
   }
 
+  if (symmetrize == TRUE) {
+    CAPI_EXPECTS((properties->is_symmetric == TRUE),
+                 CUGRAPH_INVALID_INPUT,
+                 "Invalid input arguments: The graph property must be symmetric if 'symmetrize' is "
+                 "set to True.",
+                 *error);
+  }
+
   CAPI_EXPECTS(
     (edge_type_ids == nullptr && edge_ids == nullptr) ||
       (edge_type_ids != nullptr && edge_ids != nullptr),
@@ -735,6 +793,7 @@ cugraph_error_code_t cugraph_graph_create_sg_from_csr(
                                      p_edge_ids,
                                      p_edge_type_ids,
                                      renumber,
+                                     FALSE,  // symmetrize
                                      do_expensive_check);
 
   try {
@@ -770,6 +829,7 @@ cugraph_error_code_t cugraph_sg_graph_create_from_csr(
   const cugraph_type_erased_device_array_view_t* edge_type_ids,
   bool_t store_transposed,
   bool_t renumber,
+  bool_t symmetrize,
   bool_t do_expensive_check,
   cugraph_graph_t** graph,
   cugraph_error_t** error)
@@ -783,6 +843,7 @@ cugraph_error_code_t cugraph_sg_graph_create_from_csr(
                                           edge_type_ids,
                                           store_transposed,
                                           renumber,
+                                          symmetrize,
                                           do_expensive_check,
                                           graph,
                                           error);

diff --git a/cpp/tests/c_api/create_graph_test.c b/cpp/tests/c_api/create_graph_test.c
@@ -104,6 +104,7 @@ int test_create_sg_graph_simple()
                                      FALSE,
                                      FALSE,
                                      FALSE,
+                                     FALSE,
                                      &graph,
                                      &ret_error);
   TEST_ASSERT(test_ret_value, ret_code == CUGRAPH_SUCCESS, "graph creation failed.");
@@ -213,6 +214,7 @@ int test_create_sg_graph_csr()
                                               FALSE,
                                               FALSE,
                                               FALSE,
+                                              FALSE,
                                               &graph,
                                               &ret_error);
   TEST_ASSERT(test_ret_value, ret_code == CUGRAPH_SUCCESS, "graph creation failed.");
@@ -408,6 +410,7 @@ int test_create_sg_graph_symmetric_error()
                                      FALSE,
                                      FALSE,
                                      FALSE,
+                                     FALSE,
                                      TRUE,
                                      &graph,
                                      &ret_error);
@@ -526,6 +529,7 @@ int test_create_sg_graph_with_isolated_vertices()
                                      FALSE,
                                      FALSE,
                                      FALSE,
+                                     FALSE,
                                      &graph,
                                      &ret_error);
   TEST_ASSERT(test_ret_value, ret_code == CUGRAPH_SUCCESS, "graph creation failed.");
@@ -675,6 +679,7 @@ int test_create_sg_graph_csr_with_isolated()
                                               FALSE,
                                               FALSE,
                                               FALSE,
+                                              FALSE,
                                               &graph,
                                               &ret_error);
   TEST_ASSERT(test_ret_value, ret_code == CUGRAPH_SUCCESS, "graph creation failed.");
@@ -840,6 +845,7 @@ int test_create_sg_graph_with_isolated_vertices_multi_input()
                                      TRUE,
                                      TRUE,
                                      FALSE,
+                                     FALSE,
                                      &graph,
                                      &ret_error);
   TEST_ASSERT(test_ret_value, ret_code == CUGRAPH_SUCCESS, "graph creation failed.");