|
| 1 | +#include "test_regression.h" |
| 2 | + |
| 3 | +// Regression tests for the graph builder (EvenRegularGraphBuilder) using the L2 metric. |
| 4 | +// Tests all three deglib::builder::OptimizationTarget modes: |
| 5 | +// - LowLID, HighLID, StreamingData |
| 6 | + |
| 7 | +TEST(DeglibBuilderRegressionL2, OptimizationTargetsBenchmark) |
| 8 | +{ |
| 9 | + const size_t dim = 128; |
| 10 | + const size_t base_count = 100000; |
| 11 | + const size_t query_count = 100; |
| 12 | + const size_t num_clusters = 1000; |
| 13 | + |
| 14 | + std::vector<float> base_data; |
| 15 | + std::vector<float> query_data; |
| 16 | + generate_synthetic_clustered_dataset(base_count, dim, base_data, query_data, query_count, num_clusters); |
| 17 | + |
| 18 | + auto gt_data = compute_groundtruth_l2(base_data, base_count, query_data, query_count, dim, 10); |
| 19 | + |
| 20 | + run_regression_test("LowLID", deglib::Metric::L2, 26000.0, 10.0, 0.959, |
| 21 | + base_data.data(), query_data.data(), base_count, query_count, dim, gt_data, |
| 22 | + deglib::distances::fp32_l2::L2Float{}, 100, |
| 23 | + deglib::builder::OptimizationTarget::LowLID); |
| 24 | + |
| 25 | + run_regression_test("HighLID", deglib::Metric::L2, 15000.0, 10.0, 0.888, |
| 26 | + base_data.data(), query_data.data(), base_count, query_count, dim, gt_data, |
| 27 | + deglib::distances::fp32_l2::L2Float{}, 100, |
| 28 | + deglib::builder::OptimizationTarget::HighLID); |
| 29 | + |
| 30 | + run_regression_test("StreamingData", deglib::Metric::L2, 24000.0, 24.0, 0.96, |
| 31 | + base_data.data(), query_data.data(), base_count, query_count, dim, gt_data, |
| 32 | + deglib::distances::fp32_l2::L2Float{}, 100, |
| 33 | + deglib::builder::OptimizationTarget::StreamingData); |
| 34 | +} |
| 35 | + |
| 36 | +static std::vector<uint32_t> build_graph_for_determinism( |
| 37 | + deglib::builder::OptimizationTarget optimization_target, |
| 38 | + size_t dim, size_t base_count, uint32_t edges_per_vertex, |
| 39 | + const std::vector<float>& base_data) |
| 40 | +{ |
| 41 | + const deglib::FloatSpace feature_space(dim, deglib::Metric::L2); |
| 42 | + deglib::graph::SizeBoundedGraph graph(static_cast<uint32_t>(base_count), edges_per_vertex, |
| 43 | + std::move(feature_space)); |
| 44 | + |
| 45 | + std::mt19937 rng(1337); |
| 46 | + const uint8_t extend_k = static_cast<uint8_t>(edges_per_vertex); |
| 47 | + const float extend_eps = 0.1f; |
| 48 | + const uint8_t improve_k = 0; |
| 49 | + const float improve_eps = 0.0f; |
| 50 | + const uint8_t max_path_length = 5; |
| 51 | + const uint32_t swap_tries = 0; |
| 52 | + const uint32_t additional_swap_tries = 0; |
| 53 | + |
| 54 | + deglib::builder::EvenRegularGraphBuilder builder(graph, rng, optimization_target, |
| 55 | + extend_k, extend_eps, improve_k, improve_eps, |
| 56 | + max_path_length, swap_tries, additional_swap_tries); |
| 57 | + builder.setThreadCount(1); |
| 58 | + |
| 59 | + const size_t feature_bytes = dim * sizeof(float); |
| 60 | + const std::byte* base_bytes = reinterpret_cast<const std::byte*>(base_data.data()); |
| 61 | + for (size_t i = 0; i < base_count; ++i) |
| 62 | + { |
| 63 | + const std::byte* ptr = base_bytes + i * feature_bytes; |
| 64 | + std::vector<std::byte> feat_vec(ptr, ptr + feature_bytes); |
| 65 | + builder.addEntry(static_cast<uint32_t>(i), std::move(feat_vec)); |
| 66 | + } |
| 67 | + |
| 68 | + auto build_callback = [](deglib::builder::BuilderStatus& status) {}; |
| 69 | + builder.build(build_callback); |
| 70 | + |
| 71 | + std::vector<uint32_t> neighbors; |
| 72 | + neighbors.reserve(base_count * edges_per_vertex); |
| 73 | + for (uint32_t v = 0; v < base_count; ++v) |
| 74 | + { |
| 75 | + const uint32_t* nb = graph.getNeighborIndices(v); |
| 76 | + for (uint32_t e = 0; e < edges_per_vertex; ++e) |
| 77 | + { |
| 78 | + neighbors.push_back(nb[e]); |
| 79 | + } |
| 80 | + } |
| 81 | + return neighbors; |
| 82 | +} |
| 83 | + |
| 84 | +TEST(DeglibBuilderRegressionL2, StreamingDataDeterminism) |
| 85 | +{ |
| 86 | + const size_t dim = 64; |
| 87 | + const size_t base_count = 10000; |
| 88 | + const uint32_t edges_per_vertex = 32; |
| 89 | + |
| 90 | + std::vector<float> base_data; |
| 91 | + std::vector<float> query_data; |
| 92 | + generate_synthetic_clustered_dataset(base_count, dim, base_data, query_data, 10, 50); |
| 93 | + |
| 94 | + auto graph1_neighbors = build_graph_for_determinism( |
| 95 | + deglib::builder::OptimizationTarget::StreamingData, dim, base_count, edges_per_vertex, base_data); |
| 96 | + auto graph2_neighbors = build_graph_for_determinism( |
| 97 | + deglib::builder::OptimizationTarget::StreamingData, dim, base_count, edges_per_vertex, base_data); |
| 98 | + |
| 99 | + ASSERT_EQ(graph1_neighbors.size(), graph2_neighbors.size()) |
| 100 | + << "Graph neighbor count mismatch between two builds"; |
| 101 | + |
| 102 | + size_t mismatch_count = 0; |
| 103 | + for (size_t i = 0; i < graph1_neighbors.size(); ++i) |
| 104 | + { |
| 105 | + if (graph1_neighbors[i] != graph2_neighbors[i]) |
| 106 | + mismatch_count++; |
| 107 | + } |
| 108 | + |
| 109 | + double mismatch_pct = 100.0 * mismatch_count / graph1_neighbors.size(); |
| 110 | + std::cout << "[StreamingDataDeterminism] mismatches: " << mismatch_count |
| 111 | + << " / " << graph1_neighbors.size() |
| 112 | + << " (" << mismatch_pct << "%)" << std::endl; |
| 113 | + |
| 114 | + EXPECT_EQ(0u, mismatch_count) << "Graph was not deterministic within the same process"; |
| 115 | +} |
| 116 | + |
| 117 | +TEST(DeglibBuilderRegressionL2, LowLIDDeterminism) |
| 118 | +{ |
| 119 | + const size_t dim = 64; |
| 120 | + const size_t base_count = 10000; |
| 121 | + const uint32_t edges_per_vertex = 32; |
| 122 | + |
| 123 | + std::vector<float> base_data; |
| 124 | + std::vector<float> query_data; |
| 125 | + generate_synthetic_clustered_dataset(base_count, dim, base_data, query_data, 10, 50); |
| 126 | + |
| 127 | + auto graph1_neighbors = build_graph_for_determinism( |
| 128 | + deglib::builder::OptimizationTarget::LowLID, dim, base_count, edges_per_vertex, base_data); |
| 129 | + auto graph2_neighbors = build_graph_for_determinism( |
| 130 | + deglib::builder::OptimizationTarget::LowLID, dim, base_count, edges_per_vertex, base_data); |
| 131 | + |
| 132 | + ASSERT_EQ(graph1_neighbors.size(), graph2_neighbors.size()) |
| 133 | + << "Graph neighbor count mismatch between two builds"; |
| 134 | + |
| 135 | + size_t mismatch_count = 0; |
| 136 | + for (size_t i = 0; i < graph1_neighbors.size(); ++i) |
| 137 | + { |
| 138 | + if (graph1_neighbors[i] != graph2_neighbors[i]) |
| 139 | + mismatch_count++; |
| 140 | + } |
| 141 | + |
| 142 | + double mismatch_pct = 100.0 * mismatch_count / graph1_neighbors.size(); |
| 143 | + std::cout << "[LowLIDDeterminism] mismatches: " << mismatch_count |
| 144 | + << " / " << graph1_neighbors.size() |
| 145 | + << " (" << mismatch_pct << "%)" << std::endl; |
| 146 | + |
| 147 | + EXPECT_EQ(0u, mismatch_count) << "Graph was not deterministic within the same process"; |
| 148 | +} |
| 149 | + |
| 150 | +TEST(DeglibBuilderRegressionL2, HighLIDDeterminism) |
| 151 | +{ |
| 152 | + const size_t dim = 64; |
| 153 | + const size_t base_count = 10000; |
| 154 | + const uint32_t edges_per_vertex = 32; |
| 155 | + |
| 156 | + std::vector<float> base_data; |
| 157 | + std::vector<float> query_data; |
| 158 | + generate_synthetic_clustered_dataset(base_count, dim, base_data, query_data, 10, 50); |
| 159 | + |
| 160 | + auto graph1_neighbors = build_graph_for_determinism( |
| 161 | + deglib::builder::OptimizationTarget::HighLID, dim, base_count, edges_per_vertex, base_data); |
| 162 | + auto graph2_neighbors = build_graph_for_determinism( |
| 163 | + deglib::builder::OptimizationTarget::HighLID, dim, base_count, edges_per_vertex, base_data); |
| 164 | + |
| 165 | + ASSERT_EQ(graph1_neighbors.size(), graph2_neighbors.size()) |
| 166 | + << "Graph neighbor count mismatch between two builds"; |
| 167 | + |
| 168 | + size_t mismatch_count = 0; |
| 169 | + for (size_t i = 0; i < graph1_neighbors.size(); ++i) |
| 170 | + { |
| 171 | + if (graph1_neighbors[i] != graph2_neighbors[i]) |
| 172 | + mismatch_count++; |
| 173 | + } |
| 174 | + |
| 175 | + double mismatch_pct = 100.0 * mismatch_count / graph1_neighbors.size(); |
| 176 | + std::cout << "[HighLIDDeterminism] mismatches: " << mismatch_count |
| 177 | + << " / " << graph1_neighbors.size() |
| 178 | + << " (" << mismatch_pct << "%)" << std::endl; |
| 179 | + |
| 180 | + EXPECT_EQ(0u, mismatch_count) << "Graph was not deterministic within the same process"; |
| 181 | +} |
0 commit comments