* * Pytorch torch Reference Manual](#)
`torch.searchsorted` is a function in PyTorch used to search for the insertion positions of elements in a sorted tensor. The returned value is the index position where the element should be inserted.
### Function Definition
torch.searchsorted(sorted_sequence, values, side='left', out_int32=False, right=False)
Parameter description:
* `sorted_sequence`: a sorted one-dimensional or multi-dimensional tensor
* `values`: the values to search for
* `side`: 'left' or 'right', determines whether to return the left or right insertion position
* `out_int32`: whether to return int32 type
* `right`: deprecated, use side instead
* * *
## Usage Examples
## Example
import torch
# Create sorted sequence
sorted_seq = torch.tensor([1,3,5,7,9])
# Search for positions of values
values = torch.tensor([3,6,8])
y = torch.searchsorted(sorted_seq, values)
print(y)
Output result:
tensor([1, 3, 3])
Example
import torch
# Create sorted sequence
sorted_seq = torch.tensor([1, 3, 5, 7, 9])
# Search using side='right'
values = torch.tensor([3, 6, 8])
y = torch.searchsorted(sorted_seq, values, side='right')
print(y)
Output result:
tensor([2, 3, 4])
## Example
import torch
# For multi-dimensional arrays
sorted_seq = torch.tensor([[1,3,5],[2,4,6]])
values = torch.tensor([[1.5],[3.5]])
y = torch.searchsorted(sorted_seq, values)
print(y)
Output result:
tensor([, ])
* * Pytorch torch Reference Manual](#)
YouTip