The ML.LP_NORM function

This document describes the ML.LP_NORM scalar function, which lets you compute the Lp norm for a vector, where p is the degree.

Syntax

ML.LP_NORM(vector, degree)

Arguments

ML.LP_NORM has the following arguments:

  • vector: an ARRAY<Numerical type> value that represents a vector, where Numerical type can be BIGNUMERIC, FLOAT64, INT64 or NUMERIC. For example ARRAY<BIGNUMERIC>.

    Each element of the array denotes one dimension of the vector. An example of a four-dimensional vector is [0.0, 1.0, 1.0, 0.0].

    The function calculates the p degree norm of the numerical type values in all the values in the array.

  • degree: a FLOAT64 value that specifies the degree. This can be 0.0, any value >= 1.0, or CAST('INF' AS FLOAT64) to return the L_infinity norm of the vector, which is the largest magnitude of the values in the vector.

    Commonly used values are 1.0 to calculate the Manhattan norm of the vector and 2.0 to calculate the Euclidean norm of the vector.

Output

ML.LP_NORM returns a FLOAT64 value that represents the Lp norm for the vector. Returns NULL if vector is NULL.

Example

The following example gets the Euclidean norm for vectors consisting of ARRAY<FLOAT64> values:

  1. Create the table t1:

    CREATE TABLE mydataset.t1
(
  v1 ARRAY<FLOAT64>,
  v2 ARRAY<FLOAT64>
)

  2. Populate t1:

    INSERT mydataset.t1 (v1,v2)
VALUES ([4.1,0.5,1.0], [3.0,0.0,2.5])

  3. Calculate the Euclidean norm for v1 and v2:

    SELECT v1, ML.LP_NORM(v1, 2.0) AS v1_norm, v2, ML.LP_NORM(v2, 2.0) AS v2_norm
FROM mydataset.t1;

    This query produces the following output:

    +---------------------------+-----+-------------------+
| v1  | v1_norm             | v2  | v2_norm           |
+---------------------------+-----+-------------------+
| 4.1 | 4.2497058721751557  | 3.0 | 3.905124837953327 |
+-----|                     |-----|                   |
| 0.5 |                     | 0.0 |                   |
+-----|                     |-----+                   |
| 1.0 |                     | 2.5 |                   |
+---------------------------+-----+-------------------+

What's next