-
Notifications
You must be signed in to change notification settings - Fork 8
/
bias_test.go
104 lines (90 loc) · 3.87 KB
/
bias_test.go
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
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
// Copyright 2022 Gregory Petrosyan <[email protected]>
//
// This Source Code Form is subject to the terms of the Mozilla Public
// License, v. 2.0. If a copy of the MPL was not distributed with this
// file, You can obtain one at https://mozilla.org/MPL/2.0/.
package rand_test
import (
"flag"
"fmt"
"math"
"math/bits"
"pgregory.net/rand"
"testing"
)
var (
biasFlag = flag.Bool("bias", false, "run bias-detection tests")
)
func uint16nModulo(r *rand.Rand, n uint16) uint16 {
return uint16(r.Uint32()) % n // biased
}
func uint16nFixedPoint(r *rand.Rand, n uint16, m int) uint16 {
res, frac := bits.Mul32(uint32(n), r.Uint32()&(1<<(16+m)-1))
return (uint16(res) << (16 - m)) | (uint16(frac>>16) >> m) // biased with probability 2^-m
}
func uint16nFixedPoint0(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 0) }
func uint16nFixedPoint2(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 2) }
func uint16nFixedPoint4(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 4) }
func uint16nFixedPoint8(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 8) }
func uint16nFixedPoint10(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 10) }
func uint16nFixedPoint12(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 12) }
func uint16nFixedPoint14(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 14) }
func uint16nFixedPoint16(r *rand.Rand, n uint16) uint16 { return uint16nFixedPoint(r, n, 16) }
func uint16nCanon(r *rand.Rand, n uint16) uint16 {
res, frac := bits.Mul64(uint64(n), r.Uint64())
hi, _ := bits.Mul64(uint64(n), r.Uint64())
_, carry := bits.Add64(frac, hi, 0)
return uint16(res + carry) // biased with probability 2^-64
}
func uint16nLemire(r *rand.Rand, n uint16) uint16 {
v := uint16(r.Uint32())
prod := uint32(v) * uint32(n)
low := uint16(prod)
if low < n {
thresh := -n % n
for low < thresh {
v = uint16(r.Uint32())
prod = uint32(v) * uint32(n)
low = uint16(prod)
}
}
return uint16(prod >> 16) // unbiased
}
func TestRand_Uint16nBias_Modulo(t *testing.T) { testRandUint16nBias(t, uint16nModulo) }
func TestRand_Uint16nBias_FixedPoint0(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint0) }
func TestRand_Uint16nBias_FixedPoint2(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint2) }
func TestRand_Uint16nBias_FixedPoint4(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint4) }
func TestRand_Uint16nBias_FixedPoint8(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint8) }
func TestRand_Uint16nBias_FixedPoint10(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint10) }
func TestRand_Uint16nBias_FixedPoint12(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint12) }
func TestRand_Uint16nBias_FixedPoint14(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint14) }
func TestRand_Uint16nBias_FixedPoint16(t *testing.T) { testRandUint16nBias(t, uint16nFixedPoint16) }
func TestRand_Uint16nBias_Canon(t *testing.T) { testRandUint16nBias(t, uint16nCanon) }
func TestRand_Uint16nBias_Lemire(t *testing.T) { testRandUint16nBias(t, uint16nLemire) }
func testRandUint16nBias(t *testing.T, gen func(*rand.Rand, uint16) uint16) {
t.Helper()
if !*biasFlag {
t.Skip("specify -bias flag to run bias tests")
}
const bound = math.MaxUint16 / 4 * 3
for pow := 10; pow <= 50; pow++ {
t.Run(fmt.Sprintf("%d/%dbit", bound, pow), func(t *testing.T) {
r := rand.New()
data := make([]uint64, bound)
attempts := 1 << int64(pow)
for i := 0; i < attempts; i++ {
ix := gen(r, bound)
data[ix]++
}
var chiSq float64
expected := float64(attempts) / float64(bound)
for _, n := range data {
obs := float64(n)
chiSq += (obs - expected) * (obs - expected)
}
chiSq /= expected
df := float64(bound - 1)
t.Logf("χ2 = %.1f, DoF = %v (%v attempts, delta = %.1f%%)", chiSq, df, attempts, (chiSq-df)/df*100)
})
}
}