Vector Functions Reference

Vector Functions Reference

Preview — Firestore in Native mode (with Pipeline Operations) for Enterprise Edition

This feature is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the Service Specific Terms. You can process personal data for this feature as outlined in the Cloud Data Processing Addendum, subject to the obligations and restrictions described in the agreement under which you access Google Cloud. Pre-GA features are available "as is" and might have limited support. For more information, see the launch stage descriptions.

Vector Functions

Name	Description
COSINE_DISTANCE	Returns the cosine distance between two vectors
DOT_PRODUCT	Returns the dot product between two vectors
EUCLIDEAN_DISTANCE	Returns the euclidean distance between two vectors
MANHATTAN_DISTANCE	Returns the manhattan distance between two vectors
VECTOR_LENGTH	Returns the number of elements in a vector

COSINE_DISTANCE

Syntax:

cosine_distance(x: VECTOR, y: VECTOR) -> FLOAT64

Description:

Returns the cosine distance between x and y.

Node.js

const sampleVector = [0.0, 1, 2, 3, 4, 5];
const result = await db.pipeline()
  .collection("books")
  .select(
    field("embedding").cosineDistance(sampleVector).as("cosineDistance")
  )
  .execute();
test.firestore.js

Web

const sampleVector = [0.0, 1, 2, 3, 4, 5];
const result = await execute(db.pipeline()
  .collection("books")
  .select(
    field("embedding").cosineDistance(sampleVector).as("cosineDistance")));
test.firestore.js

Swift

let sampleVector = [0.0, 1, 2, 3, 4, 5]
let result = try await db.pipeline()
  .collection("books")
  .select([
    Field("embedding").cosineDistance(sampleVector).as("cosineDistance")
  ])
  .execute()
PipelineSnippets.swift

Kotlin
Android

val sampleVector = doubleArrayOf(0.0, 1.0, 2.0, 3.0, 4.0, 5.0)
val result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").cosineDistance(sampleVector).alias("cosineDistance")
    )
    .execute()
DocSnippets.kt

Java
Android

double[] sampleVector = {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
Task<Pipeline.Snapshot> result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").cosineDistance(sampleVector).alias("cosineDistance")
    )
    .execute();
DocSnippets.java

Python

from google.cloud.firestore_v1.pipeline_expressions import Field
from google.cloud.firestore_v1.vector import Vector

sample_vector = Vector([0.0, 1.0, 2.0, 3.0, 4.0, 5.0])
result = (
    client.pipeline()
    .collection("books")
    .select(
        Field.of("embedding").cosine_distance(sample_vector).as_("cosineDistance")
    )
    .execute()
)
firestore_pipelines.py

Java

double[] sampleVector = new double[] {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
Pipeline.Snapshot result =
    firestore
        .pipeline()
        .collection("books")
        .select(cosineDistance(field("embedding"), sampleVector).as("cosineDistance"))
        .execute()
        .get();
PipelineSnippets.java

DOT_PRODUCT

Syntax:

dot_product(x: VECTOR, y: VECTOR) -> FLOAT64

Description:

Returns the dot product of x and y.

Node.js

const sampleVector = [0.0, 1, 2, 3, 4, 5];
const result = await db.pipeline()
  .collection("books")
  .select(
    field("embedding").dotProduct(sampleVector).as("dotProduct")
  )
  .execute();
test.firestore.js

Web

const sampleVector = [0.0, 1, 2, 3, 4, 5];
const result = await execute(db.pipeline()
  .collection("books")
  .select(
    field("embedding").dotProduct(sampleVector).as("dotProduct")
  )
);
test.firestore.js

Swift

let sampleVector = [0.0, 1, 2, 3, 4, 5]
let result = try await db.pipeline()
  .collection("books")
  .select([
    Field("embedding").dotProduct(sampleVector).as("dotProduct")
  ])
  .execute()
PipelineSnippets.swift

Kotlin
Android

val sampleVector = doubleArrayOf(0.0, 1.0, 2.0, 3.0, 4.0, 5.0)
val result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").dotProduct(sampleVector).alias("dotProduct")
    )
    .execute()
DocSnippets.kt

Java
Android

double[] sampleVector = {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
Task<Pipeline.Snapshot> result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").dotProduct(sampleVector).alias("dotProduct")
    )
    .execute();
DocSnippets.java

Python

from google.cloud.firestore_v1.pipeline_expressions import Field
from google.cloud.firestore_v1.vector import Vector

sample_vector = Vector([0.0, 1.0, 2.0, 3.0, 4.0, 5.0])
result = (
    client.pipeline()
    .collection("books")
    .select(Field.of("embedding").dot_product(sample_vector).as_("dotProduct"))
    .execute()
)
firestore_pipelines.py

Java

double[] sampleVector = new double[] {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
Pipeline.Snapshot result =
    firestore
        .pipeline()
        .collection("books")
        .select(dotProduct(field("embedding"), sampleVector).as("dotProduct"))
        .execute()
        .get();
PipelineSnippets.java

EUCLIDEAN_DISTANCE

Syntax:

euclidean_distance(x: VECTOR, y: VECTOR) -> FLOAT64

Description:

Computes the euclidean distance between x and y.

Node.js

const sampleVector = [0.0, 1, 2, 3, 4, 5];
const result = await db.pipeline()
  .collection("books")
  .select(
    field("embedding").euclideanDistance(sampleVector).as("euclideanDistance")
  )
  .execute();
test.firestore.js

Web

const sampleVector = [0.0, 1, 2, 3, 4, 5];
const result = await execute(db.pipeline()
  .collection("books")
  .select(
    field("embedding").euclideanDistance(sampleVector).as("euclideanDistance")
  )
);
test.firestore.js

Swift

let sampleVector = [0.0, 1, 2, 3, 4, 5]
let result = try await db.pipeline()
  .collection("books")
  .select([
    Field("embedding").euclideanDistance(sampleVector).as("euclideanDistance")
  ])
  .execute()
PipelineSnippets.swift

Kotlin
Android

val sampleVector = doubleArrayOf(0.0, 1.0, 2.0, 3.0, 4.0, 5.0)
val result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").euclideanDistance(sampleVector).alias("euclideanDistance")
    )
    .execute()
DocSnippets.kt

Java
Android

double[] sampleVector = {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
Task<Pipeline.Snapshot> result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").euclideanDistance(sampleVector).alias("euclideanDistance")
    )
    .execute();
DocSnippets.java

Python

from google.cloud.firestore_v1.pipeline_expressions import Field
from google.cloud.firestore_v1.vector import Vector

sample_vector = Vector([0.0, 1.0, 2.0, 3.0, 4.0, 5.0])
result = (
    client.pipeline()
    .collection("books")
    .select(
        Field.of("embedding")
        .euclidean_distance(sample_vector)
        .as_("euclideanDistance")
    )
    .execute()
)
firestore_pipelines.py

Java

double[] sampleVector = new double[] {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
Pipeline.Snapshot result =
    firestore
        .pipeline()
        .collection("books")
        .select(euclideanDistance(field("embedding"), sampleVector).as("euclideanDistance"))
        .execute()
        .get();
PipelineSnippets.java

MANHATTAN_DISTANCE

Syntax:

manhattan_distance(x: VECTOR, y: VECTOR) -> FLOAT64

Description:

Computes the manhattan distance between x and y.

VECTOR_LENGTH

Syntax:

vector_length(vector: VECTOR) -> INT64

Description:

Returns the number of elements in a VECTOR.

Node.js

const result = await db.pipeline()
  .collection("books")
  .select(
    field("embedding").vectorLength().as("vectorLength")
  )
  .execute();
test.firestore.js

Web

const result = await execute(db.pipeline()
  .collection("books")
  .select(
    field("embedding").vectorLength().as("vectorLength")
  )
);
test.firestore.js

Swift

let result = try await db.pipeline()
  .collection("books")
  .select([
    Field("embedding").vectorLength().as("vectorLength")
  ])
  .execute()
PipelineSnippets.swift

Kotlin
Android

val result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").vectorLength().alias("vectorLength")
    )
    .execute()
DocSnippets.kt

Java
Android

Task<Pipeline.Snapshot> result = db.pipeline()
    .collection("books")
    .select(
        field("embedding").vectorLength().alias("vectorLength")
    )
    .execute();
DocSnippets.java

Python

from google.cloud.firestore_v1.pipeline_expressions import Field

result = (
    client.pipeline()
    .collection("books")
    .select(Field.of("embedding").vector_length().as_("vectorLength"))
    .execute()
)
firestore_pipelines.py

Java

Pipeline.Snapshot result =
    firestore
        .pipeline()
        .collection("books")
        .select(vectorLength(field("embedding")).as("vectorLength"))
        .execute()
        .get();
PipelineSnippets.java

What's next

• See the Pipeline Queries overview