Migrate IVF-PQ from RAFT to cuVS

cpp/include/cuvs/core/bitset.hpp

@@ -0,0 +1,30 @@
+/*
+ * Copyright (c) 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.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include <raft/core/bitset.hpp>
+
+namespace cuvs::core {
+/* To use bitset functions containing CUDA code, include <raft/core/bitset.cuh> */
+
+template <typename bitset_t, typename index_t>
+using bitset_view = raft::core::bitset_view<bitset_t, index_t>;
+
+template <typename bitset_t, typename index_t>
+using bitset = raft::core::bitset<bitset_t, index_t>;
+
+}  // end namespace cuvs::core

cpp/include/cuvs/neighbors/ann_types.hpp

@@ -16,10 +16,48 @@
 
 #pragma once
 
+#include <cstdint>
 #include <cuvs/distance/distance_types.hpp>
 
 namespace cuvs::neighbors::ann {
 
+/** Parameters for VPQ compression. */
+struct vpq_params {
+  /**
+   * The bit length of the vector element after compression by PQ.
+   *
+   * Possible values: [4, 5, 6, 7, 8].
+   *
+   * Hint: the smaller the 'pq_bits', the smaller the index size and the better the search
+   * performance, but the lower the recall.
+   */
+  uint32_t pq_bits = 8;
+  /**
+   * The dimensionality of the vector after compression by PQ.
+   * When zero, an optimal value is selected using a heuristic.
+   *
+   * TODO: at the moment `dim` must be a multiple `pq_dim`.
+   */
+  uint32_t pq_dim = 0;
+  /**
+   * Vector Quantization (VQ) codebook size - number of "coarse cluster centers".
+   * When zero, an optimal value is selected using a heuristic.
+   */
+  uint32_t vq_n_centers = 0;
+  /** The number of iterations searching for kmeans centers (both VQ & PQ phases). */
+  uint32_t kmeans_n_iters = 25;
+  /**
+   * The fraction of data to use during iterative kmeans building (VQ phase).
+   * When zero, an optimal value is selected using a heuristic.
+   */
+  double vq_kmeans_trainset_fraction = 0;
+  /**
+   * The fraction of data to use during iterative kmeans building (PQ phase).
+   * When zero, an optimal value is selected using a heuristic.
+   */
+  double pq_kmeans_trainset_fraction = 0;
+};
+
 /**
  * @defgroup ann_types Approximate Nearest Neighbors Types
  * @{
@@ -31,7 +69,7 @@ struct index {};
 /** The base for KNN index parameters. */
 struct index_params {
   /** Distance type. */
-  cuvs::distance::DistanceType metric = distance::DistanceType::L2Expanded;
+  cuvs::distance::DistanceType metric = cuvs::distance::DistanceType::L2Expanded;
   /** The argument used by some distance metrics. */
   float metric_arg = 2.0f;
   /**