Skip to content

C++ implementation for unweighted Jaccard/Sorensen/Overlap #2750

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
rapids-bot merged 7 commits into from
Oct 4, 2022
Merged

C++ implementation for unweighted Jaccard/Sorensen/Overlap #2750

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
aef4e83
first cut at similarity implementation, waiting for primitive to be r…
ChuckHastings Sep 27, 2022
ff39a6f
Merge branch 'branch-22.10' into fea_implement_jaccard
ChuckHastings Sep 28, 2022
8b42bab
Merge branch 'branch-22.10' into fea_implement_jaccard
ChuckHastings Sep 29, 2022
13c09bb
completed implementation, still need to fix tests
ChuckHastings Sep 30, 2022
ee09e33
commit to move testing
ChuckHastings Oct 1, 2022
01944b0
debugging
ChuckHastings Oct 2, 2022
f94ff89
finished testing implementation with Seunghwa's fixes for primitives
ChuckHastings Oct 3, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 9 additions & 5 deletions cpp/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -161,9 +161,17 @@ endif()
################################################################################
# - libcugraph library target --------------------------------------------------

# NOTE: The most expensive compilations are listed first
# since ninja will run them in parallel in this order,
# which should give us a better parallel schedule.

set(CUGRAPH_SOURCES
src/detail/utility_wrappers.cu
src/detail/shuffle_wrappers.cu
src/sampling/random_walks_mg.cu
src/community/detail/common_methods_mg.cu
src/community/detail/common_methods_sg.cu
src/detail/utility_wrappers.cu
src/structure/graph_view_mg.cu
Copy link
Contributor

@seunghwak seunghwak Oct 3, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mainly for curiosity, why did you move graph_view_mg.cu to here?

Copy link
Collaborator Author

@ChuckHastings ChuckHastings Oct 3, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ninja tracing showed that it take 217 seconds to compile. I basically moved everything that was slow (by the unscientific method of my eyes looking at the graphic) up to the top to get it started earlier in the parallel schedule.

src/utilities/cython.cu
src/utilities/path_retrieval.cu
src/utilities/graph_bcast.cpp
Expand All @@ -181,8 +189,6 @@ set(CUGRAPH_SOURCES
src/layout/legacy/force_atlas2.cu
src/converters/legacy/COOtoCSR.cu
src/community/legacy/spectral_clustering.cu
src/community/detail/common_methods_sg.cu
src/community/detail/common_methods_mg.cu
src/community/louvain_sg.cu
src/community/louvain_mg.cu
src/community/leiden_sg.cu
Expand All @@ -195,7 +201,6 @@ set(CUGRAPH_SOURCES
src/community/legacy/egonet.cu
src/sampling/random_walks.cu
src/sampling/random_walks_sg.cu
src/sampling/random_walks_mg.cu
src/sampling/detail/sampling_utils_mg.cu
src/sampling/detail/sampling_utils_sg.cu
src/sampling/uniform_neighbor_sampling_mg.cpp
Expand All @@ -215,7 +220,6 @@ set(CUGRAPH_SOURCES
src/structure/graph_sg.cu
src/structure/graph_mg.cu
src/structure/graph_view_sg.cu
src/structure/graph_view_mg.cu
src/structure/coarsen_graph_sg.cu
src/structure/coarsen_graph_mg.cu
src/structure/renumber_edgelist_sg.cu
Expand Down
19 changes: 6 additions & 13 deletions cpp/src/link_prediction/jaccard_impl.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,26 +25,19 @@ namespace detail {

struct jaccard_functor_t {
template <typename weight_t>
weight_t __device__ compute_score(size_t cardinality_a,
size_t cardinality_b,
size_t cardinality_a_intersect_b,
weight_t,
weight_t,
weight_t)
weight_t __device__ compute_score(weight_t cardinality_a,
weight_t cardinality_b,
weight_t cardinality_a_intersect_b) const
{
return static_cast<weight_t>(cardinality_a_intersect_b) /
static_cast<weight_t>(cardinality_a + cardinality_b - cardinality_a_intersect_b);
return cardinality_a_intersect_b / cardinality_a + cardinality_b - cardinality_a_intersect_b;
}
};

struct weighted_jaccard_functor_t {
template <typename weight_t>
weight_t __device__ compute_score(size_t cardinality_a,
size_t cardinality_b,
size_t cardinality_a_intersect_b,
weight_t weight_a,
weight_t __device__ compute_score(weight_t weight_a,
weight_t weight_b,
weight_t min_weight_a_intersect_b)
weight_t min_weight_a_intersect_b) const
{
return min_weight_a_intersect_b / (weight_a + weight_b - min_weight_a_intersect_b);
}
Expand Down
19 changes: 6 additions & 13 deletions cpp/src/link_prediction/overlap_impl.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,26 +25,19 @@ namespace detail {

struct overlap_functor_t {
template <typename weight_t>
weight_t __device__ compute_score(size_t cardinality_a,
size_t cardinality_b,
size_t cardinality_a_intersect_b,
weight_t,
weight_t,
weight_t)
weight_t __device__ compute_score(weight_t cardinality_a,
weight_t cardinality_b,
weight_t cardinality_a_intersect_b) const
{
return static_cast<weight_t>(cardinality_a_intersect_b) /
static_cast<weight_t>(std::min(cardinality_a, cardinality_b));
return cardinality_a_intersect_b / std::min(cardinality_a, cardinality_b);
}
};

struct weighted_overlap_functor_t {
template <typename weight_t>
weight_t __device__ compute_score(size_t cardinality_a,
size_t cardinality_b,
size_t cardinality_a_intersect_b,
weight_t weight_a,
weight_t __device__ compute_score(weight_t weight_a,
weight_t weight_b,
weight_t min_weight_a_intersect_b)
weight_t min_weight_a_intersect_b) const
{
return min_weight_a_intersect_b / std::min(weight_a, weight_b);
}
Expand Down
64 changes: 54 additions & 10 deletions cpp/src/link_prediction/similarity_impl.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,9 +15,16 @@
*/
#pragma once

#include <raft/core/span.hpp>
#include <prims/per_v_pair_transform_dst_nbr_intersection.cuh>
#include <prims/update_edge_src_dst_property.cuh>

#include <cugraph/graph_view.hpp>

#include <raft/core/device_span.hpp>
#include <raft/handle.hpp>

#include <rmm/device_uvector.hpp>

#include <optional>
#include <tuple>

Expand All @@ -32,15 +39,52 @@ rmm::device_uvector<weight_t> similarity(
bool use_weights,
functor_t functor)
{
CUGRAPH_FAIL("not implemented");

// Implementation, using primitives, that computes:
// For use_weights == False: cardinality of A intersect B
// For use_weights == True: sum of minimum weight in A intersect B, sum of maximum weight in A
// intersect B
//
// Then use the functor to compute the score
//
using GraphViewType = graph_view_t<vertex_t, edge_t, weight_t, false, multi_gpu>;
constexpr bool do_expensive_check = false;

CUGRAPH_EXPECTS(std::get<0>(vertex_pairs).size() == std::get<1>(vertex_pairs).size(),
"vertex pairs have mismatched sizes");

if (use_weights)
CUGRAPH_EXPECTS(graph_view.is_weighted(), "attempting to use weights on an unweighted graph");

size_t num_vertex_pairs = std::get<0>(vertex_pairs).size();
auto vertex_pairs_begin =
thrust::make_zip_iterator(std::get<0>(vertex_pairs).data(), std::get<1>(vertex_pairs).data());

if (use_weights) {
// FIXME: need implementation, similar to unweighted
// Use compute_out_weight_sums instead of compute_out_degrees
// Sum up for each common edge compute (u,a,v): min weight ((u,a), (a,v)) and
// max weight((u,a), (a,v)).
// Use these to compute weighted score
//
CUGRAPH_FAIL("weighted similarity computations are not supported in this release");
} else {
rmm::device_uvector<weight_t> similarity_score(num_vertex_pairs, handle.get_stream());

//
// Compute vertex_degree for all vertices, then distribute to each GPU.
// Need to use this instead of the dummy properties below
//
auto out_degrees = graph_view.compute_out_degrees(handle);

per_v_pair_transform_dst_nbr_intersection(
handle,
graph_view,
vertex_pairs_begin,
vertex_pairs_begin + num_vertex_pairs,
out_degrees.begin(),
[functor] __device__(auto v1, auto v2, auto v1_degree, auto v2_degree, auto intersection) {
return functor.compute_score(static_cast<weight_t>(v1_degree),
static_cast<weight_t>(v2_degree),
static_cast<weight_t>(intersection.size()));
},
similarity_score.begin(),
do_expensive_check);

return similarity_score;
}
}

} // namespace detail
Expand Down
19 changes: 6 additions & 13 deletions cpp/src/link_prediction/sorensen_impl.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,26 +25,19 @@ namespace detail {

struct sorensen_functor_t {
template <typename weight_t>
weight_t __device__ compute_score(size_t cardinality_a,
size_t cardinality_b,
size_t cardinality_a_intersect_b,
weight_t,
weight_t,
weight_t)
weight_t __device__ compute_score(weight_t cardinality_a,
weight_t cardinality_b,
weight_t cardinality_a_intersect_b) const
{
return static_cast<weight_t>(2 * cardinality_a_intersect_b) /
static_cast<weight_t>(cardinality_a + cardinality_b);
return (2 * cardinality_a_intersect_b) / (cardinality_a + cardinality_b);
}
};

struct weighted_sorensen_functor_t {
template <typename weight_t>
weight_t __device__ compute_score(size_t cardinality_a,
size_t cardinality_b,
size_t cardinality_a_intersect_b,
weight_t weight_a,
weight_t __device__ compute_score(weight_t weight_a,
weight_t weight_b,
weight_t min_weight_a_intersect_b)
weight_t min_weight_a_intersect_b) const
{
return (2 * min_weight_a_intersect_b) / (weight_a + weight_b);
}
Expand Down
4 changes: 2 additions & 2 deletions cpp/src/prims/per_v_pair_transform_dst_nbr_intersection.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,7 +147,7 @@ struct call_intersection_op_t {
auto src_offset = GraphViewType::is_storage_transposed ? minor_offset : major_offset;
auto dst_offset = GraphViewType::is_storage_transposed ? major_offset : minor_offset;
src_prop = *(vertex_property_first + src_offset);
dst_prop = *(vertex_property_first + src_offset);
dst_prop = *(vertex_property_first + dst_offset);
}
*(major_minor_pair_value_output_first + index) =
evaluate_intersection_op<GraphViewType,
Expand Down Expand Up @@ -264,7 +264,7 @@ void per_v_pair_transform_dst_nbr_intersection(
std::vector<vertex_t> h_edge_partition_major_range_lasts(
graph_view.number_of_local_edge_partitions());
for (size_t i = 0; i < graph_view.number_of_local_edge_partitions(); ++i) {
h_edge_partition_major_range_lasts[i] = graph_view.local_edge_partition_src_range_first(i);
h_edge_partition_major_range_lasts[i] = graph_view.local_edge_partition_src_range_last(i);
}
rmm::device_uvector<vertex_t> d_edge_partition_major_range_lasts(
h_edge_partition_major_range_lasts.size(), handle.get_stream());
Expand Down
Loading