Faiss

https://faiss.ai/
Admin
Faiss is a library for efficient similarity search and clustering of dense vectors with support for...
Tokens:717,102
Snippets:6,407
Trust Score:8.6
Update:5 months ago
Show doc for...
Tokens:
Raw
### verifyPQSettings_

Source: https://faiss.ai/cpp_api/namespace/namespacefaiss_1_1gpu.html

Verifies that the Product Quantizer settings are valid. Throws an exception if they are not.

```APIDOC
## verifyPQSettings_

### Description
Throws errors if configuration settings are improper.

### Throws
An exception if PQ settings are improper.
```

--------------------------------

### AlignedTable Utility Functions

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1AlignedTable.html

Utility functions for managing the AlignedTable, including getting item size, resizing, clearing, getting the number of elements, getting the number of bytes, and accessing the data.

```APIDOC
## itemsize()

### Description
Returns the size of a single item in bytes.

### Method
`inline size_t itemsize() const`
```

```APIDOC
## resize(size_t n)

### Description
Resizes the table to hold `n` elements. If the new size is larger, memory is allocated. If smaller, memory is deallocated.

### Method
`inline void resize(size_t n)`

### Parameters
* **n** (size_t) - The new number of elements the table should hold.
```

```APIDOC
## clear()

### Description
Deallocates all memory and resets the table to an empty state.

### Method
`inline void clear()`
```

```APIDOC
## size() const

### Description
Returns the current number of elements in the table.

### Method
`inline size_t size() const`
```

```APIDOC
## nbytes() const

### Description
Returns the total number of bytes currently allocated for the table.

### Method
`inline size_t nbytes() const`
```

--------------------------------

### Initialization Methods

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexIVFLocalSearchQuantizerFastScan.html

Methods for initializing and configuring the index.

```APIDOC
## void init(AdditiveQuantizer *aq, size_t nlist, MetricType metric, int bbs)

### Description
Initializes the index with an AdditiveQuantizer and other parameters.
```

```APIDOC
## void init_fastscan(size_t M, size_t nbits, size_t nlist, MetricType metric, int bbs)

### Description
Initializes fast scan parameters for the index.
```

```APIDOC
## void init_code_packer()

### Description
Initializes the code packer for efficient code storage.
```

--------------------------------

### Install Faiss with GPU support via Conda

Source: https://faiss.ai/_sources/index.rst.txt

Install the GPU-enabled version of Faiss using Conda. This package includes CUDA-accelerated indices for faster processing on compatible hardware. Ensure you have a compatible CUDA toolkit installed.

```bash
$ conda install -c pytorch faiss-gpu
```

--------------------------------

### HeapHandler begin method

Source: https://faiss.ai/cpp_api/file/simd__result__handlers_8h.html

Called at the beginning of a search operation. It can be used to initialize any necessary state, such as providing normalization factors.

```cpp
inline virtual void begin(const float *norms)
```

--------------------------------

### set_block_origin

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd__result__handlers_1_1ReservoirHandler.html

Sets the origin of the sub-matrix currently being computed. Parameters: i0_in - starting row index, j0_in - starting column index.

```APIDOC
## void set_block_origin(size_t i0_in, size_t j0_in)

### Description
Sets the sub-matrix that is being computed.

### Parameters
- **i0_in** (size_t) - starting row index
- **j0_in** (size_t) - starting column index
```

--------------------------------

### Install Faiss GPU Version

Source: https://faiss.ai/index.html

Use this command to install the GPU-enabled version of Faiss via Conda. This package includes all CPU functionalities and adds CUDA support.

```bash
$ conda install -c pytorch faiss-gpu

```

--------------------------------

### ProductResidualQuantizer::init

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1ProductResidualQuantizer.html

Initializes the ProductResidualQuantizer with sub-quantizers.

```APIDOC
## void init(size_t d, const std::vector<AdditiveQuantizer*> &aqs, Search_type_t search_type)

### Description
Initializes the product additive quantizer.

### Parameters
* **d** (size_t) - dimensionality of the input vectors
* **aqs** (const std::vector<AdditiveQuantizer*>&) - vector of AdditiveQuantizer objects
* **search_type** (Search_type_t) - AQ search type
```

--------------------------------

### begin Method

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd__result__handlers_1_1HeapHandler.html

Called at the beginning of the search process, potentially to initialize internal states or process norms.

```APIDOC
## begin Method

### Description
Initializes the handler at the start of a search operation.

### Signature
```cpp
inline virtual void begin(const float *norms)
```

### Parameters
*   **norms** (const float *) - Pointer to an array of normalization factors, if applicable.
```

--------------------------------

### Install Faiss with CPU support via Conda

Source: https://faiss.ai/_sources/index.rst.txt

Use this command to install the CPU-only version of Faiss using Conda. This package provides the core functionality for similarity search and clustering.

```bash
$ conda install -c pytorch faiss-cpu
```

--------------------------------

### profilerStart

Source: https://faiss.ai/cpp_api/file/DeviceUtils_8h.html

Starts the CUDA profiler.

```APIDOC
## profilerStart()

### Description
Starts the CUDA profiler (exposed via SWIG).
```

--------------------------------

### init_level_0_from_entry_points

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexHNSW2Level.html

Alternative graph building using entry points.

```APIDOC
## init_level_0_from_entry_points

### Description
Alternative graph building using entry points.

### Method
void init_level_0_from_entry_points(int npt, const storage_idx_t *points, const storage_idx_t *nearests)

### Parameters
* **npt** (*int*)
* **points** (*const storage_idx_t* *)
* **nearests** (*const storage_idx_t* *)
```

--------------------------------

### get_distance_computer

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexHNSW2Level.html

Gets the distance computer for the index.

```APIDOC
## get_distance_computer

### Description
Get a DistanceComputer (defined in AuxIndexStructures) object for this kind of index. DistanceComputer is implemented for indexes that support random access of their vectors.

### Method
virtual DistanceComputer *get_distance_computer() const override

### Returns
*DistanceComputer* object
```

--------------------------------

### IndexHNSW::init_level_0_from_entry_points

Source: https://faiss.ai/cpp_api/file/IndexHNSW_8h.html

Alternative graph building method using entry points.

```APIDOC
## init_level_0_from_entry_points

### Description
alternative graph building

### Parameters
* **npt** (int) - number of entry points
* **points** (const storage_idx_t *) - entry points
* **nearests** (const storage_idx_t *) - nearest neighbors
```

--------------------------------

### Initialization

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexIVFProductLocalSearchQuantizerFastScan.html

Methods for initializing the quantizer and related structures.

```APIDOC
## void init(AdditiveQuantizer *aq, size_t nlist, MetricType metric, int bbs)

### Description
Initializes the additive quantizer, number of lists, metric, and block size.

### Parameters
* **aq** (AdditiveQuantizer *) - Pointer to the AdditiveQuantizer.
* **nlist** (size_t) - The number of lists.
* **metric** (MetricType) - The distance metric.
* **bbs** (int) - The block size for fast scanning.
```

```APIDOC
## void init_fastscan(size_t M, size_t nbits, size_t nlist, MetricType metric, int bbs)

### Description
Initializes the fast scan parameters.

### Parameters
* **M** (size_t) - The number of subquantizers.
* **nbits** (size_t) - The number of bits per subquantizer.
* **nlist** (size_t) - The number of lists.
* **metric** (MetricType) - The distance metric.
* **bbs** (int) - The block size for fast scanning.
```

```APIDOC
## void init_code_packer()

### Description
Initializes the code packer for efficient code storage.
```

--------------------------------

### combination_name

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1ParameterSpace.html

Gets the string representation of a parameter combination.

```APIDOC
## std::string combination_name(size_t cno) const

### Description
Retrieves the string representation of a specific parameter combination identified by its index.

### Method
`const` member function

### Endpoint
N/A

### Parameters
- **cno** (size_t) - The index of the combination.

### Request Example
None

### Response
- **return value** (std::string) - The string name of the combination.
```

--------------------------------

### Get Code Packer

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexFlatIP.html

Retrieves a CodePacker object.

```APIDOC
## CodePacker *get_CodePacker() const

### Description
Returns a pointer to a `CodePacker` object associated with this index.
```

--------------------------------

### initializeForDevice

Source: https://faiss.ai/cpp_api/file/StandardGpuResources_8h.html

Initializes GPU resources for a specific device. This is an internal system call.

```APIDOC
## void initializeForDevice(int device)

### Description
Internal system calls. Initialize resources for this device.

### Parameters
#### Path Parameters
- **device** (int) - The device ID to initialize.

### Method
POST
```

--------------------------------

### Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexProductLocalSearchQuantizerFastScan.html

Initializes a new instance of IndexProductLocalSearchQuantizerFastScan.

```APIDOC
## IndexProductLocalSearchQuantizerFastScan(int d, size_t nsplits, size_t Msub, size_t nbits, MetricType metric = METRIC_L2, Search_type_t search_type = AdditiveQuantizer::ST_norm_rq2x4, int bbs = 32)

### Description
Constructor for IndexProductLocalSearchQuantizerFastScan.

### Parameters
* **d** (int) - Dimensionality of the input vectors.
* **nsplits** (size_t) - Number of local search quantizers.
* **Msub** (size_t) - Number of subquantizers per LSQ.
* **nbits** (size_t) - Number of bits per subvector index.
* **metric** (MetricType) - Metric type (default: METRIC_L2).
* **search_type** (Search_type_t) - AQ search type (default: AdditiveQuantizer::ST_norm_rq2x4).
* **bbs** (int) - Block size for batch processing (default: 32).
```

--------------------------------

### filedescriptor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1FileIOReader.html

Gets the file descriptor associated with the reader.

```APIDOC
## int filedescriptor()

### Description
Returns the underlying file descriptor for the opened file.

### Returns
An integer representing the file descriptor.
```

--------------------------------

### get_iterator

Source: https://faiss.ai/cpp_api/file/InvertedLists_8h.html

Gets an iterator for lists that support iteration.

```APIDOC
## get_iterator

### Description
Get iterable for lists that use_iterator.

### Parameters
- **list_no** (size_t) - The index of the list.
- **inverted_list_context** (void *) - Optional context for the inverted list.

### Returns
- InvertedListsIterator * - An iterator for the specified list.
```

--------------------------------

### ProductLocalSearchQuantizer::init

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1ProductLocalSearchQuantizer.html

Initializes the ProductLocalSearchQuantizer with a given dimensionality, a vector of AdditiveQuantizer pointers, and a search type.

```APIDOC
## void init(size_t d, const std::vector<AdditiveQuantizer*> &aqs, Search_type_t search_type)

### Description
Initializes the product additive quantizer.

### Parameters
* **d** (size_t) - dimensionality of the input vectors
* **aqs** (const std::vector<AdditiveQuantizer*>&) - vector of additive quantizers
* **search_type** (Search_type_t) - AQ search type
```

--------------------------------

### get()

Source: https://faiss.ai/cpp_api/class/classfaiss_1_1gpu_1_1CublasHandleScope.html

Retrieves the underlying cublasHandle_t managed by the scope.

```APIDOC
## get()

### Description
Returns the cublasHandle_t managed by this RAII scope. This handle can be used for subsequent CUBLAS operations.

### Method
`inline cublasHandle_t get()`

### Parameters
None

### Returns
cublasHandle_t - The active CUBLAS handle.
```

--------------------------------

### init Method

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexIVFProductResidualQuantizerFastScan.html

Initializes the IndexIVFProductResidualQuantizerFastScan with an AdditiveQuantizer.

```APIDOC
## void init(AdditiveQuantizer *aq, size_t nlist, MetricType metric, int bbs)

### Description
Initializes the quantizer with an AdditiveQuantizer.

### Parameters
* **aq** (AdditiveQuantizer *) - The additive quantizer to use.
* **nlist** (size_t) - The number of lists.
* **metric** (MetricType) - The distance metric.
* **bbs** (int) - Block size for fast scan.
```

--------------------------------

### GpuIndexFlat Distance Computer

Source: https://faiss.ai/cpp_api/namespace/namespacefaiss_1_1gpu.html

Gets a distance computer for the index.

```APIDOC
## get_distance_computer()

### Description
Get a `DistanceComputer` (defined in AuxIndexStructures) object for this kind of index. `DistanceComputer` is implemented for indexes that support random access of their vectors.

### Method
`virtual DistanceComputer *get_distance_computer() const`
```

--------------------------------

### void display() const

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexBinaryIVF.html

Displays information about the IndexBinaryIVF class and its current state.

```APIDOC
## display

### Description
Display the actual class name and some more info.
```

--------------------------------

### get Method

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1gpu_1_1GpuIcmEncoderFactory.html

Retrieves an IcmEncoder instance for GPU processing.

```APIDOC
## lsq::IcmEncoder *get(const LocalSearchQuantizer *lsq) override

### Description
Returns an `IcmEncoder` object configured for GPU acceleration, based on the provided `LocalSearchQuantizer`.

### Parameters
* **lsq** (const LocalSearchQuantizer *) - A pointer to the LocalSearchQuantizer object to be used for encoding.
```

--------------------------------

### initialize

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1ParameterSpace.html

Initializes the parameter space with reasonable default parameters for a given index.

```APIDOC
## virtual void initialize(const Index *index)

### Description
Initializes the parameter space with sensible default values suitable for the provided Faiss index.

### Method
`virtual` member function

### Endpoint
N/A

### Parameters
- **index** (const Index *) - A pointer to the Faiss index for which to initialize parameters.

### Request Example
None

### Response
None
```

--------------------------------

### IndexIVFPQ Constructor

Source: https://faiss.ai/cpp_api/file/IndexIVFPQ_8h.html

Initializes an IndexIVFPQ object, which represents an inverted file index with Product Quantizer encoding. It sets up the index with a given quantizer, dimensionality, number of lists, product quantizer parameters, and metric type.

```APIDOC
## IndexIVFPQ(Index *quantizer, size_t d, size_t nlist, size_t M, size_t nbits_per_idx, MetricType metric = METRIC_L2)

### Description
Inverted file with Product Quantizer encoding. Each residual vector is encoded as a product quantizer code.

### Parameters
* **quantizer** (Index *) - The index used for coarse quantization.
* **d** (size_t) - The dimensionality of the vectors.
* **nlist** (size_t) - The number of inverted lists (clusters).
* **M** (size_t) - The number of subquantizers in the Product Quantizer.
* **nbits_per_idx** (size_t) - The number of bits per subquantizer index.
* **metric** (MetricType) - The distance metric to use (default: METRIC_L2).
```

--------------------------------

### Get Code Packer

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexAdditiveQuantizerFastScan.html

Retrieves the CodePacker associated with the index.

```APIDOC
## CodePacker *get_CodePacker() const

### CodePacker *get_CodePacker() const
```

--------------------------------

### GPU Resource Initialization and Handles

Source: https://faiss.ai/cpp_api/namespace/namespacefaiss_1_1gpu.html

Functions for initializing GPU resources and retrieving handles for cuBLAS and streams.

```APIDOC
## virtual void initializeForDevice(int device)

### Description
Internal system calls. Initialize resources for this device.

### Method
virtual void
```

```APIDOC
## virtual cublasHandle_t getBlasHandle(int device)

### Description
Returns the cuBLAS handle that we use for the given device.

### Method
virtual cublasHandle_t
```

```APIDOC
## cublasHandle_t getBlasHandleCurrentDevice()

### Description
Calls getBlasHandle with the current device.

### Method
cublasHandle_t
```

```APIDOC
## cudaStream_t getDefaultStreamCurrentDevice()

### Description
Calls getDefaultStream with the current device.

### Method
cudaStream_t
```

```APIDOC
## std::vector<cudaStream_t> getAlternateStreamsCurrentDevice()

### Description
Calls getAlternateStreams for the current device.

### Method
std::vector<cudaStream_t>
```

```APIDOC
## cudaStream_t getAsyncCopyStreamCurrentDevice()

### Description
Calls getAsyncCopyStream for the current device.

### Method
cudaStream_t
```

--------------------------------

### Get FlatCodesDistanceComputer

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexFlatIP.html

Retrieves a distance computer for flat codes.

```APIDOC
## FlatCodesDistanceComputer *get_FlatCodesDistanceComputer() const override

### Description
Returns a pointer to a `FlatCodesDistanceComputer` object. This computer offers a `distance_to_code` method. The default implementation explicitly decodes the vector using `sa_decode`.
```

--------------------------------

### hashtable_int64_to_int64_init

Source: https://faiss.ai/cpp_api/file/sorting_8h.html

Initializes a hash table for int64 to int64 mapping with external storage.

```APIDOC
## hashtable_int64_to_int64_init

### Description
Hashtable implementation for int64 -> int64 with external storage implemented for fast batch add and lookup.
tab is of size 2 * (1 << log2_capacity)

### Parameters
#### Parameters
- **log2_capacity** (int) - log2 of the capacity of the hash table
- **tab** (int64_t *) - pointer to the hash table storage
```

--------------------------------

### get

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1DirectMap.html

Retrieves an entry from the direct map given an ID.

```APIDOC
## idx_t get(idx_t id) const

### Description
Retrieves an entry from the direct map using the provided ID. This is a core function for looking up mapped values.

### Method
`idx_t`

### Parameters
* **id** (idx_t) - The ID to look up.

### Returns
The mapped entry for the given ID.
```

--------------------------------

### pq4_get_packed_element

Source: https://faiss.ai/cpp_api/file/pq4__fast__scan_8h.html

Get a single element from a packed codes table.

```APIDOC
## pq4_get_packed_element

### Description
Get a single element from a packed codes table.

### Parameters
#### Path Parameters
- **data** (const uint8_t *) - packed codes array
- **bbs** (size_t) - size of database blocks (multiple of 32)
- **nsq** (size_t) - number of sub-quantizers (multiple of 2)
- **vector_id** (size_t) - vector id
- **sq** (size_t) - subquantizer (< nsq).
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/class/classfaiss_1_1gpu_1_1GpuIndexIVFPQ.html

Constructors for GpuIndexIVFPQ allow initialization from existing faiss::IndexIVFPQ instances or by specifying parameters for a new index.

```APIDOC
## GpuIndexIVFPQ(GpuResourcesProvider *provider, const faiss::IndexIVFPQ *index, GpuIndexIVFPQConfig config = GpuIndexIVFPQConfig())

### Description
Construct from a pre-existing faiss::IndexIVFPQ instance, copying data over to the given GPU, if the input index is trained.

## GpuIndexIVFPQ(GpuResourcesProvider *provider, int dims, idx_t nlist, idx_t subQuantizers, idx_t bitsPerCode, faiss::MetricType metric = faiss::METRIC_L2, GpuIndexIVFPQConfig config = GpuIndexIVFPQConfig())

### Description
Constructs a new instance with an empty flat quantizer; the user provides the number of IVF lists desired.

## GpuIndexIVFPQ(GpuResourcesProvider *provider, Index *coarseQuantizer, int dims, idx_t nlist, idx_t subQuantizers, idx_t bitsPerCode, faiss::MetricType metric = faiss::METRIC_L2, GpuIndexIVFPQConfig config = GpuIndexIVFPQConfig())

### Description
Constructs a new instance with a provided CPU or GPU coarse quantizer; the user provides the number of IVF lists desired.
```

--------------------------------

### GpuIndexIVFFlat Initialization

Source: https://faiss.ai/cpp_api/file/GpuIndexIVFFlat_8h.html

Initializes the GpuIndexIVFFlat with specified parameters. This method sets up the index structure on the GPU.

```APIDOC
## void setIndex_(GpuResources *resources, int dim, int nlist, faiss::MetricType metric, float metricArg, bool useResidual, faiss::ScalarQuantizer *scalarQ, bool interleavedLayout, IndicesOptions indicesOptions, MemorySpace space)

### Description
Initialize appropriate index.

### Parameters
- **resources** (GpuResources *) - Pointer to GPU resources.
- **dim** (int) - The dimension of the vectors.
- **nlist** (int) - The number of inverted lists.
- **metric** (faiss::MetricType) - The distance metric to use.
- **metricArg** (float) - Argument for the distance metric.
- **useResidual** (bool) - Whether to use residual quantization.
- **scalarQ** (faiss::ScalarQuantizer *) - Optional ScalarQuantizer for quantization.
- **interleavedLayout** (bool) - Whether to use interleaved layout for data.
- **indicesOptions** (IndicesOptions) - Options for index storage.
- **space** (MemorySpace) - The memory space to use (e.g., GPU).
```

--------------------------------

### Get Centroids

Source: https://faiss.ai/cpp_api/file/ProductQuantizer_8h.html

Retrieves the centroids associated with a specific subvector.

```APIDOC
## float *get_centroids(size_t m, size_t i)

### Description
Returns a pointer to the centroids associated with subvector `m` and centroid index `i`.

### Method
`float *get_centroids(size_t m, size_t i)`

### Parameters
- **m** (size_t) - The index of the subquantizer.
- **i** (size_t) - The index of the centroid within the subquantizer.
```

```APIDOC
## const float *get_centroids(size_t m, size_t i) const

### Description
Returns a const pointer to the centroids associated with subvector `m` and centroid index `i`.

### Method
`const float *get_centroids(size_t m, size_t i) const`

### Parameters
- **m** (size_t) - The index of the subquantizer.
- **i** (size_t) - The index of the centroid within the subquantizer.
```

--------------------------------

### ProductAdditiveQuantizer::init

Source: https://faiss.ai/cpp_api/file/ProductAdditiveQuantizer_8h.html

Initializes the product additive quantizer with specified parameters.

```APIDOC
## void init(size_t d, const std::vector<AdditiveQuantizer*> &aqs, Search_type_t search_type)

### Description
Initializes the product additive quantizer.

### Parameters
* **d** (size_t) - dimensionality of the input vectors
* **aqs** (const std::vector<AdditiveQuantizer*>&) - sub-additive quantizers
* **search_type** (Search_type_t) - AQ search type
```

--------------------------------

### IndexFlatCodes::get_distance_computer

Source: https://faiss.ai/cpp_api/file/IndexFlatCodes_8h.html

Gets a distance computer object for the index.

```APIDOC
## DistanceComputer *get_distance_computer() const

### Description
Retrieves a `DistanceComputer` object for this index. This is available for indexes that support random access to their vectors. The `DistanceComputer` is defined in `AuxIndexStructures`.
```

--------------------------------

### begin Method

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1Top1BlockResultHandler_1_1SingleResultHandler.html

Marks the beginning of results processing for a specific query.

```APIDOC
## begin Method

### Description
Marks the beginning of results processing for a specific query.

### Signature
`inline void begin(const size_t current_idx_2)`

### Parameters
*   **current_idx_2** (*const size_t*) - The index of the current query being processed.
```

--------------------------------

### CpuTimer Constructor

Source: https://faiss.ai/cpp_api/class/classfaiss_1_1gpu_1_1CpuTimer.html

Creates and starts a new CpuTimer instance.

```APIDOC
## CpuTimer()

### Description
Creates and starts a new timer.

### Method
Constructor

### Parameters
None
```

--------------------------------

### begin

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd__result__handlers_1_1SingleResultHandler.html

Called at the beginning of the result handling process, potentially receiving normalization factors.

```APIDOC
## begin

### Description
Called at the beginning of the result handling process, potentially receiving normalization factors.

### Signature
`inline virtual void begin(const float *norms)`

### Parameters
- **norms** (const float *) - Pointer to normalization factors, if applicable.
```

--------------------------------

### Get Code Packer

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexIVFFlatDedup.html

Returns the `CodePacker` object used by this index.

```APIDOC
## get_CodePacker

Returns the `CodePacker` object used by this index.

### Method
virtual CodePacker *get_CodePacker() const
```

--------------------------------

### Get Distance Computer

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexFlatIP.html

Retrieves a distance computer object for this index.

```APIDOC
## DistanceComputer *get_distance_computer() const override

### Description
Returns a `DistanceComputer` object for this index. This is available for indexes that support random access of their vectors. The `DistanceComputer` interface is defined in `AuxIndexStructures`.
```

--------------------------------

### verifyIVFSettings_

Source: https://faiss.ai/cpp_api/namespace/namespacefaiss_1_1gpu.html

Verifies the settings for the Inverted File (IVF) index. Throws an exception if settings are invalid.

```APIDOC
## verifyIVFSettings_

### Description
Verifies the settings for the Inverted File (IVF) index. Throws an exception if settings are invalid.

### Throws
An exception if IVF settings are invalid.
```

--------------------------------

### Get Code Size

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexFlatIP.html

Returns the size of the encoded vectors in bytes.

```APIDOC
## size_t sa_code_size() const override

### Description
Returns the size in bytes of the encoded representation for a single vector.
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/class/classfaiss_1_1gpu_1_1GpuIndexIVFScalarQuantizer.html

Provides different ways to construct a GpuIndexIVFScalarQuantizer, either from an existing CPU index, or by specifying parameters for a new index.

```APIDOC
## Constructor Overloads

### GpuIndexIVFScalarQuantizer(GpuResourcesProvider *provider, const faiss::IndexIVFScalarQuantizer *index, GpuIndexIVFScalarQuantizerConfig config = GpuIndexIVFScalarQuantizerConfig())

**Description**: Constructs from a pre-existing faiss::IndexIVFScalarQuantizer instance, copying data over to the given GPU, if the input index is trained.

### GpuIndexIVFScalarQuantizer(GpuResourcesProvider *provider, int dims, idx_t nlist, faiss::ScalarQuantizer::QuantizerType qtype, faiss::MetricType metric = MetricType::METRIC_L2, bool encodeResidual = true, GpuIndexIVFScalarQuantizerConfig config = GpuIndexIVFScalarQuantizerConfig())

**Description**: Constructs a new instance with an empty flat quantizer; the user provides the number of IVF lists desired.

### GpuIndexIVFScalarQuantizer(GpuResourcesProvider *provider, Index *coarseQuantizer, int dims, idx_t nlist, faiss::ScalarQuantizer::QuantizerType qtype, faiss::MetricType metric = MetricType::METRIC_L2, bool encodeResidual = true, GpuIndexIVFScalarQuantizerConfig config = GpuIndexIVFScalarQuantizerConfig())

**Description**: Constructs a new instance with a provided CPU or GPU coarse quantizer; the user provides the number of IVF lists desired.
```

--------------------------------

### Get Base Vectors

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexFlatIP.html

Provides access to the underlying base vectors.

```APIDOC
## float *get_xb()

### Description
Returns a pointer to the beginning of the underlying base vectors array.
```

```APIDOC
## const float *get_xb() const

### Description
Returns a const pointer to the beginning of the underlying base vectors array.
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd64uint8.html

Various constructors are available to initialize a simd64uint8 object.

```APIDOC
## Constructors

### Default Constructor
```cpp
inline simd64uint8()
```

### Constructor from __m512i
```cpp
inline explicit simd64uint8(__m512i i)
```

### Constructor from int
```cpp
inline explicit simd64uint8(int x)
```

### Constructor from uint8_t
```cpp
inline explicit simd64uint8(uint8_t x)
```

### Constructor from simd256bit (low part)
```cpp
inline explicit simd64uint8(simd256bit lo)
```

### Constructor from two simd256bit vectors
```cpp
inline explicit simd64uint8(simd256bit lo, simd256bit hi)
```

### Constructor from simd512bit
```cpp
inline explicit simd64uint8(simd512bit x)
```

### Constructor from const uint8_t pointer
```cpp
inline explicit simd64uint8(const uint8_t *x)
```
```

--------------------------------

### reconstruct_n

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexProductResidualQuantizer.html

Reconstructs a sequence of vectors from the index, starting from a given index.

```APIDOC
## void reconstruct_n(idx_t i0, idx_t ni, float *recons) const

### Description
Reconstruct vectors i0 to i0 + ni - 1. This function may not be defined for some indexes.

### Parameters
- **i0** (idx_t) - index of the first vector in the sequence
- **ni** (idx_t) - number of vectors in the sequence
- **recons** (float *) - reconstructed vector (size ni * d)
```

--------------------------------

### begin

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd__result__handlers_1_1ReservoirHandler.html

Called at the beginning of processing, potentially with normalization information.

```APIDOC
## virtual void begin(const float *norms)

### Description
Called at the beginning of processing, potentially with normalization information.

### Parameters
- **norms** (const float *) - pointer to normalization values.
```

--------------------------------

### add

Source: https://faiss.ai/cpp_api/namespace/namespacefaiss_1_1gpu.html

`x` can be resident on the CPU or any GPU; copies are performed as needed. Handles paged adds if the add set is too large; calls addInternal_.

```APIDOC
## add

### Description
Adds vectors to the index. `x` can be resident on the CPU or any GPU; copies are performed as needed. Handles paged adds if the add set is too large; calls addInternal_.

### Method
`virtual void add(idx_t, const float *x) override`
```

--------------------------------

### get() Method

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1InvertedLists_1_1ScopedIds.html

Retrieves a raw pointer to the beginning of the inverted list IDs.

```APIDOC
## get() Method

### Description
Returns a const pointer to the array of IDs for the specified inverted list.

### Signature
`inline const idx_t *get()`

### Returns
`const idx_t *` - A pointer to the first ID in the list.
```

--------------------------------

### Get List Size

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexIVFFlatDedup.html

Returns the number of vectors in a specific inverted list.

```APIDOC
## get_list_size

Returns the number of vectors in a specific inverted list.

### Method
inline size_t get_list_size(size_t list_no) const

### Parameters
* **list_no** (size_t) - The index of the inverted list.
```

--------------------------------

### Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexIVFProductLocalSearchQuantizer.html

Constructs an IndexIVFProductLocalSearchQuantizer. This constructor initializes the index with a given quantizer, dimensionality, number of lists, number of splits for local search, subquantizer configuration, bit depth, metric type, and search type.

```APIDOC
## IndexIVFProductLocalSearchQuantizer(Index *quantizer, size_t d, size_t nlist, size_t nsplits, size_t Msub, size_t nbits, MetricType metric = METRIC_L2, Search_type_t search_type = AdditiveQuantizer::ST_decompress)

### Description
Constructor for IndexIVFProductLocalSearchQuantizer.

### Parameters
* **quantizer** - The main quantizer index.
* **d** - Dimensionality of the input vectors.
* **nlist** - Number of inverted lists.
* **nsplits** - Number of local search quantizers.
* **Msub** - Number of subquantizers per Local Search Quantizer (LSQ).
* **nbits** - Number of bits per subvector index.
* **metric** - The distance metric to use (default: METRIC_L2).
* **search_type** - The type of search to perform (default: AdditiveQuantizer::ST_decompress).
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd32uint16.html

Various constructors are available to initialize a simd32uint16 object.

```APIDOC
## Constructors

### Default Constructor
```cpp
inline simd32uint16()
```

### Constructor from __m512i
```cpp
inline explicit simd32uint16(__m512i i)
```

### Constructor from int
```cpp
inline explicit simd32uint16(int x)
```

### Constructor from uint16_t
```cpp
inline explicit simd32uint16(uint16_t x)
```

### Constructor from simd512bit
```cpp
inline explicit simd32uint16(simd512bit x)
```

### Constructor from const uint16_t*
```cpp
inline explicit simd32uint16(const uint16_t *x)
```

### Constructor from simd256bit (low)
```cpp
inline explicit simd32uint16(simd256bit lo)
```

### Constructor from two simd256bit (low and high)
```cpp
inline explicit simd32uint16(simd256bit lo, simd256bit hi)
```
```

--------------------------------

### SliceInvertedLists::list_size

Source: https://faiss.ai/cpp_api/file/InvertedLists_8h.html

Gets the size of a specific list within the sliced `InvertedLists`.

```APIDOC
## SliceInvertedLists::list_size

### Description
Gets the size of a list within the slice.

### Signature
virtual size_t list_size(size_t list_no) const override
```

--------------------------------

### IndexPQStats Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexPQStats.html

Initializes a new instance of the IndexPQStats struct.

```APIDOC
## IndexPQStats()

### Description
Constructs a new IndexPQStats object with default values.

### Method
`inline IndexPQStats()`

### Parameters
None
```

--------------------------------

### LSQTimerScope Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1lsq_1_1LSQTimerScope.html

Constructs an LSQTimerScope object, starting the timer with a given name.

```APIDOC
## LSQTimerScope(LSQTimer *timer, std::string name)

### Description
Initializes a new instance of the LSQTimerScope class, associating it with a timer and a name for the timed operation.

### Parameters
#### Path Parameters
- **timer** (LSQTimer *) - A pointer to the LSQTimer object to be used for timing.
- **name** (std::string) - The name of the operation being timed.
```

--------------------------------

### search_level_0

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexHNSW2Level.html

Performs search only on level 0, given the starting points for each vertex.

```APIDOC
## search_level_0

### Description
Perform search only on level 0, given the starting points for each vertex.

### Method
void search_level_0(idx_t n, const float *x, idx_t k, const storage_idx_t *nearest, const float *nearest_d, float *distances, idx_t *labels, int nprobe = 1, int search_type = 1, const SearchParameters *params = nullptr) const

### Parameters
* **n** (*idx_t*)
* **x** (*const float* *)
* **k** (*idx_t*)
* **nearest** (*const storage_idx_t* *)
* **nearest_d** (*const float* *)
* **distances** (*float* *)
* **labels** (*idx_t* *)
* **nprobe** (*int*) - Defaults to 1
* **search_type** (*int*) - 1:perform one search per nprobe, 2: enqueue all entry points. Defaults to 1
* **params** (*const SearchParameters* *) - Optional search parameters
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd16uint16.html

Various constructors for initializing simd16uint16 objects.

```APIDOC
## Constructors

### Default Constructor
```cpp
simd16uint16()
```

### Constructor from __m256i
```cpp
explicit simd16uint16(__m256i i)
```

### Constructor from int
```cpp
explicit simd16uint16(int x)
```

### Constructor from uint16_t
```cpp
explicit simd16uint16(uint16_t x)
```

### Constructor from simd256bit
```cpp
explicit simd16uint16(simd256bit x)
```

### Constructor from const uint16_t pointer
```cpp
explicit simd16uint16(const uint16_t *x)
```

### Constructor with 16 uint16_t values
```cpp
simd16uint16(uint16_t u0, uint16_t u1, uint16_t u2, uint16_t u3, uint16_t u4, uint16_t u5, uint16_t u6, uint16_t u7, uint16_t u8, uint16_t u9, uint16_t u10, uint16_t u11, uint16_t u12, uint16_t u13, uint16_t u14, uint16_t u15)
```

### Constructor from uint16x8x2_t
```cpp
explicit simd16uint16(const uint16x8x2_t &v)
```

### Template Constructor from any simd256bit type
```cpp
template<typename T, typename std::enable_if<detail::simdlib::is_simd256bit<T>::value, std::nullptr_t>::type = nullptr>
inline explicit simd16uint16(const T &x)
```
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1simd16uint16.html

Various constructors for initializing simd16uint16 objects.

```APIDOC
## Constructors

### Default Constructor
```cpp
inline simd16uint16()
```

### Integer Constructor
```cpp
inline explicit simd16uint16(int x)
```

### Unsigned 16-bit Integer Constructor
```cpp
inline explicit simd16uint16(uint16_t x)
```

### 256-bit SIMD Constructor
```cpp
inline explicit simd16uint16(const simd256bit &x)
```

### Pointer Constructor (uint16_t)
```cpp
inline explicit simd16uint16(const uint16_t *x)
```

### Variadic Constructor (16 uint16_t values)
```cpp
inline explicit simd16uint16(uint16_t u0, uint16_t u1, uint16_t u2, uint16_t u3, uint16_t u4, uint16_t u5, uint16_t u6, uint16_t u7, uint16_t u8, uint16_t u9, uint16_t u10, uint16_t u11, uint16_t u12, uint16_t u13, uint16_t u14, uint16_t u15)
```
```

--------------------------------

### Get most significant bits of each byte

Source: https://faiss.ai/cpp_api/file/simdlib__emulated_8h.html

Extracts the most significant bit from each byte of a simd32uint8 vector.

```cpp
inline uint32_t get_MSBs(const simd32uint8 &a) {
    
}
```

--------------------------------

### Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexProductResidualQuantizerFastScan.html

Initializes a new instance of the IndexProductResidualQuantizerFastScan struct.

```APIDOC
## IndexProductResidualQuantizerFastScan(int d, size_t nsplits, size_t Msub, size_t nbits, MetricType metric = METRIC_L2, Search_type_t search_type = AdditiveQuantizer::ST_norm_rq2x4, int bbs = 32)

### Description
Constructor for IndexProductResidualQuantizerFastScan.

### Parameters
* **d** (int) - dimensionality of the input vectors
* **nsplits** (size_t) - number of residual quantizers
* **Msub** (size_t) - number of subquantizers per RQ
* **nbits** (size_t) - number of bit per subvector index
* **metric** (MetricType) - metric type (default: METRIC_L2)
* **search_type** (Search_type_t) - AQ search type (default: AdditiveQuantizer::ST_norm_rq2x4)
* **bbs** (int) - block size for search (default: 32)
```

--------------------------------

### IcmEncoderFactory::get

Source: https://faiss.ai/cpp_api/file/LocalSearchQuantizer_8h.html

Factory method to get an IcmEncoder instance. This is an inline virtual function.

```APIDOC
## get

### Description
Factory method to get an IcmEncoder instance.

### Signature
`inline virtual IcmEncoder *get(const LocalSearchQuantizer *lsq)`

### Parameters
* **lsq** (const LocalSearchQuantizer *) - Pointer to the LocalSearchQuantizer object.
```

--------------------------------

### ProductQuantizer Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1ProductQuantizer.html

Constructs a ProductQuantizer object. You can initialize it with dimensions or use the default constructor.

```APIDOC
## ProductQuantizer Constructor

### Description
Constructs a ProductQuantizer object. You can initialize it with dimensions or use the default constructor.

### Signature
ProductQuantizer(size_t d, size_t M, size_t nbits)
ProductQuantizer()

### Parameters
- **d** (size_t) - The total dimensionality of the vectors.
- **M** (size_t) - The number of subquantizers (subvectors).
- **nbits** (size_t) - The number of bits to use for each subquantizer's code.

### Usage
```cpp
// Example with dimensions
faiss::ProductQuantizer pq(d, M, nbits);

// Example with default constructor (dimensions must be set later)
faiss::ProductQuantizer pq_default;
```
```

--------------------------------

### get_InvertedListScanner

Source: https://faiss.ai/cpp_api/file/IndexIVFFlat_8h.html

Gets a scanner for this index. The default search implementation uses this to compute the distances.

```APIDOC
## get_InvertedListScanner

### Description
Gets a scanner for this index. The default search implementation uses this to compute the distances.

### Signature
`virtual InvertedListScanner *get_InvertedListScanner(bool store_pairs, const IDSelector *sel) const override`

### Parameters
*   **store_pairs** (bool) - If true, ignore labels.
*   **sel** (const IDSelector *) - The ID selector.
```

--------------------------------

### ProductQuantizer Constructor

Source: https://faiss.ai/cpp_api/file/ProductQuantizer_8h.html

Constructs a ProductQuantizer object. The default constructor initializes an empty quantizer.

```APIDOC
## ProductQuantizer()

### Description
Default constructor for ProductQuantizer.

### Method
ProductQuantizer()

### Parameters
None
```

--------------------------------

### reconstruct_n

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexNSGPQ.html

Reconstructs a contiguous sequence of vectors from the index, specified by a starting index and a count.

```APIDOC
## void reconstruct_n(idx_t i0, idx_t ni, float *recons) const

### Description
Reconstructs vectors from index i0 to i0 + ni - 1. This function may not be defined for some indexes.

### Method
`void reconstruct_n(idx_t i0, idx_t ni, float *recons) const`

### Parameters:
* **i0** - The index of the first vector in the sequence.
* **ni** - The number of vectors in the sequence.
* **recons** - Output buffer for the reconstructed vectors (size ni * d).
```

--------------------------------

### Constructors

Source: https://faiss.ai/cpp_api/class/classfaiss_1_1gpu_1_1GpuIndexFlat.html

Constructs a GpuIndexFlat instance. It can be initialized from an existing faiss::IndexFlat, or as an empty instance.

```APIDOC
## Constructors

### GpuIndexFlat(GpuResourcesProvider *provider, const faiss::IndexFlat *index, GpuIndexFlatConfig config = GpuIndexFlatConfig())

Construct from a pre-existing faiss::IndexFlat instance, copying data over to the given GPU.

### GpuIndexFlat(std::shared_ptr<GpuResources> resources, const faiss::IndexFlat *index, GpuIndexFlatConfig config = GpuIndexFlatConfig())

### GpuIndexFlat(GpuResourcesProvider *provider, int dims, faiss::MetricType metric, GpuIndexFlatConfig config = GpuIndexFlatConfig())

Construct an empty instance that can be added to.

### GpuIndexFlat(std::shared_ptr<GpuResources> resources, int dims, faiss::MetricType metric, GpuIndexFlatConfig config = GpuIndexFlatConfig())


```

--------------------------------

### IndexHNSWSQ::search_level_0

Source: https://faiss.ai/cpp_api/file/IndexHNSW_8h.html

Performs a search only on level 0, given the starting points for each vertex.

```APIDOC
## search_level_0

### Description
Perform search only on level 0, given the starting points for each vertex.

### Parameters
* **n** (idx_t) - number of vectors
* **x** (const float *) - input vectors to search, size n * d
* **k** (idx_t) - number of nearest neighbors to search for
* **nearest** (const storage_idx_t *) - starting points for each vertex
* **nearest_d** (const float *) - distances to the starting points
* **distances** (float *) - output array for distances
* **labels** (idx_t *) - output array for labels
* **nprobe** (int) - number of probes (default: 1)
* **search_type** (int) - type of search (1: perform one search per nprobe, 2: enqueue all entry points) (default: 1)
* **params** (const SearchParameters *) - additional search parameters (default: nullptr)
```

--------------------------------

### IndexHNSW::search_level_0

Source: https://faiss.ai/cpp_api/file/IndexHNSW_8h.html

Performs a search only on level 0, given the starting points for each vertex.

```APIDOC
## search_level_0

### Description
Perform search only on level 0, given the starting points for each vertex.

### Parameters
* **n** (idx_t) - number of vectors
* **x** (const float *) - input vectors to search, size n * d
* **k** (idx_t) - number of nearest neighbors to search for
* **nearest** (const storage_idx_t *) - starting points for each vertex
* **nearest_d** (const float *) - distances to the starting points
* **distances** (float *) - output array for distances
* **labels** (idx_t *) - output array for labels
* **nprobe** (int) - number of probes (default: 1)
* **search_type** (int) - type of search (1: one search per nprobe, 2: enqueue all entry points) (default: 1)
* **params** (const SearchParameters *) - additional search parameters (default: nullptr)
```

--------------------------------

### Constructor

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1CodePackerFlat.html

Constructs a CodePackerFlat object with a specified code size.

```APIDOC
## Constructor

### Description
Initializes a new instance of the `CodePackerFlat` class.

### Signature
`explicit CodePackerFlat(size_t code_size)`

### Parameters
*   **code_size** (size_t) - The size of each code to be packed or unpacked.
```

--------------------------------

### Information and Code Size

Source: https://faiss.ai/cpp_api/struct/structfaiss_1_1IndexBinaryFromFloat.html

Methods to display index information and get the size of the produced codes.

```APIDOC
## Information and Code Size

### `void display() const`

Display the actual class name and some more info.

### `virtual size_t sa_code_size() const`

Size of the produced codes in bytes.
```
Tokens:
Raw