Lower Triangular Mask

Returns a boolean matrix that is TRUE on and below the given diagonal, mirroring base R's lower.tri(). Use nv_tril() to zero out the other triangle of an existing array instead.

Usage

nv_lower_tri(shape, diagonal = -1L, device = NULL)

nv_lower_tri_like(like, diagonal = -1L, shape = NULL, device = NULL)

Arguments

shape

(integer())
Shape of the result: exactly two axis sizes, since the result is a matrix.

diagonal

(integer(1))
Diagonal offset, with the same meaning as in nv_tril(). The default -1 excludes the main diagonal, matching lower.tri(); use 0 to include it, matching lower.tri(diag = TRUE).

device

(NULL | character(1) | device)
The device the data lives on, given either as:

• a device string naming the platform (e.g. "cpu", "cuda", "cuda:<n>"), which is resolved against the backend in use, or

• a device object as returned by nv_device(): a PJRTDevice for the "pjrt" backend or a quickr_device for the "quickr" backend. Because a device object is backend-specific, it also determines the backend.

The default (NULL) uses default_device().

like

(AnvlArray)
Existing array whose attributes are used as defaults (only for nv_lower_tri_like()).

Value

(arrayish)
Has the given shape and boolean data type. It is a mask over positions, so no array data enters it – pass it to nv_ifelse() or multiply by it to use it.

See also

nv_upper_tri(), nv_tril(), prim_iota() for the underlying primitive.

Examples

# a boolean mask, whatever the array it is later used with
nv_lower_tri(c(3, 3))
#> AnvlArray
#>  0 0 0
#>  1 0 0
#>  1 1 0
#> [ CPUbool{3,3} ] 
nv_lower_tri(c(3, 3), diagonal = 0L)
#> AnvlArray
#>  1 0 0
#>  1 1 0
#>  1 1 1
#> [ CPUbool{3,3} ] 
x <- nv_fill(0, shape = c(3, 3))
nv_lower_tri_like(x)
#> AnvlArray
#>  0 0 0
#>  1 0 0
#>  1 1 0
#> [ CPUbool{3,3} ]