Evenly Spaced Sequence

Creates a 1-D array with length_out evenly spaced values from from to to (both inclusive), like R's seq(from, to, length.out = length_out).

The spacing (to - from) / (length_out - 1) is generally not a whole number, so the result is a float.

nv_linspace_like() is a variant where dtype and device default to those of like.

Usage

nv_linspace(from, to, length_out, dtype = NULL, device = NULL)

nv_linspace_like(like, from, to, length_out, dtype = NULL, device = NULL)

Arguments

from, to

(numeric(1))
First and last value of the sequence. to may lie below from, in which case the values decrease.

length_out

(integer(1))
Number of values to generate. Must be at least 1; for length_out = 1 the result is from.

dtype

(NULL | character(1) | DataType)
Data type of the result. Must be a float data type; NULL (default) uses the default float data type (see default_dtypes()), since the spacing is fractional. For nv_linspace_like(), NULL uses dtype(like), which must then be a float too.

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. It must belong to the active backend (active_backend()); a device of another backend is an error.

The default (NULL) uses default_device().

like

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

Value

(arrayish)
Has dtype and shape length_out.

See also

nv_seq() for consecutive integers, nv_iota() for values increasing along an axis of any shape, dtype_categories for the data type categories.

Examples

nv_linspace(0, 1, length_out = 5L)
#> AnvlArray
#>  0.0000
#>  0.2500
#>  0.5000
#>  0.7500
#>  1.0000
#> [ CPUf32{5} ] 

# to below from counts down
nv_linspace(1, 0, length_out = 3L)
#> AnvlArray
#>  1.0000
#>  0.5000
#>  0.0000
#> [ CPUf32{3} ] 

# length_out = 1 gives from alone
nv_linspace(2.5, 10, length_out = 1L)
#> AnvlArray
#>  2.5000
#> [ CPUf32{1} ] 

# the data type must be a float; convert afterwards for integers
nv_convert(nv_linspace(0, 10, length_out = 5L), "i32")
#> AnvlArray
#>   0
#>   2
#>   5
#>   7
#>  10
#> [ CPUi32{5} ] 

# nv_linspace_like() takes the data type and device from an existing array
x <- nv_array(c(1, 2, 3), dtype = "f64")
nv_linspace_like(x, 0, 1, length_out = 3L)
#> AnvlArray
#>  0.0000
#>  0.5000
#>  1.0000
#> [ CPUf64{3} ]