gomoku/libtorch_test.cpp at master · yffbit/gomoku · GitHub

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
#include <vector>
#include <deque>
#include <iostream>
#include <torch/torch.h>
#include <torch/script.h>

using namespace std;

int main()
{
	at::Tensor a = torch::tensor({ {1,2},{3,4} }, torch::Dtype::Int);
	at::Tensor b = torch::tensor({ {1,0},{2,4} }, torch::Dtype::Int);
	cout << "a:" << endl << a << endl;
	cout << "b:" << endl << b << endl;

	// reshape 没有复制数据
	a.reshape({ 4 })[0] = 0;
	cout << "a.reshape({ 4 })[0] = 0" << endl << a << endl;
	a.reshape({ 4 })[2] = 10;
	cout << "a.reshape({ 4 })[2] = 10" << endl << a << endl;

	// 对应元素操作
	cout << "a + b" << endl << a + b << endl;
	cout << "a * b" << endl << a * b << endl;
	cout << "a.mul(b)" << endl << a.mul(b) << endl;
	// 矩阵运算
	cout << "a.matmul(b)" << endl << a.matmul(b) << endl;
	cout << "b.matmul(a)" << endl << b.matmul(a) << endl;
	// 广播操作
	cout << "a + 1" << endl << a + 1 << endl;
	cout << "(a > 1)" << endl << (a > 1) << endl;
	cout << "a.gt(1)" << endl << a.gt(1) << endl;
	cout << "(a < 1)" << endl << (a < 1) << endl;
	cout << "a.lt(1)" << endl << a.lt(1) << endl;
	// 索引
	at::Tensor c = a.select(0, 0); // 0 轴上取 index=0 的数据，也就是第一行数据（共享内存）
	at::Tensor d = a.select(1, 0); // 1 轴上取 index=0 的数据，也就是第一列数据（共享内存）
	cout << "a:" << endl << a << endl;
	cout << "c:" << endl << c << endl;
	cout << "d:" << endl << d << endl;
	a[0][0] = 0;
	cout << "a:" << endl << a << endl;
	cout << "c:" << endl << c << endl;
	cout << "d:" << endl << d << endl;

	at::Tensor e = a.index_select(0, torch::tensor({1,1,0})); // 0 轴上取 index={1,1,0} 的数据，也就是两次第二行数据，一次第一行数据（不共享内存）
	at::Tensor f = a.index_select(1, torch::tensor({1,0}));	// 1 轴上取 index={1,0} 的数据，也就是第二列数据，第一列数据（不共享内存）
	cout << "a:" << endl << a << endl;
	cout << "e:" << endl << e << endl;
	cout << "f:" << endl << f << endl;
	a[0][0] = 1; e[1][0] = 8; f[1][0] = 9;
	cout << "a:" << endl << a << endl;
	cout << "e:" << endl << e << endl;
	cout << "f:" << endl << f << endl;

	at::Tensor g = torch::cat({ a,b }, 0);	// 不增加维度，按照指定维度堆叠（不共享内存）
	cout << "a:" << endl << a << endl;
	cout << "b:" << endl << b << endl;
	cout << "g:" << endl << g << endl;
	g[0][1] = 12; g[2][0] = 13;
	cout << "a:" << endl << a << endl;
	cout << "b:" << endl << b << endl;
	cout << "g:" << endl << g << endl;
	cout << " torch::cat({ a,b }, 1):" << endl << torch::cat({ a,b }, 1) << endl;

	at::Tensor h = torch::stack({ a,b }, 0);	// 增加维度，按照指定维度堆叠（不共享内存）
	cout << "a:" << endl << a << endl;
	cout << "b:" << endl << b << endl;
	cout << "h:" << endl << h << endl;
	h[0][0][1] = 14; h[1][0][0] = 15;
	cout << "a:" << endl << a << endl;
	cout << "b:" << endl << b << endl;
	cout << "h:" << endl << h << endl;
	cout << " torch::stack({ a,b }, 1):" << endl << torch::stack({ a,b }, 1) << endl;// a 的第一行和 b 的第一行堆叠，a 的第二行和 b 的第二行堆叠

	at::Tensor i = g.slice(0, 0, 100, 2);	// 0 轴上从 index=0 开始取数据 index=100 结束 index步进2 取第一行，第三行... （共享内存）
	cout << "g:" << endl << g << endl;
	cout << "i:" << endl << i << endl;
	i[1][1] = 20;
	cout << "g:" << endl << g << endl;
	cout << "i:" << endl << i << endl;

	vector<int> j(3, 0);
	j[1] = 10; j[2] = 20;
	at::Tensor k = torch::tensor(j, torch::Dtype::Int);	// 不共享内存
	cout << "j:" << endl << j << endl;
	cout << "k:" << endl << k << endl;
	j[2] = 3; j[1] = 2; j[0] = 1;
	cout << "j:" << endl << j << endl;
	cout << "k:" << endl << k << endl;

	at::Tensor l = a;	// 共享内存
	cout << "a:" << endl << a << endl;
	cout << "l:" << endl << l << endl;
	a[0][0] = 11;
	cout << "a:" << endl << a << endl;
	cout << "l:" << endl << l << endl;

	deque<at::Tensor> m(2, torch::zeros({ 2,3 }, torch::Dtype::Int));
	m[0][0] = 1; m[1][1] = 2;	// 两个元素共享内存
	cout << "m[0]:" << endl << m[0] << endl;
	cout << "m[1]:" << endl << m[1] << endl;
	m[1][1][0] = 3;
	cout << "m[0]:" << endl << m[0] << endl;
	cout << "m[1]:" << endl << m[1] << endl;

	vector<at::Tensor> n(2, torch::zeros({ 2,3 }, torch::Dtype::Int));
	n[0][0] = 3; n[1][1] = 4;	// 两个元素共享内存
	cout << "n[0]:" << endl << n[0] << endl;
	cout << "n[1]:" << endl << n[1] << endl;
	n[1][1][0] = 5;
	cout << "n[0]:" << endl << n[0] << endl;
	cout << "n[1]:" << endl << n[1] << endl;
	n.clear();
	n.push_back(a);	// 没有申请内存
	n.push_back(a);
	cout << "a:" << endl << a << endl;
	cout << "n[0]:" << endl << n[0] << endl;
	cout << "n[1]:" << endl << n[1] << endl;
	a[0][0] = 14;
	cout << "a:" << endl << a << endl;
	cout << "n[0]:" << endl << n[0] << endl;
	cout << "n[1]:" << endl << n[1] << endl;

	n.clear();
	n.emplace_back(a);	// 没有申请内存
	n.emplace_back(a);
	cout << "a:" << endl << a << endl;
	cout << "n[0]:" << endl << n[0] << endl;
	cout << "n[1]:" << endl << n[1] << endl;
	a[0][0] = 18;
	cout << "a:" << endl << a << endl;
	cout << "n[0]:" << endl << n[0] << endl;
	cout << "n[1]:" << endl << n[1] << endl;

	at::Tensor o0 = torch::zeros({ 0,2,3,3 }, torch::Dtype::Int);
	cout << "torch::zeros({ 0,2,3,3 }, torch::Dtype::Int)" << endl << o0 << endl;

	at::Tensor o1 = torch::range(1, 18, torch::Dtype::Int).reshape({1,2,3,3});
	at::Tensor o2 = torch::range(19, 36, torch::Dtype::Int).reshape({1,2,3,3});
	at::Tensor o = torch::cat({ o0,o1,o2 }, 0);
	cout << "o1:" << endl << o1 << endl;
	cout << "o2:" << endl << o2 << endl;
	cout << "o:" << endl << o << endl;
	// {2,3}轴逆时针旋转90度*1，也就是每个通道逆时针旋转90度
	cout << "o.rot90(1, {2,3}):" << endl << o.rot90(1, {2,3}) << endl;
	// 在2轴上翻转，也就是每个通道上下翻转
	cout << "o.flip(2):" << endl << o.flip(2) << endl;
	cout << "o.reshape({2,2,9}):" << endl << o.reshape({2,2,9}) << endl;
	cout << "o.reshape({2,2,9}).reshape({2,2,3,3}):" << endl << o.reshape({2,2,9}).reshape({2,2,3,3}) << endl;

	cout << "torch::get_num_threads() : " << torch::get_num_threads() << endl;

	cout << torch::randn({2,3}) << endl;
	cout << torch::randperm(10) << endl;
	cout << torch::randperm(10) << endl;
	cout << torch::randperm(10) << endl;
	cout << torch::randperm(10).slice(0, 0, 2) << endl;
	cout << torch::randperm(10).slice(0,0,100) << endl;
	cout << torch::cat({ torch::randperm(10).slice(0,0,2),torch::randperm(10).slice(0,0,100) }, 0) << endl;

	at::Tensor idx = torch::tensor({ 1,0,1,0 });
	cout << idx << endl;
	at::Tensor p = a.index(idx);	// 不共享内存
	cout << "a:" << endl << a << endl;
	cout << "p:" << endl << p << endl;
	a[0][1] = 16; a[1][0] = 17; p[1][1] = 19;
	cout << "a:" << endl << a << endl;
	cout << "p:" << endl << p << endl;
	cout << a.index(torch::tensor({1})) << endl;
	cout << a[1].index(torch::tensor({1})) << endl;

	at::Tensor q = torch::range(1, 30, torch::Dtype::Int).reshape({ 10,3 });
	cout << "q:" << endl << q << endl;
	cout << "idx:" << endl << idx << endl;
	cout << "q.index(idx):" << endl << q.index(idx) << endl;

	idx = torch::randperm(q.size(0), torch::Dtype::Long);	// 必须是 CPULongType
	//idx = torch::randperm(q.size(0), torch::Dtype::Int);	// CPUIntType 运行出错
	cout << "q:" << endl << q << endl;
	cout << "idx:" << endl << idx << endl;
	cout << "q.index(idx):" << endl << q.index(idx) << endl;

	return 0;
}