In this article we describe the indexing operator for torch tensors and how it compares to the R indexing operator for arrays.
Torch’s indexing semantics are closer to numpy’s semantics than R’s. You will find a lot of similarities between this article and the
numpy indexing article available here.
Single element indexing for a 1-D tensors works mostly as expected. Like R, it is 1-based. Unlike R though, it accepts negative indices for indexing from the end of the array. (In R, negative indices are used to remove elements.)
<- torch_tensor(1:10) x 1] x[-1]x[
You can also subset matrices and higher dimensions arrays using the same syntax:
<- x$reshape(shape = c(2,5)) x x1,3] x[1,-1]x[
Note that if one indexes a multidimensional tensor with fewer indices than dimensions, one gets an error, unlike in R that would flatten the array. For example:
It is possible to slice and stride arrays to extract sub-arrays of the same number of dimensions, but of different sizes than the original. This is best illustrated by a few examples:
<- torch_tensor(1:10) x x2:5] x[1:(-7)]x[
You can also use the
1:10:2 syntax which means: In the range from 1 to 10, take every second item. For example:
Another special syntax is the
N, meaning the size of the specified dimension.
Note: the slicing behavior relies on Non Standard Evaluation. It requires that the expression is passed to the
[not exactly the resulting R vector.
To allow dynamic dynamic indices, you can create a new slice using the
slc function. For example:
is equivalent to:
slc(start = 1, end = 5, step = 2)]x[
Like in R, you can take all elements in a dimension by leaving an index empty.
Consider a matrix:
<- torch_randn(2, 3) x x
The following syntax will give you the first row:
And this would give you the first 2 columns:
By default, when indexing by a single integer, this dimension will be dropped to avoid the singleton dimension:
<- torch_randn(2, 3) x 1,]$shapex[
You can optionally use the
drop = FALSE argument to avoid dropping the dimension.
1,,drop = FALSE]$shapex[
It’s possible to add a new dimension to a tensor using index-like syntax:
<- torch_tensor(c(10)) x $shape x$shape x[, newaxis]$shapex[, newaxis, newaxis]
You can also use
NULL instead of
Sometimes we don’t know how many dimensions a tensor has, but we do know what to do with the last available dimension, or the first one. To subsume all others, we can use
<- torch_tensor(1:125)$reshape(c(5,5,5)) z 1,..] zz[..,
Vector indexing is also supported but care must be taken regarding performance as, in general its much less performant than slice based indexing.
Note: Starting from version 0.5.0, vector indexing in torch follows R semantics, prior to that the behavior was similar to numpy’s advanced indexing. To use the old behavior, consider using
<- torch_randn(4,4) x c(1,3), c(1,3)]x[
You can also use boolean vectors, for example:
c(TRUE, FALSE, TRUE, FALSE), c(TRUE, FALSE, TRUE, FALSE)]x[
The above examples also work if the index were long or boolean tensors, instead of R vectors. It’s also possible to index with multi-dimensional boolean tensors:
<- torch_tensor(rbind( x c(1,2,3), c(4,5,6) ))>3]x[x