PyTorch torch.cumsum 函数
Pytorch torch 参考手册torch.cumsum 是 PyTorch 中用于计算累积和的函数。它返回沿指定维度的累积和，即从开始到当前位置所有元素的和。
函数定义
torch.cumsum(input, dim, dtype=None)
使用示例
实例
import torch
# 计算累积和
x = torch.tensor([1, 2, 3, 4, 5])
result = torch.cumsum(x, dim=0)
print("输入:", x)
print("累积和:", result)
# 输出: tensor([1, 3, 6, 10, 15])
# 说明: 1, 1+2=3, 1+2+3=6, 1+2+3+4=10, 1+2+3+4+5=15
# 2 维张量
x = torch.tensor([[1, 2, 3], [4, 5, 6]], dtype=torch.float32)
# 按列累积
result_col = torch.cumsum(x, dim=0)
print("n按列累积和:")
print(result_col)
# tensor([[ 1, 2, 3],
# [ 5, 7, 9]])
# 按行累积
result_row = torch.cumsum(x, dim=1)
print("n按行累积和:")
print(result_row)
# tensor([[ 1, 3, 6],
# [ 4, 9, 15]])
# 计算累积和
x = torch.tensor([1, 2, 3, 4, 5])
result = torch.cumsum(x, dim=0)
print("输入:", x)
print("累积和:", result)
# 输出: tensor([1, 3, 6, 10, 15])
# 说明: 1, 1+2=3, 1+2+3=6, 1+2+3+4=10, 1+2+3+4+5=15
# 2 维张量
x = torch.tensor([[1, 2, 3], [4, 5, 6]], dtype=torch.float32)
# 按列累积
result_col = torch.cumsum(x, dim=0)
print("n按列累积和:")
print(result_col)
# tensor([[ 1, 2, 3],
# [ 5, 7, 9]])
# 按行累积
result_row = torch.cumsum(x, dim=1)
print("n按行累积和:")
print(result_row)
# tensor([[ 1, 3, 6],
# [ 4, 9, 15]])