Vecsim normalize

src/VecSim/algorithms/brute_force/brute_force.cpp

@@ -157,13 +157,7 @@ double BruteForceIndex::getDistanceFrom(size_t label, const void *vector_data) {
     }
     idType id = optionalId->second;
     VectorBlockMember *vector_index = this->idToVectorBlockMemberMapping[id];
-    float normalized_blob[this->dim]; // This will be use only if metric == VecSimMetric_Cosine
-    if (this->metric == VecSimMetric_Cosine) {
-        // TODO: need more generic
-        memcpy(normalized_blob, vector_data, this->dim * sizeof(float));
-        float_vector_normalize(normalized_blob, this->dim);
-        vector_data = normalized_blob;
-    }
+
     return this->dist_func(vector_index->block->getVector(vector_index->index), vector_data,
                            &this->dim);
 }

src/VecSim/algorithms/hnsw/hnsw_wrapper.cpp

@@ -83,13 +83,6 @@ VecSimResolveCode HNSWIndex::resolveParams(VecSimRawParam *rparams, int paramNum
 }
 
 double HNSWIndex::getDistanceFrom(size_t label, const void *vector_data) {
-    if (this->metric == VecSimMetric_Cosine) {
-        // TODO: need more generic
-        float normalized_data[this->dim];
-        memcpy(normalized_data, vector_data, this->dim * sizeof(float));
-        float_vector_normalize(normalized_data, this->dim);
-        return this->hnsw->getDistanceByLabelFromPoint(label, normalized_data);
-    }
     return this->hnsw->getDistanceByLabelFromPoint(label, vector_data);
 }
 

src/VecSim/vec_sim.cpp

@@ -46,6 +46,12 @@ extern "C" double VecSimIndex_GetDistanceFrom(VecSimIndex *index, size_t id, con
     return index->getDistanceFrom(id, blob);
 }
 
+extern "C" void VecSim_Normalize(void *blob, size_t dim, VecSimType type) {
+    // TODO: need more generic
+    assert(type == VecSimType_FLOAT32);
+    float_vector_normalize((float *)blob, dim);
+}
+
 extern "C" size_t VecSimIndex_IndexSize(VecSimIndex *index) { return index->indexSize(); }
 
 extern "C" VecSimResolveCode VecSimIndex_ResolveParams(VecSimIndex *index, VecSimRawParam *rparams,

src/VecSim/vec_sim.h

@@ -44,17 +44,28 @@ int VecSimIndex_AddVector(VecSimIndex *index, const void *blob, size_t id);
 int VecSimIndex_DeleteVector(VecSimIndex *index, size_t id);
 
 /**
- * @brief Calculate the distance of a vector from an index to a vector.
+ * @brief Calculate the distance of a vector from an index to a vector. This function assumes that
+ * the vector fits the index - its type and dimension are the same as the index's, and if the
+ * index's distance metric is cosine, the vector is already normalized.
  * @param index the index from which the first vector is located, and that defines the distance
  * metric.
  * @param id the id of the vector in the index.
  * @param blob binary representation of the second vector. Blob size should match the index data
- * type and dimension.
+ * type and dimension, and pre-normalized if needed.
  * @return The distance (according to the index's distance metric) between `blob` and the vector
  * with id `id`.
  */
 double VecSimIndex_GetDistanceFrom(VecSimIndex *index, size_t id, const void *blob);
 
+/**
+ * @brief normalize the vector blob in place.
+ * @param blob binary representation of a vector. Blob size should match the specified type and
+ * dimension.
+ * @param dim vector dimension.
+ * @param type vector type.
+ */
+void VecSim_Normalize(void *blob, size_t dim, VecSimType type);
+
 /**
  * @brief Return the number of vectors in the index.
  * @param index the index whose size is requested.

src/VecSim/vec_sim_index.h

@@ -47,7 +47,7 @@ struct VecSimIndex : public VecsimBaseObject {
      * metric.
      * @param id the id of the vector in the index.
      * @param blob binary representation of the second vector. Blob size should match the index data
-     * type and dimension.
+     * type and dimension, and pre-normalized if needed.
      * @return The distance (according to the index's distance metric) between `blob` and the vector
      * with id `id`.
      */

tests/unit/test_bruteforce.cpp

@@ -971,6 +971,10 @@ TEST_F(BruteForceTest, brute_get_distance) {
     }
 
     void *query = v1;
+    void *norm = v2;                                 // {e, e}
+    VecSim_Normalize(norm, dim, VecSimType_FLOAT32); // now {1/sqrt(2), 1/sqrt(2)}
+    ASSERT_FLOAT_EQ(((float *)norm)[0], 1.0f / sqrt(2.0f));
+    ASSERT_FLOAT_EQ(((float *)norm)[1], 1.0f / sqrt(2.0f));
     double dist;
 
     // VecSimMetric_L2
@@ -990,12 +994,12 @@ TEST_F(BruteForceTest, brute_get_distance) {
     // VecSimMetric_Cosine
     distances = {5.9604644775390625e-08, 5.9604644775390625e-08, 0.0025991201400756836, 1};
     for (size_t i = 0; i < n; i++) {
-        dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], i + 1, query);
+        dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], i + 1, norm);
         ASSERT_DOUBLE_EQ(dist, distances[i]);
     }
 
     // Bad values
-    dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], 0, query);
+    dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], 0, norm);
     ASSERT_TRUE(std::isnan(dist));
     dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_L2], 46, query);
     ASSERT_TRUE(std::isnan(dist));

tests/unit/test_hnswlib.cpp

@@ -1140,6 +1140,10 @@ TEST_F(HNSWLibTest, hnsw_get_distance) {
     }
 
     void *query = v1;
+    void *norm = v2;                                 // {e, e}
+    VecSim_Normalize(norm, dim, VecSimType_FLOAT32); // now {1/sqrt(2), 1/sqrt(2)}
+    ASSERT_FLOAT_EQ(((float *)norm)[0], 1.0f / sqrt(2.0f));
+    ASSERT_FLOAT_EQ(((float *)norm)[1], 1.0f / sqrt(2.0f));
     double dist;
 
     // VecSimMetric_L2
@@ -1159,12 +1163,12 @@ TEST_F(HNSWLibTest, hnsw_get_distance) {
     // VecSimMetric_Cosine
     distances = {5.9604644775390625e-08, 5.9604644775390625e-08, 0.0025991201400756836, 1};
     for (size_t i = 0; i < n; i++) {
-        dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], i + 1, query);
+        dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], i + 1, norm);
         ASSERT_DOUBLE_EQ(dist, distances[i]);
     }
 
     // Bad values
-    dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], 0, query);
+    dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_Cosine], 0, norm);
     ASSERT_TRUE(std::isnan(dist));
     dist = VecSimIndex_GetDistanceFrom(index[VecSimMetric_L2], 46, query);
     ASSERT_TRUE(std::isnan(dist));