Loading TOC...
JavaScript vec.precision

MarkLogic 12 Product Documentation
vec:precision

vec:precision(
   $vector as vec:vector?,
   [$precision as xs:integer?]
) as vec:vector

Summary

Returns a new vector which is a copy of the input vector with reduced precision. The precision reduction is achieved by clearing the bottom (32 - precision) bits of the mantissa for each dimension's float value. This can be useful for reducing storage requirements or for creating approximate vector representations.

Parameters
vector The input vector to reduce precision. Can be a vector or an empty sequence.
precision The number of mantissa bits to preserve (9-32 inclusive). Default is 16. Higher values preserve more precision. If the value is outside the valid range, throw VEC-INVALIDPRECISION.

Example

  xquery version "1.0-ml";
  let $vec := vec:vector((3.14159265, 2.71828182, 1.41421356))
  return vec:precision($vec, 16)

  => [ 3.14062, 2.70312, 1.41406 ]

Example

  xquery version "1.0-ml";
  let $vec := vec:vector((3.14159265, 2.71828182, 1.41421356))
  let $reduced := vec:precision($vec)
  return ($vec, $reduced)

  [ 3.14159, 2.71828, 1.41421 ]
  [ 3.14062, 2.70312, 1.41406 ]

Stack Overflow iconStack Overflow: Get the most useful answers to questions from the MarkLogic community, or ask your own question.