Pytorch 学习 - 1.pytorch 张量基本操作

1.张量基本操作简介与创建

1.1 torch.zeros

import torch
out_t =torch.tensor([1])
t = torch.zeros((4,4),out=out_t)

print('t :', t, '\n','out_t :', out_t)
print(id(t),id(out_t))

output

t : tensor([[0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0]]) 
 out_t : tensor([[0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0]])
2777186850912 2777186850912

out_t 与t 共用内存

1.2 torch.ones

import torch
out_t =torch.tensor([1])
t = torch.ones((3,3),out=out_t)
print('t :', t, '\n', out_t)
print(id(t),id(out_t))

output
t : tensor([[1, 1, 1],
        [1, 1, 1],
        [1, 1, 1]]) 
 tensor([[1, 1, 1],
        [1, 1, 1],
        [1, 1, 1]])
2777187916752 2777187916752

1.3 torch.full

import torch
out_t =torch.tensor([1])
t = torch.full((3,3),4,out=out_t)
print('t :', t, '\n', out_t)
print(id(t),id(out_t))

output

t : tensor([[4, 4, 4],
        [4, 4, 4],
        [4, 4, 4]]) 
 tensor([[4, 4, 4],
        [4, 4, 4],
        [4, 4, 4]])
2776837717504 2776837717504

1.4 torch.full_like

import torch
input = torch.empty(2,3)
t = torch.full_like(input,4)
print('t :', t, '\n', input)

output：

t : tensor([[4., 4., 4.],
        [4., 4., 4.]]) 
 tensor([[7.5670e-44, 8.1275e-44, 6.7262e-44],
        [6.7262e-44, 8.1275e-44, 7.2868e-44]])

1.5 torch.arange 目的创建等差数列

import torch
t = torch.arange(1,21,2)
print('t :', t)

start = 1

end < 21

step =2 

output

t : tensor([ 1,  3,  5,  7,  9, 11, 13, 15, 17, 19])

1.6 torch.linsapce 目的创建等差数列

import torch
t = torch.linspace(1,20,39)
print('t :', t)

start = 1

end <= 20   <<<< 注意，是闭合

step =39 代表 点数 共计 39点  

t : tensor([ 1.0000,  1.5000,  2.0000,  2.5000,  3.0000,  3.5000,  4.0000,  4.5000,
         5.0000,  5.5000,  6.0000,  6.5000,  7.0000,  7.5000,  8.0000,  8.5000,
         9.0000,  9.5000, 10.0000, 10.5000, 11.0000, 11.5000, 12.0000, 12.5000,
        13.0000, 13.5000, 14.0000, 14.5000, 15.0000, 15.5000, 16.0000, 16.5000,
        17.0000, 17.5000, 18.0000, 18.5000, 19.0000, 19.5000, 20.0000])

import torch
t = torch.linspace(1,40,40)
print('t :', t)

t : tensor([ 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.])

1.7 torch.eye() 对角矩阵

import torch
t = torch.eye(5,6)
print('output t :', t)

row * col
output 
t : tensor([[1., 0., 0., 0., 0., 0.],
        [0., 1., 0., 0., 0., 0.],
        [0., 0., 1., 0., 0., 0.],
        [0., 0., 0., 1., 0., 0.],
        [0., 0., 0., 0., 1., 0.]])

import torch
t = torch.eye(5)
print('output t :', t)

可省略 col 
output t : tensor([[1., 0., 0., 0., 0.],
        [0., 1., 0., 0., 0.],
        [0., 0., 1., 0., 0.],
        [0., 0., 0., 1., 0.],
        [0., 0., 0., 0., 1.]])

1.8 torch.normal() 正态分布抽样，有多种方法

import torch
std = torch.arange(1,5,1,dtype = torch.float)
mean = torch.arange(1,5,1,dtype = torch.float)
t = torch.normal(mean,std)
print('output std :', std,'mean',mean,'t:',t)

output std : tensor([1., 2., 3., 4.]) mean tensor([1., 2., 3., 4.]) t: tensor([2.2844, 5.3268, 7.8541, 8.5367])

import torch
std = torch.arange(1,5,1,dtype = torch.float)
mean = torch.arange(1,5,1,dtype = torch.float)
t = torch.normal(0.,1.,size=(10,))
print('output std :', std,'mean',mean,'t:',t)

output std : tensor([1., 2., 3., 4.]) mean tensor([1., 2., 3., 4.]) t: tensor([ 0.6912, -0.0292, -1.9124,  0.4636, -1.4063,  0.3485, -1.1667, -0.3790,  1.1603,  0.4437])

import torch
std = torch.arange(1,5,1,dtype = torch.float)
mean = 1
t = torch.normal(mean,std)
print('output std :', std,'mean',mean,'t:',t)

output std : tensor([1., 2., 3., 4.]) mean 1 t: tensor([ 1.4478,  2.1242, -1.8429, -2.5465])

1.9 torch.randn()  rand + normal

import torch
t = torch.randn(size=(4,))
print('output t:',t)

output t: tensor([ 0.3946,  1.7896, -0.5532,  0.3768])

t = torch.randn(size=(4,))  与 t = torch.normal(0.,1.,size=(4,)) 是一样的

import torch
t = torch.rand(size=[40,])
print('output t:',t)

import torch

t = torch.rand(size=[40,])

print('output t:',t)

output t: tensor([0.3416, 0.5088, 0.2788, 0.0533, 0.7062, 0.0581, 0.7742, 0.2045, 0.1741,
        0.8933, 0.1506, 0.4629, 0.8797, 0.1002, 0.5638, 0.5921, 0.0862, 0.8362,
        0.8205, 0.6718, 0.8183, 0.0334, 0.7365, 0.1660, 0.8979, 0.1517, 0.3308,
        0.2383, 0.1495, 0.9984, 0.4724, 0.7952, 0.6377, 0.3555, 0.8515, 0.2382,
        0.7898, 0.4520, 0.4491, 0.8748])

import torch
input = torch.empty(2,3)
t = torch.rand_like(input)
print('output t:',t)

output t: tensor([[0.9993, 0.2470, 0.4895],
        [0.2251, 0.1244, 0.0183]])

还有 torch.randint() , torch.randint_like()

import torch
t = torch.randint(1,20,size=[40,])
print('output t:',t)

output t: tensor([ 8, 15,  3, 18, 17, 19,  2, 10,  6, 13, 10,  3,  2,  7, 15, 11,  7, 12,  14,  3,  3,  1, 18, 15, 10, 10,  2,  3, 11, 14, 18,  4, 17,  7, 14, 15,
         7,  5, 10, 16])

import torch
t = torch.randperm(20)
print('output t:',t)

output t: tensor([12,  8, 17, 15, 10,  5, 11, 16,  7,  2,  6,  9,  4, 13, 18,  3,  1, 19, 0, 14])

生成 0-19 乱序的整数索引

import torch
input = torch.empty(2,10)
t = torch.bernoulli(input,0.2)
print('output t :', t)

output t : tensor([[1., 0., 0., 0., 0., 0., 1., 1., 1., 0.],
        [0., 0., 1., 0., 0., 0., 0., 0., 0., 1.]])
生成 2x10 矩阵 0，1 概率0.2