Package io.github.jbellis.jvector.vector

Interface VectorUtilSupport

public interface VectorUtilSupport

Interface for implementations of VectorUtil support.

Method Summary

Modifier and Type	Method	Description
void	addInPlace(float[] v1, float[] v2)	Adds v2 into v1, in place (v1 will be modified)
float	assembleAndSum(float[] data, int baseIndex, byte[] baseOffsets)	Calculates the sum of sparse points in a vector.
float	cosine(byte[] a, byte[] b)	Returns the cosine similarity between the two byte vectors.
float	cosine(float[] v1, float[] v2)	Returns the cosine similarity between the two vectors.
void	divInPlace(float[] vector, float divisor)	Divide vector by divisor, in place (vector will be modified)
int	dotProduct(byte[] a, byte[] b)	Returns the dot product computed over signed bytes.
float	dotProduct(float[] a, float[] b)	Calculates the dot product of the given float arrays.
float	dotProduct(float[] a, int aoffset, float[] b, int boffset, int length)	Calculates the dot product of float arrays of differing sizes, or a subset of the data
int	hammingDistance(long[] v1, long[] v2)
int	squareDistance(byte[] a, byte[] b)	Returns the sum of squared differences of the two byte vectors.
float	squareDistance(float[] a, float[] b)	Returns the sum of squared differences of the two vectors.
float	squareDistance(float[] a, int aoffset, float[] b, int boffset, int length)	Calculates the sum of squared differences of float arrays of differing sizes, or a subset of the data
float[]	sub(float[] lhs, float[] rhs)
void	subInPlace(float[] v1, float[] v2)	Subtracts v2 from v1, in place (v1 will be modified)
float	sum(float[] vector)	return the sum of the components of the vector
float[]	sum(List<float[]> vectors)	returns the sum of the given vectors.

Method Details

dotProduct

float dotProduct(float[] a,
 float[] b)

Calculates the dot product of the given float arrays.

dotProduct

float dotProduct(float[] a,
 int aoffset,
 float[] b,
 int boffset,
 int length)

Calculates the dot product of float arrays of differing sizes, or a subset of the data

cosine

float cosine(float[] v1,
 float[] v2)

Returns the cosine similarity between the two vectors.

squareDistance

float squareDistance(float[] a,
 float[] b)

Returns the sum of squared differences of the two vectors.

squareDistance

float squareDistance(float[] a,
 int aoffset,
 float[] b,
 int boffset,
 int length)

Calculates the sum of squared differences of float arrays of differing sizes, or a subset of the data

dotProduct

int dotProduct(byte[] a,
 byte[] b)

Returns the dot product computed over signed bytes.

cosine

float cosine(byte[] a,
 byte[] b)

Returns the cosine similarity between the two byte vectors.

squareDistance

int squareDistance(byte[] a,
 byte[] b)

Returns the sum of squared differences of the two byte vectors.

sum

float[] sum(List<float[]> vectors)

returns the sum of the given vectors.

sum

float sum(float[] vector)

return the sum of the components of the vector

divInPlace

void divInPlace(float[] vector,
 float divisor)

Divide vector by divisor, in place (vector will be modified)

addInPlace

void addInPlace(float[] v1,
 float[] v2)

Adds v2 into v1, in place (v1 will be modified)

subInPlace

void subInPlace(float[] v1,
 float[] v2)

Subtracts v2 from v1, in place (v1 will be modified)

sub

float[] sub(float[] lhs,
 float[] rhs)

Returns:

lhs - rhs, element-wise

assembleAndSum

float assembleAndSum(float[] data,
 int baseIndex,
 byte[] baseOffsets)

Calculates the sum of sparse points in a vector. This assumes the data vector is a 2d matrix which has been flattened into 1 dimension so rather than data[n][m] it's data[n * m]. With this layout this method can quickly assemble the data from this heap and sum it.

Parameters:

data - the vector of all datapoints

baseIndex - the start of the data in the offset table (scaled by the index of the lookup table)

baseOffsets - bytes that represent offsets from the baseIndex

Returns:

the sum of the points

hammingDistance

int hammingDistance(long[] v1,
 long[] v2)