change the resize index factor to 1.1 instead of 2 (#148)

Author: alonre24

src/VecSim/algorithms/brute_force/brute_force.cpp

@@ -10,6 +10,7 @@
 #include <cstring>
 #include <queue>
 #include <cassert>
+#include <cmath>
 
 using namespace std;
 
@@ -76,7 +77,7 @@ int BruteForceIndex::addVector(const void *vector_data, size_t label) {
 
     // See if new id is bigger than current vector count. Needs to resize the index.
     if (id >= this->idToVectorBlockMemberMapping.size()) {
-        this->idToVectorBlockMemberMapping.resize(this->count * 2);
+        this->idToVectorBlockMemberMapping.resize(std::ceil(this->count * 1.1));
     }
 
     // Get vector block to store the vector in.

src/VecSim/algorithms/brute_force/brute_force.h

@@ -44,8 +44,9 @@ class BruteForceIndex : public VecSimIndex {
     DISTFUNC<float> dist_func;
     VecSearchMode last_mode;
 #ifdef BUILD_TESTS
-    // Allow the following tests to access the index size private member.
+    // Allow the following tests to access the index private members.
     friend class BruteForceTest_preferAdHocOptimization_Test;
     friend class BruteForceTest_test_dynamic_bf_info_iterator_Test;
+    friend class BruteForceTest_resizeIndex_Test;
 #endif
 };

src/VecSim/algorithms/hnsw/hnsw_wrapper.cpp

@@ -40,7 +40,8 @@ int HNSWIndex::addVector(const void *vector_data, size_t id) {
             vector_data = normalized_data;
         }
         if (hnsw->getIndexSize() == this->hnsw->getIndexCapacity()) {
-            this->hnsw->resizeIndex(std::max<size_t>(this->hnsw->getIndexCapacity() * 2, 2));
+            this->hnsw->resizeIndex(
+                std::max<size_t>(std::ceil(this->hnsw->getIndexCapacity() * 1.1), 2));
         }
         this->hnsw->addPoint(vector_data, id);
         return true;

tests/unit/test_bruteforce.cpp

@@ -35,6 +35,35 @@ TEST_F(BruteForceTest, brute_force_vector_add_test) {
     VecSimIndex_Free(index);
 }
 
+TEST_F(BruteForceTest, resizeIndex) {
+    size_t dim = 4;
+    size_t n = 15;
+    VecSimParams params{.algo = VecSimAlgo_BF,
+                        .bfParams = BFParams{.type = VecSimType_FLOAT32,
+                                             .dim = dim,
+                                             .metric = VecSimMetric_L2,
+                                             .initialCapacity = n}};
+    VecSimIndex *index = VecSimIndex_New(&params);
+    ASSERT_EQ(VecSimIndex_IndexSize(index), 0);
+
+    float a[dim];
+    for (size_t i = 0; i < n; i++) {
+        for (size_t j = 0; j < dim; j++) {
+            a[j] = (float)i;
+        }
+        VecSimIndex_AddVector(index, (const void *)a, i);
+    }
+    ASSERT_EQ(reinterpret_cast<BruteForceIndex *>(index)->idToVectorBlockMemberMapping.size(), n);
+
+    // Add another vector, since index size equals to the capacity, this should cause resizing
+    // (by 10% factor from the new index size).
+    VecSimIndex_AddVector(index, (const void *)a, n + 1);
+    ASSERT_EQ(VecSimIndex_IndexSize(index), n + 1);
+    ASSERT_EQ(reinterpret_cast<BruteForceIndex *>(index)->idToVectorBlockMemberMapping.size(),
+              std::ceil(1.1 * (n + 1)));
+    VecSimIndex_Free(index);
+}
+
 TEST_F(BruteForceTest, brute_force_vector_search_test_ip) {
     size_t dim = 4;
     size_t n = 100;

tests/unit/test_hnswlib.cpp

@@ -40,6 +40,35 @@ TEST_F(HNSWLibTest, hnswlib_vector_add_test) {
     VecSimIndex_Free(index);
 }
 
+TEST_F(HNSWLibTest, resizeIndex) {
+    size_t dim = 4;
+    size_t n = 15;
+    VecSimParams params{.algo = VecSimAlgo_HNSWLIB,
+                        .hnswParams = HNSWParams{.type = VecSimType_FLOAT32,
+                                                 .dim = dim,
+                                                 .metric = VecSimMetric_L2,
+                                                 .initialCapacity = n}};
+    VecSimIndex *index = VecSimIndex_New(&params);
+    ASSERT_EQ(VecSimIndex_IndexSize(index), 0);
+
+    float a[dim];
+    for (size_t i = 0; i < n; i++) {
+        for (size_t j = 0; j < dim; j++) {
+            a[j] = (float)i;
+        }
+        VecSimIndex_AddVector(index, (const void *)a, i);
+    }
+    ASSERT_EQ(reinterpret_cast<HNSWIndex *>(index)->getHNSWIndex()->getIndexCapacity(), n);
+
+    // Add another vector, since index size equals to the capacity, this should cause resizing
+    // (by 10% factor from the index size).
+    VecSimIndex_AddVector(index, (const void *)a, n + 1);
+    ASSERT_EQ(VecSimIndex_IndexSize(index), n + 1);
+    ASSERT_EQ(reinterpret_cast<HNSWIndex *>(index)->getHNSWIndex()->getIndexCapacity(),
+              std::ceil(1.1 * n));
+    VecSimIndex_Free(index);
+}
+
 TEST_F(HNSWLibTest, hnswlib_vector_search_test) {
     size_t n = 100;
     size_t k = 11;