PyTorch torch.cummin function
Pytorch torch reference manual
torch.cumminis a function in PyTorch used to compute the cumulative minimum. It returns the cumulative minimum along the specified dimension along with the corresponding indices.
Function definition
torch.cummin(input, dim, dtype=None)
Usage examples
Example
import torch
# Compute cumulative minimum
x = torch.tensor([3, 1, 4, 1, 5, 9, 2, 6])
values, indices = torch.cummin(x, dim=0)
print("Input:", x)
print("Cumulative minimum:", values)
print("Minimum indices:", indices)
# 2D tensor
x = torch.tensor([[3, 1, 4], [1, 5, 9], [2, 6, 5]], dtype=torch.float32)
# Accumulate along columns
values_col, indices_col = torch.cummin(x, dim=0)
print("nAccumulate along columns:")
print("Cumulative minimum:")
print(values_col)
print("Indices:")
print(indices_col)
# Accumulate along rows
values_row, indices_row = torch.cummin(x, dim=1)
print("nAccumulate along rows:")
print("Cumulative minimum:")
print(values_row)
# Compute cumulative minimum
x = torch.tensor([3, 1, 4, 1, 5, 9, 2, 6])
values, indices = torch.cummin(x, dim=0)
print("Input:", x)
print("Cumulative minimum:", values)
print("Minimum indices:", indices)
# 2D tensor
x = torch.tensor([[3, 1, 4], [1, 5, 9], [2, 6, 5]], dtype=torch.float32)
# Accumulate along columns
values_col, indices_col = torch.cummin(x, dim=0)
print("nAccumulate along columns:")
print("Cumulative minimum:")
print(values_col)
print("Indices:")
print(indices_col)
# Accumulate along rows
values_row, indices_row = torch.cummin(x, dim=1)
print("nAccumulate along rows:")
print("Cumulative minimum:")
print(values_row)
Other extensions