liuyu/pgvector
1
0
Fork 0
mirror of https://github.com/pgvector/pgvector.git synced 2026-07-13 16:16:56 +08:00
Code Issues Packages Projects Releases Wiki Activity

Improved types and scoping for k-means

Browse Source
This commit is contained in:
Andrew Kane
2023-09-12 18:33:31 -07:00
parent 466c556b1a
commit e367155732
1 changed files with 9 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
src/ivfkmeans.c
View File

@@ -29,7 +29,7 @@ InitCenters(Relation index, VectorArray samples, VectorArray centers, float *low
centers->length++; centers->length++;
for (j = 0; j < numSamples; j++) for (j = 0; j < numSamples; j++)
weight[j] = DBL_MAX; weight[j] = FLT_MAX;
for (int i = 0; i < numCenters; i++) for (int i = 0; i < numCenters; i++)
{ {
@@ -181,7 +181,6 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
float *halfcdist; float *halfcdist;
float *newcdist; float *newcdist;
int closestCenter; int closestCenter;
double distance;
bool rj; bool rj;
bool rjreset; bool rjreset;
double dxcx; double dxcx;
@@ -243,7 +242,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
/* Assign each x to its closest initial center c(x) = argmin d(x,c) */ /* Assign each x to its closest initial center c(x) = argmin d(x,c) */
for (j = 0; j < numSamples; j++) for (j = 0; j < numSamples; j++)
{ {
double minDistance = DBL_MAX; float minDistance = FLT_MAX;
closestCenter = 0; closestCenter = 0;
@@ -251,7 +250,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
for (k = 0; k < numCenters; k++) for (k = 0; k < numCenters; k++)
{ {
/* TODO Use Lemma 1 in k-means++ initialization */ /* TODO Use Lemma 1 in k-means++ initialization */
distance = lowerBound[j * numCenters + k]; float distance = lowerBound[j * numCenters + k];
if (distance < minDistance) if (distance < minDistance)
{ {
@@ -279,7 +278,8 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
for (k = j + 1; k < numCenters; k++) for (k = j + 1; k < numCenters; k++)
{ {
distance = 0.5 * DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(vec), PointerGetDatum(VectorArrayGet(centers, k)))); float distance = 0.5 * DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(vec), PointerGetDatum(VectorArrayGet(centers, k))));
halfcdist[j * numCenters + k] = distance; halfcdist[j * numCenters + k] = distance;
halfcdist[k * numCenters + j] = distance; halfcdist[k * numCenters + j] = distance;
} }
@@ -288,10 +288,12 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
/* For all centers c, compute s(c) */ /* For all centers c, compute s(c) */
for (j = 0; j < numCenters; j++) for (j = 0; j < numCenters; j++)
{ {
double minDistance = DBL_MAX; float minDistance = FLT_MAX;
for (k = 0; k < numCenters; k++) for (k = 0; k < numCenters; k++)
{ {
float distance;
if (j == k) if (j == k)
continue; continue;
@@ -423,7 +425,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
{ {
for (k = 0; k < numCenters; k++) for (k = 0; k < numCenters; k++)
{ {
distance = lowerBound[j * numCenters + k] - newcdist[k]; float distance = lowerBound[j * numCenters + k] - newcdist[k];
if (distance < 0) if (distance < 0)
distance = 0; distance = 0;