iqdb-quantize 1.0.0

Vector quantization (scalar, product, binary) for memory-efficient vector search - part of the iQDB family.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

//! Shared input-validation helpers.
//!
//! Surface bad input as typed [`iqdb_types::IqdbError`] variants so the
//! quantizer hot paths never panic on empty, non-finite, or
//! dimension-mismatched inputs.

use iqdb_types::{IqdbError, Result};

/// Reject empty vectors and vectors with NaN / infinite components.
pub(crate) fn finite_non_empty(vector: &[f32]) -> Result<()> {
    if vector.is_empty() {
        return Err(IqdbError::InvalidVector);
    }
    if vector.iter().any(|v| !v.is_finite()) {
        return Err(IqdbError::InvalidVector);
    }
    Ok(())
}

/// Check that `actual == expected`, returning
/// [`IqdbError::DimensionMismatch`] when they disagree.
pub(crate) fn dim_eq(expected: usize, actual: usize) -> Result<()> {
    if actual != expected {
        return Err(IqdbError::DimensionMismatch {
            expected,
            found: actual,
        });
    }
    Ok(())
}

/// Validate a training set as a unit: non-empty, every vector valid, every
/// vector the same dimension. Returns the shared dimension on success.
pub(crate) fn training_set(vectors: &[&[f32]]) -> Result<usize> {
    let first = match vectors.first() {
        Some(first) => first,
        None => {
            return Err(IqdbError::InvalidConfig {
                reason: "quantizer training set is empty",
            });
        }
    };
    finite_non_empty(first)?;
    let dim = first.len();
    for v in &vectors[1..] {
        finite_non_empty(v)?;
        dim_eq(dim, v.len())?;
    }
    Ok(dim)
}