From f107bf0eff2b99c1758113a72ea662cb326db0da Mon Sep 17 00:00:00 2001
From: Dmitry Kolesnikov <dmkolesnikov@gmail.com>
Date: Sat, 16 Mar 2024 18:22:00 +0200
Subject: [PATCH 1/2] Fix NaN for Cosine similarity on Zero Vector

---
 internal/noasm/cosine.go |  4 ++++
 internal/pure/cosine.go  |  4 ++++
 vector_test.go           | 45 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 53 insertions(+)

diff --git a/internal/noasm/cosine.go b/internal/noasm/cosine.go
index 31f5f53..411f2bd 100644
--- a/internal/noasm/cosine.go
+++ b/internal/noasm/cosine.go
@@ -22,6 +22,7 @@ func CosineF32(a, b []float32) (d float32) {
 		panic("vector length must be multiple of 4")
 	}
 
+	e := float32(1e-8)
 	ab := float32(0.0)
 	aa := float32(0.0)
 	bb := float32(0.0)
@@ -53,6 +54,9 @@ func CosineF32(a, b []float32) (d float32) {
 	}
 
 	s := math32.Sqrt(aa) * math32.Sqrt(bb)
+	if s < e {
+		s = e
+	}
 
 	d = (1 - ab/s) / 2
 
diff --git a/internal/pure/cosine.go b/internal/pure/cosine.go
index 4150ab4..dbd5547 100644
--- a/internal/pure/cosine.go
+++ b/internal/pure/cosine.go
@@ -20,6 +20,7 @@ func CosineF32(a, b []float32) (d float32) {
 		panic("vectors must have equal lengths")
 	}
 
+	e := float32(1e-8)
 	ab := float32(0.0)
 	aa := float32(0.0)
 	bb := float32(0.0)
@@ -31,6 +32,9 @@ func CosineF32(a, b []float32) (d float32) {
 	}
 
 	s := math32.Sqrt(aa) * math32.Sqrt(bb)
+	if s < e {
+		s = e
+	}
 
 	d = (1 - ab/s) / 2
 
diff --git a/vector_test.go b/vector_test.go
index 9dac45d..c54dae5 100644
--- a/vector_test.go
+++ b/vector_test.go
@@ -9,6 +9,7 @@
 package vector_test
 
 import (
+	"math"
 	"math/rand"
 	"testing"
 
@@ -45,6 +46,14 @@ func init() {
 	n2 = Node{ID: 2, Vector: b}
 }
 
+func zeroF32() vector.F32 {
+	v := make(vector.F32, n)
+	for i := 0; i < n; i++ {
+		v[i] = 0.0
+	}
+	return v
+}
+
 func randF32() vector.F32 {
 	v := make(vector.F32, n)
 	for i := 0; i < n; i++ {
@@ -133,6 +142,34 @@ func TestNoAsmCosineF32(t *testing.T) {
 	}
 }
 
+func TestPureCosineZeroF32(t *testing.T) {
+	sut := pure.Cosine(0)
+
+	for i := 0; i < n*100; i++ {
+		a := zeroF32()
+		b := randF32()
+
+		d := sut.Distance(a, b)
+		if math.IsNaN(float64(d)) {
+			t.Errorf("failed distance")
+		}
+	}
+}
+
+func TestNoAsmCosineZeroF32(t *testing.T) {
+	sut := noasm.Cosine(0)
+
+	for i := 0; i < n*100; i++ {
+		a := zeroF32()
+		b := randF32()
+
+		d := sut.Distance(a, b)
+		if math.IsNaN(float64(d)) {
+			t.Errorf("failed distance")
+		}
+	}
+}
+
 //
 // Benchmark
 //
@@ -153,6 +190,14 @@ func BenchmarkNoAsmEuclideanF32(t *testing.B) {
 	}
 }
 
+func BenchmarkNoAsmEuclideanUn(t *testing.B) {
+	euc := noasm.EuclideanU(0)
+
+	for i := t.N; i > 0; i-- {
+		d = euc.Distance(a, b)
+	}
+}
+
 func BenchmarkSIMDEuclideanF32(t *testing.B) {
 	euc := simd.Euclidean{}
 

From 761646306f97e9bd93e1bc70241d5dc8a7c45aae Mon Sep 17 00:00:00 2001
From: Dmitry Kolesnikov <dmkolesnikov@gmail.com>
Date: Sat, 16 Mar 2024 19:00:29 +0200
Subject: [PATCH 2/2] clean up code

---
 vector_test.go | 8 --------
 1 file changed, 8 deletions(-)

diff --git a/vector_test.go b/vector_test.go
index c54dae5..8c14001 100644
--- a/vector_test.go
+++ b/vector_test.go
@@ -190,14 +190,6 @@ func BenchmarkNoAsmEuclideanF32(t *testing.B) {
 	}
 }
 
-func BenchmarkNoAsmEuclideanUn(t *testing.B) {
-	euc := noasm.EuclideanU(0)
-
-	for i := t.N; i > 0; i-- {
-		d = euc.Distance(a, b)
-	}
-}
-
 func BenchmarkSIMDEuclideanF32(t *testing.B) {
 	euc := simd.Euclidean{}