Permute the ranks of a tensor — permute • tidytensor

Permute the ranks of a tensor, for example to convert between "channels first" and "channels last" representations.

Ranknames are respected for both inputs and return values.

permute(tensor, ..., .dots = NULL)

Arguments

tensor: the tidytensor permute.
...: ranknames or integers to permute by (quoted or unquoted).
.dots: character or integer vector to permute by.

Value

a new tidytensor.

Details

The rank parameter may be an integer numeric vector (for permuting by index), or character vector (for permuting by rankname).

Examples

# shape [20, 26, 26]
t <- as.tidytensor(array(rnorm(20 * 26 * 26), dim = c(20, 26, 26)))
ranknames(t) <- c("sample", "row", "col")
print(t)
#> # Rank 3 tensor, shape: (20, 26, 26), ranknames: sample, row, col
#> |  # Rank 2 tensor, shape: (26, 26)
#> |        1.83  -0.721  -1.42  -0.864   -2.15    1.01  ... 
#> |        1.13  -0.397  0.998   -1.08  -0.199  0.0444  ... 
#> |        1.37   -1.01  0.865  0.0816   0.921  -0.728  ... 
#> |        -0.6    1.14  -1.59    2.07    2.02   -1.05  ... 
#> |      0.0928   -1.11  -3.21    3.37   -2.21  -0.721  ... 
#> |       -1.66   0.853  0.843  -0.116   -1.68  -0.374  ... 
#> |         ...     ...    ...     ...     ...     ...  ... 
#> |  # ...

t2 <- permute(t, col, sample, row)
t2 <- permute(t, 3, 1, 2)
t2 <- permute(t, .dots = c(3, 1, 2))
t2 <- permute(t, .dots = c("col", "sample", "row"))