Module org.elasticsearch.simdvec

Package org.elasticsearch.simdvec

Class ES940OSQVectorsScorer

java.lang.Object

org.elasticsearch.simdvec.ES940OSQVectorsScorer

Direct Known Subclasses:

MemorySegmentES940OSQVectorsScorer

public class ES940OSQVectorsScorer
extends Object

Scorer for quantized vectors stored as an IndexInput.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	ES940OSQVectorsScorer.SymmetricInt4Encoding

Field Summary

Fields

Modifier and Type	Field	Description
protected final float[]	additionalCorrections
protected static final float[]	BIT_SCALES
static final int	BULK_SIZE
protected final int	bulkSize
protected final int	dimensions
protected final org.apache.lucene.store.IndexInput	in	The wrapper IndexInput.
protected final byte	indexBits
protected final ES940OSQVectorsScorer.SymmetricInt4Encoding	int4Encoding
protected final int	length
protected final float[]	lowerIntervals
protected final byte	queryBits
protected final int[]	targetComponentSums
protected final float[]	upperIntervals

Constructor Summary

Constructors

Constructor	Description
ES940OSQVectorsScorer(org.apache.lucene.store.IndexInput in, byte queryBits, byte indexBits, int dimensions, int dataLength)
ES940OSQVectorsScorer(org.apache.lucene.store.IndexInput in, byte queryBits, byte indexBits, int dimensions, int dataLength, int bulkSize)
ES940OSQVectorsScorer(org.apache.lucene.store.IndexInput in, byte queryBits, byte indexBits, int dimensions, int dataLength, int bulkSize, ES940OSQVectorsScorer.SymmetricInt4Encoding int4Encoding)

Method Summary

Modifier and Type	Method	Description
float	applyCorrectionsIndividually(float queryLowerInterval, float queryUpperInterval, int queryComponentSum, float queryAdditionalCorrection, org.apache.lucene.index.VectorSimilarityFunction similarityFunction, float centroidDp, float lowerInterval, float upperInterval, int targetComponentSum, float additionalCorrection, float qcDist)	Computes the final score by applying the necessary corrections to the provided quantized distance.
long	quantizeScore(byte[] q)	compute the quantize distance between the provided quantized query and the quantized vector that is read from the wrapped IndexInput.
void	quantizeScoreBulk(byte[] q, int count, float[] scores)	Compute the quantize distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput.
void	quantizeScoreBulkOffsets(byte[] q, int[] offsets, int offsetsCount, float[] scores, int count)	Compute the quantize distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput.
float	scoreBulk(byte[] q, float queryLowerInterval, float queryUpperInterval, int queryComponentSum, float queryAdditionalCorrection, org.apache.lucene.index.VectorSimilarityFunction similarityFunction, float centroidDp, float[] scores)	Bulk score overload; same as scoreBulk(byte[], float, float, int, float, VectorSimilarityFunction, float, float[], int) with bulkSize as the default bulkSize param value.
float	scoreBulk(byte[] q, float queryLowerInterval, float queryUpperInterval, int queryComponentSum, float queryAdditionalCorrection, org.apache.lucene.index.VectorSimilarityFunction similarityFunction, float centroidDp, float[] scores, int bulkSize)	Compute the distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput.
float	scoreBulkOffsets(byte[] q, float queryLowerInterval, float queryUpperInterval, int queryComponentSum, float queryAdditionalCorrection, org.apache.lucene.index.VectorSimilarityFunction similarityFunction, float centroidDp, int[] offsets, int offsetsCount, float[] scores, int count)	Compute the distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

BULK_SIZE

public static final int BULK_SIZE

See Also:

• Constant Field Values

BIT_SCALES

protected static final float[] BIT_SCALES

in

protected final org.apache.lucene.store.IndexInput in

The wrapper IndexInput.

queryBits

protected final byte queryBits

indexBits

protected final byte indexBits

length

protected final int length

dimensions

protected final int dimensions

bulkSize

protected final int bulkSize

int4Encoding

protected final ES940OSQVectorsScorer.SymmetricInt4Encoding int4Encoding

lowerIntervals

protected final float[] lowerIntervals

upperIntervals

protected final float[] upperIntervals

targetComponentSums

protected final int[] targetComponentSums

additionalCorrections

protected final float[] additionalCorrections

Constructor Details

ES940OSQVectorsScorer

public ES940OSQVectorsScorer(org.apache.lucene.store.IndexInput in,
 byte queryBits,
 byte indexBits,
 int dimensions,
 int dataLength,
 int bulkSize,
 ES940OSQVectorsScorer.SymmetricInt4Encoding int4Encoding)

ES940OSQVectorsScorer

public ES940OSQVectorsScorer(org.apache.lucene.store.IndexInput in,
 byte queryBits,
 byte indexBits,
 int dimensions,
 int dataLength,
 int bulkSize)

ES940OSQVectorsScorer

public ES940OSQVectorsScorer(org.apache.lucene.store.IndexInput in,
 byte queryBits,
 byte indexBits,
 int dimensions,
 int dataLength)

Method Details

quantizeScore

public long quantizeScore(byte[] q)
                   throws IOException

compute the quantize distance between the provided quantized query and the quantized vector that is read from the wrapped IndexInput.

Throws:

IOException

quantizeScoreBulk

public void quantizeScoreBulk(byte[] q,
 int count,
 float[] scores)
                       throws IOException

Compute the quantize distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput. The number of quantized vectors to read is determined by and the results are stored in the provided array.

Throws:

IOException

quantizeScoreBulkOffsets

public void quantizeScoreBulkOffsets(byte[] q,
 int[] offsets,
 int offsetsCount,
 float[] scores,
 int count)
                              throws IOException

Compute the quantize distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput. The number of quantized vectors to read is determined by and the results are stored in the provided array. Only the vectors that are indexed in the provided are scored; the others are skipped.

Throws:

IOException

applyCorrectionsIndividually

public float applyCorrectionsIndividually(float queryLowerInterval,
 float queryUpperInterval,
 int queryComponentSum,
 float queryAdditionalCorrection,
 org.apache.lucene.index.VectorSimilarityFunction similarityFunction,
 float centroidDp,
 float lowerInterval,
 float upperInterval,
 int targetComponentSum,
 float additionalCorrection,
 float qcDist)

Computes the final score by applying the necessary corrections to the provided quantized distance.

scoreBulk

public float scoreBulk(byte[] q,
 float queryLowerInterval,
 float queryUpperInterval,
 int queryComponentSum,
 float queryAdditionalCorrection,
 org.apache.lucene.index.VectorSimilarityFunction similarityFunction,
 float centroidDp,
 float[] scores)
                throws IOException

Bulk score overload; same as scoreBulk(byte[], float, float, int, float, VectorSimilarityFunction, float, float[], int) with bulkSize as the default bulkSize param value.

Throws:

IOException

scoreBulk

public float scoreBulk(byte[] q,
 float queryLowerInterval,
 float queryUpperInterval,
 int queryComponentSum,
 float queryAdditionalCorrection,
 org.apache.lucene.index.VectorSimilarityFunction similarityFunction,
 float centroidDp,
 float[] scores,
 int bulkSize)
                throws IOException

Compute the distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput.

The number of vectors to score is defined by . The expected format of the input is as follows: First the quantized vectors are read from the input,then all the lower intervals as floats, then all the upper intervals as floats, then all the target component sums as shorts, and finally all the additional corrections as floats.

The results are stored in the provided array.

Throws:

IOException

scoreBulkOffsets

public float scoreBulkOffsets(byte[] q,
 float queryLowerInterval,
 float queryUpperInterval,
 int queryComponentSum,
 float queryAdditionalCorrection,
 org.apache.lucene.index.VectorSimilarityFunction similarityFunction,
 float centroidDp,
 int[] offsets,
 int offsetsCount,
 float[] scores,
 int count)
                       throws IOException

Compute the distance between the provided quantized query and the quantized vectors that are read from the wrapped IndexInput.

Similar to scoreBulk(byte[], float, float, int, float, org.apache.lucene.index.VectorSimilarityFunction, float, float[]), but only the vectors indexed by the provided are scored; the others are skipped.

The results are stored in the provided array.

Throws:

IOException