Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/ankane/pgvector

Open-source vector similarity search for Postgres
https://github.com/ankane/pgvector

approximate-nearest-neighbor-search nearest-neighbor-search

Last synced: 3 months ago
JSON representation

Open-source vector similarity search for Postgres

Awesome Lists containing this project

README 

        
# pgvector



Open-source vector similarity search for Postgres



Store your vectors with the rest of your data. Supports:



- exact and approximate nearest neighbor search

- single-precision, half-precision, binary, and sparse vectors

- L2 distance, inner product, cosine distance, L1 distance, Hamming distance, and Jaccard distance

- any [language](#languages) with a Postgres client



Plus [ACID](https://en.wikipedia.org/wiki/ACID) compliance, point-in-time recovery, JOINs, and all of the other [great features](https://www.postgresql.org/about/) of Postgres



[![Build Status](https://github.com/pgvector/pgvector/actions/workflows/build.yml/badge.svg)](https://github.com/pgvector/pgvector/actions)



## Installation



### Linux and Mac



Compile and install the extension (supports Postgres 12+)



```sh

cd /tmp

git clone --branch v0.7.3 https://github.com/pgvector/pgvector.git

cd pgvector

make

make install # may need sudo

```



See the [installation notes](#installation-notes---linux-and-mac) if you run into issues



You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#pgxn), [APT](#apt), [Yum](#yum), [pkg](#pkg), or [conda-forge](#conda-forge), and it comes preinstalled with [Postgres.app](#postgresapp) and many [hosted providers](#hosted-postgres). There are also instructions for [GitHub Actions](https://github.com/pgvector/setup-pgvector).



### Windows



Ensure [C++ support in Visual Studio](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#download-and-install-the-tools) is installed, and run:



```cmd

call "C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvars64.bat"

```



Note: The exact path will vary depending on your Visual Studio version and edition



Then use `nmake` to build:



```cmd

set "PGROOT=C:\Program Files\PostgreSQL\16"

cd %TEMP%

git clone --branch v0.7.3 https://github.com/pgvector/pgvector.git

cd pgvector

nmake /F Makefile.win

nmake /F Makefile.win install

```



See the [installation notes](#installation-notes---windows) if you run into issues



You can also install it with [Docker](#docker) or [conda-forge](#conda-forge).



## Getting Started



Enable the extension (do this once in each database where you want to use it)



```tsql

CREATE EXTENSION vector;

```



Create a vector column with 3 dimensions



```sql

CREATE TABLE items (id bigserial PRIMARY KEY, embedding vector(3));

```



Insert vectors



```sql

INSERT INTO items (embedding) VALUES ('[1,2,3]'), ('[4,5,6]');

```



Get the nearest neighbors by L2 distance



```sql

SELECT * FROM items ORDER BY embedding <-> '[3,1,2]' LIMIT 5;

```



Also supports inner product (`<#>`), cosine distance (`<=>`), and L1 distance (`<+>`, added in 0.7.0)



Note: `<#>` returns the negative inner product since Postgres only supports `ASC` order index scans on operators



## Storing



Create a new table with a vector column



```sql

CREATE TABLE items (id bigserial PRIMARY KEY, embedding vector(3));

```



Or add a vector column to an existing table



```sql

ALTER TABLE items ADD COLUMN embedding vector(3);

```



Insert vectors



```sql

INSERT INTO items (embedding) VALUES ('[1,2,3]'), ('[4,5,6]');

```



Or load vectors in bulk using `COPY` ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/bulk_loading.py))



```sql

COPY items (embedding) FROM STDIN WITH (FORMAT BINARY);

```



Upsert vectors



```sql

INSERT INTO items (id, embedding) VALUES (1, '[1,2,3]'), (2, '[4,5,6]')

    ON CONFLICT (id) DO UPDATE SET embedding = EXCLUDED.embedding;

```



Update vectors



```sql

UPDATE items SET embedding = '[1,2,3]' WHERE id = 1;

```



Delete vectors



```sql

DELETE FROM items WHERE id = 1;

```



## Querying



Get the nearest neighbors to a vector



```sql

SELECT * FROM items ORDER BY embedding <-> '[3,1,2]' LIMIT 5;

```



Supported distance functions are:



- `<->` - L2 distance

- `<#>` - (negative) inner product

- `<=>` - cosine distance

- `<+>` - L1 distance (added in 0.7.0)



Get the nearest neighbors to a row



```sql

SELECT * FROM items WHERE id != 1 ORDER BY embedding <-> (SELECT embedding FROM items WHERE id = 1) LIMIT 5;

```



Get rows within a certain distance



```sql

SELECT * FROM items WHERE embedding <-> '[3,1,2]' < 5;

```



Note: Combine with `ORDER BY` and `LIMIT` to use an index



#### Distances



Get the distance



```sql

SELECT embedding <-> '[3,1,2]' AS distance FROM items;

```



For inner product, multiply by -1 (since `<#>` returns the negative inner product)



```tsql

SELECT (embedding <#> '[3,1,2]') * -1 AS inner_product FROM items;

```



For cosine similarity, use 1 - cosine distance



```sql

SELECT 1 - (embedding <=> '[3,1,2]') AS cosine_similarity FROM items;

```



#### Aggregates



Average vectors



```sql

SELECT AVG(embedding) FROM items;

```



Average groups of vectors



```sql

SELECT category_id, AVG(embedding) FROM items GROUP BY category_id;

```



## Indexing



By default, pgvector performs exact nearest neighbor search, which provides perfect recall.



You can add an index to use approximate nearest neighbor search, which trades some recall for speed. Unlike typical indexes, you will see different results for queries after adding an approximate index.



Supported index types are:



- [HNSW](#hnsw) - added in 0.5.0

- [IVFFlat](#ivfflat)



## HNSW



An HNSW index creates a multilayer graph. It has better query performance than IVFFlat (in terms of speed-recall tradeoff), but has slower build times and uses more memory. Also, an index can be created without any data in the table since there isn’t a training step like IVFFlat.



Add an index for each distance function you want to use.



L2 distance



```sql

CREATE INDEX ON items USING hnsw (embedding vector_l2_ops);

```



Note: Use `halfvec_l2_ops` for `halfvec` and `sparsevec_l2_ops` for `sparsevec` (and similar with the other distance functions)



Inner product



```sql

CREATE INDEX ON items USING hnsw (embedding vector_ip_ops);

```



Cosine distance



```sql

CREATE INDEX ON items USING hnsw (embedding vector_cosine_ops);

```



L1 distance - added in 0.7.0



```sql

CREATE INDEX ON items USING hnsw (embedding vector_l1_ops);

```



Hamming distance - added in 0.7.0



```sql

CREATE INDEX ON items USING hnsw (embedding bit_hamming_ops);

```



Jaccard distance - added in 0.7.0



```sql

CREATE INDEX ON items USING hnsw (embedding bit_jaccard_ops);

```



Supported types are:



- `vector` - up to 2,000 dimensions

- `halfvec` - up to 4,000 dimensions (added in 0.7.0)

- `bit` - up to 64,000 dimensions (added in 0.7.0)

- `sparsevec` - up to 1,000 non-zero elements (added in 0.7.0)



### Index Options



Specify HNSW parameters



- `m` - the max number of connections per layer (16 by default)

- `ef_construction` - the size of the dynamic candidate list for constructing the graph (64 by default)



```sql

CREATE INDEX ON items USING hnsw (embedding vector_l2_ops) WITH (m = 16, ef_construction = 64);

```



A higher value of `ef_construction` provides better recall at the cost of index build time / insert speed.



### Query Options



Specify the size of the dynamic candidate list for search (40 by default)



```sql

SET hnsw.ef_search = 100;

```



A higher value provides better recall at the cost of speed.



Use `SET LOCAL` inside a transaction to set it for a single query



```sql

BEGIN;

SET LOCAL hnsw.ef_search = 100;

SELECT ...

COMMIT;

```



### Index Build Time



Indexes build significantly faster when the graph fits into `maintenance_work_mem`



```sql

SET maintenance_work_mem = '8GB';

```



A notice is shown when the graph no longer fits



```text

NOTICE:  hnsw graph no longer fits into maintenance_work_mem after 100000 tuples

DETAIL:  Building will take significantly more time.

HINT:  Increase maintenance_work_mem to speed up builds.

```



Note: Do not set `maintenance_work_mem` so high that it exhausts the memory on the server



Like other index types, it’s faster to create an index after loading your initial data



Starting with 0.6.0, you can also speed up index creation by increasing the number of parallel workers (2 by default)



```sql

SET max_parallel_maintenance_workers = 7; -- plus leader

```



For a large number of workers, you may also need to increase `max_parallel_workers` (8 by default)



### Indexing Progress



Check [indexing progress](https://www.postgresql.org/docs/current/progress-reporting.html#CREATE-INDEX-PROGRESS-REPORTING) with Postgres 12+



```sql

SELECT phase, round(100.0 * blocks_done / nullif(blocks_total, 0), 1) AS "%" FROM pg_stat_progress_create_index;

```



The phases for HNSW are:



1. `initializing`

2. `loading tuples`



## IVFFlat



An IVFFlat index divides vectors into lists, and then searches a subset of those lists that are closest to the query vector. It has faster build times and uses less memory than HNSW, but has lower query performance (in terms of speed-recall tradeoff).



Three keys to achieving good recall are:



1. Create the index *after* the table has some data

2. Choose an appropriate number of lists - a good place to start is `rows / 1000` for up to 1M rows and `sqrt(rows)` for over 1M rows

3. When querying, specify an appropriate number of [probes](#query-options) (higher is better for recall, lower is better for speed) - a good place to start is `sqrt(lists)`



Add an index for each distance function you want to use.



L2 distance



```sql

CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100);

```



Note: Use `halfvec_l2_ops` for `halfvec` (and similar with the other distance functions)



Inner product



```sql

CREATE INDEX ON items USING ivfflat (embedding vector_ip_ops) WITH (lists = 100);

```



Cosine distance



```sql

CREATE INDEX ON items USING ivfflat (embedding vector_cosine_ops) WITH (lists = 100);

```



Hamming distance - added in 0.7.0



```sql

CREATE INDEX ON items USING ivfflat (embedding bit_hamming_ops) WITH (lists = 100);

```



Supported types are:



- `vector` - up to 2,000 dimensions

- `halfvec` - up to 4,000 dimensions (added in 0.7.0)

- `bit` - up to 64,000 dimensions (added in 0.7.0)



### Query Options



Specify the number of probes (1 by default)



```sql

SET ivfflat.probes = 10;

```



A higher value provides better recall at the cost of speed, and it can be set to the number of lists for exact nearest neighbor search (at which point the planner won’t use the index)



Use `SET LOCAL` inside a transaction to set it for a single query



```sql

BEGIN;

SET LOCAL ivfflat.probes = 10;

SELECT ...

COMMIT;

```



### Index Build Time



Speed up index creation on large tables by increasing the number of parallel workers (2 by default)



```sql

SET max_parallel_maintenance_workers = 7; -- plus leader

```



For a large number of workers, you may also need to increase `max_parallel_workers` (8 by default)



### Indexing Progress



Check [indexing progress](https://www.postgresql.org/docs/current/progress-reporting.html#CREATE-INDEX-PROGRESS-REPORTING) with Postgres 12+



```sql

SELECT phase, round(100.0 * tuples_done / nullif(tuples_total, 0), 1) AS "%" FROM pg_stat_progress_create_index;

```



The phases for IVFFlat are:



1. `initializing`

2. `performing k-means`

3. `assigning tuples`

4. `loading tuples`



Note: `%` is only populated during the `loading tuples` phase



## Filtering



There are a few ways to index nearest neighbor queries with a `WHERE` clause



```sql

SELECT * FROM items WHERE category_id = 123 ORDER BY embedding <-> '[3,1,2]' LIMIT 5;

```



Create an index on one [or more](https://www.postgresql.org/docs/current/indexes-multicolumn.html) of the `WHERE` columns for exact search



```sql

CREATE INDEX ON items (category_id);

```



Or a [partial index](https://www.postgresql.org/docs/current/indexes-partial.html) on the vector column for approximate search



```sql

CREATE INDEX ON items USING hnsw (embedding vector_l2_ops) WHERE (category_id = 123);

```



Use [partitioning](https://www.postgresql.org/docs/current/ddl-partitioning.html) for approximate search on many different values of the `WHERE` columns



```sql

CREATE TABLE items (embedding vector(3), category_id int) PARTITION BY LIST(category_id);

```



## Half-Precision Vectors



*Added in 0.7.0*



Use the `halfvec` type to store half-precision vectors



```sql

CREATE TABLE items (id bigserial PRIMARY KEY, embedding halfvec(3));

```



## Half-Precision Indexing



*Added in 0.7.0*



Index vectors at half precision for smaller indexes



```sql

CREATE INDEX ON items USING hnsw ((embedding::halfvec(3)) halfvec_l2_ops);

```



Get the nearest neighbors



```sql

SELECT * FROM items ORDER BY embedding::halfvec(3) <-> '[1,2,3]' LIMIT 5;

```



## Binary Vectors



Use the `bit` type to store binary vectors ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/hash_image_search.py))



```sql

CREATE TABLE items (id bigserial PRIMARY KEY, embedding bit(3));

INSERT INTO items (embedding) VALUES ('000'), ('111');

```



Get the nearest neighbors by Hamming distance (added in 0.7.0)



```sql

SELECT * FROM items ORDER BY embedding <~> '101' LIMIT 5;

```



Or (before 0.7.0)



```sql

SELECT * FROM items ORDER BY bit_count(embedding # '101') LIMIT 5;

```



Also supports Jaccard distance (`<%>`)



## Binary Quantization



*Added in 0.7.0*



Use expression indexing for binary quantization



```sql

CREATE INDEX ON items USING hnsw ((binary_quantize(embedding)::bit(3)) bit_hamming_ops);

```



Get the nearest neighbors by Hamming distance



```sql

SELECT * FROM items ORDER BY binary_quantize(embedding)::bit(3) <~> binary_quantize('[1,-2,3]') LIMIT 5;

```



Re-rank by the original vectors for better recall



```sql

SELECT * FROM (

    SELECT * FROM items ORDER BY binary_quantize(embedding)::bit(3) <~> binary_quantize('[1,-2,3]') LIMIT 20

) ORDER BY embedding <=> '[1,-2,3]' LIMIT 5;

```



## Sparse Vectors



*Added in 0.7.0*



Use the `sparsevec` type to store sparse vectors



```sql

CREATE TABLE items (id bigserial PRIMARY KEY, embedding sparsevec(5));

```



Insert vectors



```sql

INSERT INTO items (embedding) VALUES ('{1:1,3:2,5:3}/5'), ('{1:4,3:5,5:6}/5');

```



The format is `{index1:value1,index2:value2}/dimensions` and indices start at 1 like SQL arrays



Get the nearest neighbors by L2 distance



```sql

SELECT * FROM items ORDER BY embedding <-> '{1:3,3:1,5:2}/5' LIMIT 5;

```



## Hybrid Search



Use together with Postgres [full-text search](https://www.postgresql.org/docs/current/textsearch-intro.html) for hybrid search.



```sql

SELECT id, content FROM items, plainto_tsquery('hello search') query

    WHERE textsearch @@ query ORDER BY ts_rank_cd(textsearch, query) DESC LIMIT 5;

```



You can use [Reciprocal Rank Fusion](https://github.com/pgvector/pgvector-python/blob/master/examples/hybrid_search_rrf.py) or a [cross-encoder](https://github.com/pgvector/pgvector-python/blob/master/examples/hybrid_search.py) to combine results.



## Indexing Subvectors



*Added in 0.7.0*



Use expression indexing to index subvectors



```sql

CREATE INDEX ON items USING hnsw ((subvector(embedding, 1, 3)::vector(3)) vector_cosine_ops);

```



Get the nearest neighbors by cosine distance



```sql

SELECT * FROM items ORDER BY subvector(embedding, 1, 3)::vector(3) <=> subvector('[1,2,3,4,5]'::vector, 1, 3) LIMIT 5;

```



Re-rank by the full vectors for better recall



```sql

SELECT * FROM (

    SELECT * FROM items ORDER BY subvector(embedding, 1, 3)::vector(3) <=> subvector('[1,2,3,4,5]'::vector, 1, 3) LIMIT 20

) ORDER BY embedding <=> '[1,2,3,4,5]' LIMIT 5;

```



## Performance



### Tuning



Use a tool like [PgTune](https://pgtune.leopard.in.ua/) to set initial values for Postgres server parameters. For instance, `shared_buffers` should typically be 25% of the server’s memory. You can find the config file with:



```sql

SHOW config_file;

```



And check individual settings with:



```sql

SHOW shared_buffers;

```



Be sure to restart Postgres for changes to take effect.



### Loading



Use `COPY` for bulk loading data ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/bulk_loading.py)).



```sql

COPY items (embedding) FROM STDIN WITH (FORMAT BINARY);

```



Add any indexes *after* loading the initial data for best performance.



### Indexing



See index build time for [HNSW](#index-build-time) and [IVFFlat](#index-build-time-1).



In production environments, create indexes concurrently to avoid blocking writes.



```sql

CREATE INDEX CONCURRENTLY ...

```



### Querying



Use `EXPLAIN ANALYZE` to debug performance.



```sql

EXPLAIN ANALYZE SELECT * FROM items ORDER BY embedding <-> '[3,1,2]' LIMIT 5;

```



#### Exact Search



To speed up queries without an index, increase `max_parallel_workers_per_gather`.



```sql

SET max_parallel_workers_per_gather = 4;

```



If vectors are normalized to length 1 (like [OpenAI embeddings](https://platform.openai.com/docs/guides/embeddings/which-distance-function-should-i-use)), use inner product for best performance.



```tsql

SELECT * FROM items ORDER BY embedding <#> '[3,1,2]' LIMIT 5;

```



#### Approximate Search



To speed up queries with an IVFFlat index, increase the number of inverted lists (at the expense of recall).



```sql

CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 1000);

```



### Vacuuming



Vacuuming can take a while for HNSW indexes. Speed it up by reindexing first.



```sql

REINDEX INDEX CONCURRENTLY index_name;

VACUUM table_name;

```



## Monitoring



Monitor performance with [pg_stat_statements](https://www.postgresql.org/docs/current/pgstatstatements.html) (be sure to add it to `shared_preload_libraries`).



```sql

CREATE EXTENSION pg_stat_statements;

```



Get the most time-consuming queries with:



```sql

SELECT query, calls, ROUND((total_plan_time + total_exec_time) / calls) AS avg_time_ms,

    ROUND((total_plan_time + total_exec_time) / 60000) AS total_time_min

    FROM pg_stat_statements ORDER BY total_plan_time + total_exec_time DESC LIMIT 20;

```



Note: Replace `total_plan_time + total_exec_time` with `total_time` for Postgres < 13



Monitor recall by comparing results from approximate search with exact search.



```sql

BEGIN;

SET LOCAL enable_indexscan = off; -- use exact search

SELECT ...

COMMIT;

```



## Scaling



Scale pgvector the same way you scale Postgres.



Scale vertically by increasing memory, CPU, and storage on a single instance. Use existing tools to [tune parameters](#tuning) and [monitor performance](#monitoring).



Scale horizontally with [replicas](https://www.postgresql.org/docs/current/hot-standby.html), or use [Citus](https://github.com/citusdata/citus) or another approach for sharding ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/citus.py)).



## Languages



Use pgvector from any language with a Postgres client. You can even generate and store vectors in one language and query them in another.



Language | Libraries / Examples

--- | ---

C | [pgvector-c](https://github.com/pgvector/pgvector-c)

C++ | [pgvector-cpp](https://github.com/pgvector/pgvector-cpp)

C#, F#, Visual Basic | [pgvector-dotnet](https://github.com/pgvector/pgvector-dotnet)

Crystal | [pgvector-crystal](https://github.com/pgvector/pgvector-crystal)

Dart | [pgvector-dart](https://github.com/pgvector/pgvector-dart)

Elixir | [pgvector-elixir](https://github.com/pgvector/pgvector-elixir)

Go | [pgvector-go](https://github.com/pgvector/pgvector-go)

Haskell | [pgvector-haskell](https://github.com/pgvector/pgvector-haskell)

Java, Kotlin, Groovy, Scala | [pgvector-java](https://github.com/pgvector/pgvector-java)

JavaScript, TypeScript | [pgvector-node](https://github.com/pgvector/pgvector-node)

Julia | [pgvector-julia](https://github.com/pgvector/pgvector-julia)

Lisp | [pgvector-lisp](https://github.com/pgvector/pgvector-lisp)

Lua | [pgvector-lua](https://github.com/pgvector/pgvector-lua)

Nim | [pgvector-nim](https://github.com/pgvector/pgvector-nim)

OCaml | [pgvector-ocaml](https://github.com/pgvector/pgvector-ocaml)

Perl | [pgvector-perl](https://github.com/pgvector/pgvector-perl)

PHP | [pgvector-php](https://github.com/pgvector/pgvector-php)

Python | [pgvector-python](https://github.com/pgvector/pgvector-python)

R | [pgvector-r](https://github.com/pgvector/pgvector-r)

Ruby | [pgvector-ruby](https://github.com/pgvector/pgvector-ruby), [Neighbor](https://github.com/ankane/neighbor)

Rust | [pgvector-rust](https://github.com/pgvector/pgvector-rust)

Swift | [pgvector-swift](https://github.com/pgvector/pgvector-swift)

Zig | [pgvector-zig](https://github.com/pgvector/pgvector-zig)



## Frequently Asked Questions



#### How many vectors can be stored in a single table?



A non-partitioned table has a limit of 32 TB by default in Postgres. A partitioned table can have thousands of partitions of that size.



#### Is replication supported?



Yes, pgvector uses the write-ahead log (WAL), which allows for replication and point-in-time recovery.



#### What if I want to index vectors with more than 2,000 dimensions?



You can use [half-precision indexing](#half-precision-indexing) to index up to 4,000 dimensions or [binary quantization](#binary-quantization) to index up to 64,000 dimensions. Another option is [dimensionality reduction](https://en.wikipedia.org/wiki/Dimensionality_reduction).



#### Can I store vectors with different dimensions in the same column?



You can use `vector` as the type (instead of `vector(3)`).



```sql

CREATE TABLE embeddings (model_id bigint, item_id bigint, embedding vector, PRIMARY KEY (model_id, item_id));

```



However, you can only create indexes on rows with the same number of dimensions (using [expression](https://www.postgresql.org/docs/current/indexes-expressional.html) and [partial](https://www.postgresql.org/docs/current/indexes-partial.html) indexing):



```sql

CREATE INDEX ON embeddings USING hnsw ((embedding::vector(3)) vector_l2_ops) WHERE (model_id = 123);

```



and query with:



```sql

SELECT * FROM embeddings WHERE model_id = 123 ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;

```



#### Can I store vectors with more precision?



You can use the `double precision[]` or `numeric[]` type to store vectors with more precision.



```sql

CREATE TABLE items (id bigserial PRIMARY KEY, embedding double precision[]);



-- use {} instead of [] for Postgres arrays

INSERT INTO items (embedding) VALUES ('{1,2,3}'), ('{4,5,6}');

```



Optionally, add a [check constraint](https://www.postgresql.org/docs/current/ddl-constraints.html) to ensure data can be converted to the `vector` type and has the expected dimensions.



```sql

ALTER TABLE items ADD CHECK (vector_dims(embedding::vector) = 3);

```



Use [expression indexing](https://www.postgresql.org/docs/current/indexes-expressional.html) to index (at a lower precision):



```sql

CREATE INDEX ON items USING hnsw ((embedding::vector(3)) vector_l2_ops);

```



and query with:



```sql

SELECT * FROM items ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;

```



#### Do indexes need to fit into memory?



No, but like other index types, you’ll likely see better performance if they do. You can get the size of an index with:



```sql

SELECT pg_size_pretty(pg_relation_size('index_name'));

```



## Troubleshooting



#### Why isn’t a query using an index?



The query needs to have an `ORDER BY` and `LIMIT`, and the `ORDER BY` must be the result of a distance operator (not an expression) in ascending order.



```sql

-- index

ORDER BY embedding <=> '[3,1,2]' LIMIT 5;



-- no index

ORDER BY 1 - (embedding <=> '[3,1,2]') DESC LIMIT 5;

```



You can encourage the planner to use an index for a query with:



```sql

BEGIN;

SET LOCAL enable_seqscan = off;

SELECT ...

COMMIT;

```



Also, if the table is small, a table scan may be faster.



#### Why isn’t a query using a parallel table scan?



The planner doesn’t consider [out-of-line storage](https://www.postgresql.org/docs/current/storage-toast.html) in cost estimates, which can make a serial scan look cheaper. You can reduce the cost of a parallel scan for a query with:



```sql

BEGIN;

SET LOCAL min_parallel_table_scan_size = 1;

SET LOCAL parallel_setup_cost = 1;

SELECT ...

COMMIT;

```



or choose to store vectors inline:



```sql

ALTER TABLE items ALTER COLUMN embedding SET STORAGE PLAIN;

```



#### Why are there less results for a query after adding an HNSW index?



Results are limited by the size of the dynamic candidate list (`hnsw.ef_search`). There may be even less results due to dead tuples or filtering conditions in the query. We recommend setting `hnsw.ef_search` to at least twice the `LIMIT` of the query. If you need more than 500 results, use an IVFFlat index instead.



Also, note that `NULL` vectors are not indexed (as well as zero vectors for cosine distance).



#### Why are there less results for a query after adding an IVFFlat index?



The index was likely created with too little data for the number of lists. Drop the index until the table has more data.



```sql

DROP INDEX index_name;

```



Results can also be limited by the number of probes (`ivfflat.probes`).



Also, note that `NULL` vectors are not indexed (as well as zero vectors for cosine distance).



## Reference



- [Vector](#vector-type)

- [Halfvec](#halfvec-type)

- [Bit](#bit-type)

- [Sparsevec](#sparsevec-type)



### Vector Type



Each vector takes `4 * dimensions + 8` bytes of storage. Each element is a single-precision floating-point number (like the `real` type in Postgres), and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Vectors can have up to 16,000 dimensions.



### Vector Operators



Operator | Description | Added

--- | --- | ---

\+ | element-wise addition |

\- | element-wise subtraction |

\* | element-wise multiplication | 0.5.0

\|\| | concatenate | 0.7.0

<-> | Euclidean distance |

<#> | negative inner product |

<=> | cosine distance |

<+> | taxicab distance | 0.7.0



### Vector Functions



Function | Description | Added

--- | --- | ---

binary_quantize(vector) → bit | binary quantize | 0.7.0

cosine_distance(vector, vector) → double precision | cosine distance |

inner_product(vector, vector) → double precision | inner product |

l1_distance(vector, vector) → double precision | taxicab distance | 0.5.0

l2_distance(vector, vector) → double precision | Euclidean distance |

l2_normalize(vector) → vector | Normalize with Euclidean norm | 0.7.0

subvector(vector, integer, integer) → vector | subvector | 0.7.0

vector_dims(vector) → integer | number of dimensions |

vector_norm(vector) → double precision | Euclidean norm |



### Vector Aggregate Functions



Function | Description | Added

--- | --- | ---

avg(vector) → vector | average |

sum(vector) → vector | sum | 0.5.0



### Halfvec Type



Each half vector takes `2 * dimensions + 8` bytes of storage. Each element is a half-precision floating-point number, and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Half vectors can have up to 16,000 dimensions.



### Halfvec Operators



Operator | Description | Added

--- | --- | ---

\+ | element-wise addition | 0.7.0

\- | element-wise subtraction | 0.7.0

\* | element-wise multiplication | 0.7.0

\|\| | concatenate | 0.7.0

<-> | Euclidean distance | 0.7.0

<#> | negative inner product | 0.7.0

<=> | cosine distance | 0.7.0

<+> | taxicab distance | 0.7.0



### Halfvec Functions



Function | Description | Added

--- | --- | ---

binary_quantize(halfvec) → bit | binary quantize | 0.7.0

cosine_distance(halfvec, halfvec) → double precision | cosine distance | 0.7.0

inner_product(halfvec, halfvec) → double precision | inner product | 0.7.0

l1_distance(halfvec, halfvec) → double precision | taxicab distance | 0.7.0

l2_distance(halfvec, halfvec) → double precision | Euclidean distance | 0.7.0

l2_norm(halfvec) → double precision | Euclidean norm | 0.7.0

l2_normalize(halfvec) → halfvec | Normalize with Euclidean norm | 0.7.0

subvector(halfvec, integer, integer) → halfvec | subvector | 0.7.0

vector_dims(halfvec) → integer | number of dimensions | 0.7.0



### Halfvec Aggregate Functions



Function | Description | Added

--- | --- | ---

avg(halfvec) → halfvec | average | 0.7.0

sum(halfvec) → halfvec | sum | 0.7.0



### Bit Type



Each bit vector takes `dimensions / 8 + 8` bytes of storage. See the [Postgres docs](https://www.postgresql.org/docs/current/datatype-bit.html) for more info.



### Bit Operators



Operator | Description | Added

--- | --- | ---

<~> | Hamming distance | 0.7.0

<%> | Jaccard distance | 0.7.0



### Bit Functions



Function | Description | Added

--- | --- | ---

hamming_distance(bit, bit) → double precision | Hamming distance | 0.7.0

jaccard_distance(bit, bit) → double precision | Jaccard distance | 0.7.0



### Sparsevec Type



Each sparse vector takes `8 * non-zero elements + 16` bytes of storage. Each element is a single-precision floating-point number, and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Sparse vectors can have up to 16,000 non-zero elements.



### Sparsevec Operators



Operator | Description | Added

--- | --- | ---

<-> | Euclidean distance | 0.7.0

<#> | negative inner product | 0.7.0

<=> | cosine distance | 0.7.0

<+> | taxicab distance | 0.7.0



### Sparsevec Functions



Function | Description | Added

--- | --- | ---

cosine_distance(sparsevec, sparsevec) → double precision | cosine distance | 0.7.0

inner_product(sparsevec, sparsevec) → double precision | inner product | 0.7.0

l1_distance(sparsevec, sparsevec) → double precision | taxicab distance | 0.7.0

l2_distance(sparsevec, sparsevec) → double precision | Euclidean distance | 0.7.0

l2_norm(sparsevec) → double precision | Euclidean norm | 0.7.0

l2_normalize(sparsevec) → sparsevec | Normalize with Euclidean norm | 0.7.0



## Installation Notes - Linux and Mac



### Postgres Location



If your machine has multiple Postgres installations, specify the path to [pg_config](https://www.postgresql.org/docs/current/app-pgconfig.html) with:



```sh

export PG_CONFIG=/Library/PostgreSQL/16/bin/pg_config

```



Then re-run the installation instructions (run `make clean` before `make` if needed). If `sudo` is needed for `make install`, use:



```sh

sudo --preserve-env=PG_CONFIG make install

```



A few common paths on Mac are:



- EDB installer - `/Library/PostgreSQL/16/bin/pg_config`

- Homebrew (arm64) - `/opt/homebrew/opt/postgresql@16/bin/pg_config`

- Homebrew (x86-64) - `/usr/local/opt/postgresql@16/bin/pg_config`



Note: Replace `16` with your Postgres server version



### Missing Header



If compilation fails with `fatal error: postgres.h: No such file or directory`, make sure Postgres development files are installed on the server.



For Ubuntu and Debian, use:



```sh

sudo apt install postgresql-server-dev-16

```



Note: Replace `16` with your Postgres server version



### Missing SDK



If compilation fails and the output includes `warning: no such sysroot directory` on Mac, reinstall Xcode Command Line Tools.



### Portability



By default, pgvector compiles with `-march=native` on some platforms for best performance. However, this can lead to `Illegal instruction` errors if trying to run the compiled extension on a different machine.



To compile for portability, use:



```sh

make OPTFLAGS=""

```



## Installation Notes - Windows



### Missing Header



If compilation fails with `Cannot open include file: 'postgres.h': No such file or directory`, make sure `PGROOT` is correct.



### Permissions



If installation fails with `Access is denied`, re-run the installation instructions as an administrator.



## Additional Installation Methods



### Docker



Get the [Docker image](https://hub.docker.com/r/pgvector/pgvector) with:



```sh

docker pull pgvector/pgvector:pg16

```



This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (replace `16` with your Postgres server version, and run it the same way).



You can also build the image manually:



```sh

git clone --branch v0.7.3 https://github.com/pgvector/pgvector.git

cd pgvector

docker build --pull --build-arg PG_MAJOR=16 -t myuser/pgvector .

```



### Homebrew



With Homebrew Postgres, you can use:



```sh

brew install pgvector

```



Note: This only adds it to the `postgresql@14` formula



### PGXN



Install from the [PostgreSQL Extension Network](https://pgxn.org/dist/vector) with:



```sh

pgxn install vector

```



### APT



Debian and Ubuntu packages are available from the [PostgreSQL APT Repository](https://wiki.postgresql.org/wiki/Apt). Follow the [setup instructions](https://wiki.postgresql.org/wiki/Apt#Quickstart) and run:



```sh

sudo apt install postgresql-16-pgvector

```



Note: Replace `16` with your Postgres server version



### Yum



RPM packages are available from the [PostgreSQL Yum Repository](https://yum.postgresql.org/). Follow the [setup instructions](https://www.postgresql.org/download/linux/redhat/) for your distribution and run:



```sh

sudo yum install pgvector_16

# or

sudo dnf install pgvector_16

```



Note: Replace `16` with your Postgres server version



### pkg



Install the FreeBSD package with:



```sh

pkg install postgresql15-pgvector

```



or the port with:



```sh

cd /usr/ports/databases/pgvector

make install

```



### conda-forge



With Conda Postgres, install from [conda-forge](https://anaconda.org/conda-forge/pgvector) with:



```sh

conda install -c conda-forge pgvector

```



This method is [community-maintained](https://github.com/conda-forge/pgvector-feedstock) by [@mmcauliffe](https://github.com/mmcauliffe)



### Postgres.app



Download the [latest release](https://postgresapp.com/downloads.html) with Postgres 15+.



## Hosted Postgres



pgvector is available on [these providers](https://github.com/pgvector/pgvector/issues/54).



## Upgrading



[Install](#installation) the latest version (use the same method as the original installation). Then in each database you want to upgrade, run:



```sql

ALTER EXTENSION vector UPDATE;

```



You can check the version in the current database with:



```sql

SELECT extversion FROM pg_extension WHERE extname = 'vector';

```



## Upgrade Notes



### 0.6.0



#### Postgres 12



If upgrading with Postgres 12, remove this line from `sql/vector--0.5.1--0.6.0.sql`:



```sql

ALTER TYPE vector SET (STORAGE = external);

```



Then run `make install` and `ALTER EXTENSION vector UPDATE;`.



#### Docker



The Docker image is now published in the `pgvector` org, and there are tags for each supported version of Postgres (rather than a `latest` tag).



```sh

docker pull pgvector/pgvector:pg16

# or

docker pull pgvector/pgvector:0.6.0-pg16

```



Also, if you’ve increased `maintenance_work_mem`, make sure `--shm-size` is at least that size to avoid an error with parallel HNSW index builds.



```sh

docker run --shm-size=1g ...

```



## Thanks



Thanks to:



- [PASE: PostgreSQL Ultra-High-Dimensional Approximate Nearest Neighbor Search Extension](https://dl.acm.org/doi/pdf/10.1145/3318464.3386131)

- [Faiss: A Library for Efficient Similarity Search and Clustering of Dense Vectors](https://github.com/facebookresearch/faiss)

- [Using the Triangle Inequality to Accelerate k-means](https://cdn.aaai.org/ICML/2003/ICML03-022.pdf)

- [k-means++: The Advantage of Careful Seeding](https://theory.stanford.edu/~sergei/papers/kMeansPP-soda.pdf)

- [Concept Decompositions for Large Sparse Text Data using Clustering](https://www.cs.utexas.edu/users/inderjit/public_papers/concept_mlj.pdf)

- [Efficient and Robust Approximate Nearest Neighbor Search using Hierarchical Navigable Small World Graphs](https://arxiv.org/ftp/arxiv/papers/1603/1603.09320.pdf)



## History



View the [changelog](https://github.com/pgvector/pgvector/blob/master/CHANGELOG.md)



## Contributing



Everyone is encouraged to help improve this project. Here are a few ways you can help:



- [Report bugs](https://github.com/pgvector/pgvector/issues)

- Fix bugs and [submit pull requests](https://github.com/pgvector/pgvector/pulls)

- Write, clarify, or fix documentation

- Suggest or add new features



To get started with development:



```sh

git clone https://github.com/pgvector/pgvector.git

cd pgvector

make

make install

```



To run all tests:



```sh

make installcheck        # regression tests

make prove_installcheck  # TAP tests

```



To run single tests:



```sh

make installcheck REGRESS=functions                            # regression test

make prove_installcheck PROVE_TESTS=test/t/001_ivfflat_wal.pl  # TAP test

```



To enable assertions:



```sh

make clean && PG_CFLAGS="-DUSE_ASSERT_CHECKING" make && make install

```



To enable benchmarking:



```sh

make clean && PG_CFLAGS="-DIVFFLAT_BENCH" make && make install

```



To show memory usage:



```sh

make clean && PG_CFLAGS="-DHNSW_MEMORY -DIVFFLAT_MEMORY" make && make install

```



To get k-means metrics:



```sh

make clean && PG_CFLAGS="-DIVFFLAT_KMEANS_DEBUG" make && make install

```



Resources for contributors



- [Extension Building Infrastructure](https://www.postgresql.org/docs/current/extend-pgxs.html)

- [Index Access Method Interface Definition](https://www.postgresql.org/docs/current/indexam.html)

- [Generic WAL Records](https://www.postgresql.org/docs/current/generic-wal.html)