-
Notifications
You must be signed in to change notification settings - Fork 24
/
cholesky_test.go
375 lines (343 loc) · 9.13 KB
/
cholesky_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
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
package sparse
import (
"fmt"
"math"
"testing"
"golang.org/x/exp/rand"
"gonum.org/v1/gonum/mat"
"gonum.org/v1/gonum/stat/distmat"
)
func TestCholesky(t *testing.T) {
t.Parallel()
for _, test := range []struct {
a *mat.SymDense
cond float64
want *mat.TriDense
posdef bool
}{
{
a: mat.NewSymDense(3, []float64{
4, 1, 1,
0, 2, 3,
0, 0, 6,
}),
cond: 37,
want: mat.NewTriDense(3, true, []float64{
2, 0.5, 0.5,
0, 1.3228756555322954, 2.0788046015507495,
0, 0, 1.195228609334394,
}),
posdef: true,
},
} {
if !cholMatches(test.a) {
t.Error("chol mismatch")
}
}
iters := 16
src := rand.NewSource(1)
for i := 0; i < iters; i++ {
n := 128
frac := 0.05
m := randomSymDensePosDefinite(n, frac, src)
if !cholMatches(m) {
t.Error("mismatch on random matrix")
}
}
}
func TestCholeskySolveVecTo(t *testing.T) {
t.Parallel()
for idx, test := range []struct {
a *mat.SymDense
b *mat.VecDense
ans *mat.VecDense
}{
{
a: mat.NewSymDense(2, []float64{
1, 0,
0, 1,
}),
b: mat.NewVecDense(2, []float64{5, 6}),
ans: mat.NewVecDense(2, []float64{5, 6}),
},
{
a: mat.NewSymDense(3, []float64{
53, 59, 37,
0, 83, 71,
0, 0, 101,
}),
b: mat.NewVecDense(3, []float64{5, 6, 7}),
ans: mat.NewVecDense(3, []float64{0.20745069393718094, -0.17421475529583694, 0.11577794010226464}),
},
} {
var chol mat.Cholesky
ok := chol.Factorize(test.a)
if !ok {
t.Fatal("unexpected Cholesky factorization failure: not positive definite")
}
var x mat.VecDense
err := chol.SolveVecTo(&x, test.b)
if err != nil {
t.Errorf("unexpected error from Cholesky solve: %v", err)
}
if !mat.EqualApprox(&x, test.ans, 1e-12) {
t.Error("incorrect Cholesky solve solution")
}
var ans mat.VecDense
ans.MulVec(test.a, &x)
if !mat.EqualApprox(&ans, test.b, 1e-12) {
t.Error("incorrect Cholesky solve solution product")
}
// if !cholMatches(test.a) {
// t.Error("chol mismatch in solvevecto test")
// }
n := test.a.Symmetric()
aCOO := matToCOO(test.a, 1e-10)
aCSR := aCOO.ToCSR()
var sc Cholesky
sc.Factorize(aCSR)
xs := mat.NewVecDense(n, nil)
sc.SolveVecTo(xs, test.b)
if !mat.EqualApprox(xs, test.ans, 1e-12) {
t.Error("incorrect sparse Cholesky solution", idx)
}
}
}
func TestCholeskyAt(t *testing.T) {
t.Parallel()
for _, test := range []*mat.SymDense{
mat.NewSymDense(3, []float64{
53, 59, 37,
59, 83, 71,
37, 71, 101,
}),
} {
var chol Cholesky
csr := matToCSR(test, 1e-8)
chol.Factorize(csr)
n := test.Symmetric()
cn := chol.Symmetric()
if cn != n {
t.Errorf("Cholesky size does not match. Got %d, want %d", cn, n)
}
for i := 0; i < n; i++ {
for j := 0; j < n; j++ {
got := chol.At(i, j)
want := test.At(i, j)
if math.Abs(got-want) > 1e-12 {
t.Errorf("Cholesky at does not match at %d, %d. Got %v, want %v", i, j, got, want)
}
}
}
}
}
func TestCholeskySolveTo(t *testing.T) {
t.Parallel()
for _, test := range []struct {
a *mat.SymDense
b *mat.Dense
ans *mat.Dense
}{
{
a: mat.NewSymDense(2, []float64{
1, 0,
0, 1,
}),
b: mat.NewDense(2, 1, []float64{5, 6}),
ans: mat.NewDense(2, 1, []float64{5, 6}),
},
{
a: mat.NewSymDense(3, []float64{
53, 59, 37,
0, 83, 71,
37, 71, 101,
}),
b: mat.NewDense(3, 1, []float64{5, 6, 7}),
ans: mat.NewDense(3, 1, []float64{0.20745069393718094, -0.17421475529583694, 0.11577794010226464}),
},
} {
var chol Cholesky
csr := matToCSR(test.a, 1e-8)
chol.Factorize(csr)
var x mat.Dense
err := chol.SolveTo(&x, test.b)
if err != nil {
t.Errorf("unexpected error from Cholesky solve: %v", err)
}
if !mat.EqualApprox(&x, test.ans, 1e-12) {
t.Error("incorrect Cholesky solve solution")
}
var ans mat.Dense
ans.Mul(test.a, &x)
if !mat.EqualApprox(&ans, test.b, 1e-12) {
t.Error("incorrect Cholesky solve solution product")
}
}
}
func cholMatches(a *mat.SymDense) bool {
_, n := a.Dims()
var chol mat.Cholesky
ok := chol.Factorize(a)
if !ok {
fmt.Println("cannot factorize")
return false
}
var L mat.TriDense
chol.LTo(&L)
simpleRes := mat.NewTriDense(n, false, nil)
cholSimple(a, simpleRes)
if !mat.EqualApprox(&L, simpleRes, 1e-10) {
return false
}
coo := NewCOO(n, n, nil, nil, nil)
csrRes := coo.ToCSR()
aCOO := matToCOO(a, 1e-10)
aCSR := aCOO.ToCSR()
cholCSR(aCSR, csrRes)
if !mat.EqualApprox(&L, csrRes, 1e-10) {
for i := 0; i < n; i++ {
for j := 0; j < n; j++ {
fmt.Println(L.At(i, j), csrRes.At(i, j))
}
}
return false
}
return true
}
// computes a permutation matrix where each non-zero is rand.Float64() instead of 1
// and then return aat of that
func randomScaledPermutationMatrixAAT(n int, src rand.Source) *mat.SymDense {
// p contains a permutation matrix
p := mat.NewDense(n, n, nil)
u := distmat.NewUniformPermutation(src)
u.PermTo(p)
// s is a scale matrix with rnd.Float64() down the diagonal
s := mat.NewDense(n, n, nil)
rnd := rand.New(src)
for i := 0; i < n; i++ {
s.Set(i, i, rnd.Float64())
}
// a = s p
// compute aat and convert to symdense
a := mat.NewDense(n, n, nil)
a.Mul(p, s)
at := a.T()
aat := mat.NewDense(n, n, nil)
aat.Mul(a, at)
aatSym := mat.NewSymDense(n, aat.RawMatrix().Data)
return aatSym
}
func randomSymDensePosDefinite(n int, fracNZ float64, src rand.Source) *mat.SymDense {
ok := false
for !ok {
m := randomSymDensePosDefiniteInternal(n, fracNZ, src)
var chol mat.Cholesky
ok = chol.Factorize(m)
if ok {
return m
}
}
return nil
}
func randomSymDensePosDefiniteInternal(n int, fracNZ float64, src rand.Source) *mat.SymDense {
rnd := rand.New(src)
m := mat.NewDense(n, n, nil)
nnz := int(float64(n) * float64(n) * fracNZ)
rList := make([]int, nnz)
cList := make([]int, nnz)
for i := range rList {
rList[i] = rnd.Intn(n)
cList[i] = rnd.Intn(n)
}
for i := range rList {
r := rList[i]
c := cList[i]
m.Set(r, c, rnd.Float64())
m.Set(c, r, rnd.Float64())
}
mt := m.T()
mmt := mat.NewDense(n, n, nil)
mmt.Mul(m, mt)
mmtSym := mat.NewSymDense(n, mmt.RawMatrix().Data)
return mmtSym
}
func matToCOO(m mat.Matrix, tol float64) *COO {
r, c := m.Dims()
newMat := NewCOO(r, c, nil, nil, nil)
for i := 0; i < r; i++ {
for j := 0; j < c; j++ {
v := m.At(i, j)
if v > tol || v < -tol {
newMat.Set(i, j, v)
}
}
}
// nSize := float64(r * c)
// nnz := float64(newMat.NNZ())
// ratio := nnz / nSize
// fmt.Println("frac=", ratio)
return newMat
}
func matToCSR(m mat.Matrix, tol float64) *CSR {
coo := matToCOO(m, tol)
return coo.ToCSR()
}
func BenchmarkCholSimple800(b *testing.B) { cholSimpleBench(b, 800, 0) }
func BenchmarkCholGoNum400S3(b *testing.B) { cholGoNumBench(b, 400, 3.0/400) }
func BenchmarkCholGoNum400S5(b *testing.B) { cholGoNumBench(b, 400, 5.0/400) }
func BenchmarkCholGoNum400S7(b *testing.B) { cholGoNumBench(b, 400, 7.0/400) }
func BenchmarkCholGoNum400S9(b *testing.B) { cholGoNumBench(b, 400, 9.0/400) }
func BenchmarkCholGoNum400(b *testing.B) { cholGoNumBench(b, 400, 0.0) }
func BenchmarkCholGoNum800(b *testing.B) { cholGoNumBench(b, 800, 0.0) }
func BenchmarkCholGoNum1600(b *testing.B) { cholGoNumBench(b, 1600, 0.0) }
func BenchmarkCholGoNum3200(b *testing.B) { cholGoNumBench(b, 3200, 0.0) }
func BenchmarkCholGoNum6400(b *testing.B) { cholGoNumBench(b, 6400, 0.0) }
//func BenchmarkCholGoNum12800(b *testing.B) { cholGoNumBench(b, 12800, 0.0) }
func BenchmarkCholSparse400S3(b *testing.B) { sparseCholBench(b, 400, 3.0/400) }
func BenchmarkCholSparse400S5(b *testing.B) { sparseCholBench(b, 400, 5.0/400) }
func BenchmarkCholSparse400S7(b *testing.B) { sparseCholBench(b, 400, 7.0/400) }
func BenchmarkCholSparse400S9(b *testing.B) { sparseCholBench(b, 400, 9.0/400) }
func BenchmarkCholSparse400(b *testing.B) { sparseCholBench(b, 400, 0.0) }
func BenchmarkCholSparse800(b *testing.B) { sparseCholBench(b, 800, 0.0) }
func BenchmarkCholSparse1600(b *testing.B) { sparseCholBench(b, 1600, 0.0) }
func BenchmarkCholSparse3200(b *testing.B) { sparseCholBench(b, 3200, 0.0) }
func BenchmarkCholSparse6400(b *testing.B) { sparseCholBench(b, 6400, 0.0) }
//func BenchmarkCholSparse12800(b *testing.B) { sparseCholBench(b, 12800, 0.0) }
func cholGoNumBench(b *testing.B, size int, frac float64) {
src := rand.NewSource(1)
m := randomScaledPermutationMatrixAAT(size, src)
if frac != 0.0 {
m = randomSymDensePosDefinite(size, frac, src)
}
b.ResetTimer()
for i := 0; i < b.N; i++ {
var chol mat.Cholesky
chol.Factorize(m)
}
}
func cholSimpleBench(b *testing.B, size int, frac float64) {
src := rand.NewSource(1)
m := randomScaledPermutationMatrixAAT(size, src)
if frac != 0.0 {
m = randomSymDensePosDefinite(size, frac, src)
}
simpleRes := mat.NewTriDense(size, false, nil)
b.ResetTimer()
for i := 0; i < b.N; i++ {
cholSimple(m, simpleRes)
}
}
func sparseCholBench(b *testing.B, size int, frac float64) {
src := rand.NewSource(1)
mDense := randomScaledPermutationMatrixAAT(size, src)
if frac != 0.0 {
mDense = randomSymDensePosDefinite(size, frac, src)
}
coo := NewCOO(size, size, nil, nil, nil)
csrRes := coo.ToCSR()
aCSR := matToCSR(mDense, 1e-8)
b.ResetTimer()
for i := 0; i < b.N; i++ {
cholCSR(aCSR, csrRes)
}
}