import torch

x = torch.randn(2, 3)
y = torch.randn(2, 3)

z = x + y
print(z)

z = torch.add(x, y)
print(z)

y.add_(x)
print(y)

# 二維矩陣相乘
a = torch.full([2, 2], 3, dtype=torch.long)
b = torch.ones(2, 2, dtype=torch.long)

print(a)
print(b)
print(torch.mm(a, b))

#  matmul 和 @ 可以用于二維矩陣計(jì)算，也可以是多維
# 四維，計(jì)算的時(shí)候就是前兩維不變，后兩維進(jìn)行計(jì)算。
a = torch.rand(1, 1, 3, 2)
b = torch.rand(1, 1, 2, 4)

c = torch.matmul(a, b)
print(a)
print(b)
print(c)

a = torch.full([2, 2], 6)
print(a.pow(3))

a = torch.full([2, 2], 6)
print(a**2)

a = torch.full([2, 2], 1024)

print(a.sqrt())
print(a.rsqrt())
print(a**0.5)

a = torch.ones(2, 2)

print(torch.exp(a))
print(torch.log(a))
print(torch.log2(a))

a = torch.tensor(3.1415926)

print(a.floor())
print(a.ceil())
print(a.trunc())
print(a.frac())

a = torch.tensor(3.1415926)

print(a.round())

x = torch.randn(1)

print(x)
print(x.item())

a = torch.randn(4, 3)
print(a)

# 取第二列
print(a[:, 1])

# 取前兩列
print(a[:, :2])

x = torch.randn(4, 4)
y = x.view(16)

# -1, 自動匹配個(gè)數(shù)
z = x.view(-1, 8)

print(x)
print(y)
print(z)

# 底層內(nèi)存共享
x = torch.ones(5)
print(x)

y = x.numpy()
print(y)

x.add_(1)
print(y)

import numpy as np
x = np.ones(5)
y = torch.from_numpy(x)
print(y)

if torch.cuda.is_available():
    
    device = torch.device("cuda")
    x = torch.randn(2, 3)
    print(x)

    y = x.to(device)
    print(y)

    z = torch.randn(2, 3, device="cuda")
    print(z)
    
    # 同時(shí)在GPU上才能相加
    print(y + z)

    # 轉(zhuǎn)換會cpu
    print(z.to("cpu"))