Declare a multiscale Gaussian spatial model for GazeWeave — gaze_gaussian_mixture • eyesim

The resulting spatial cost is the negative log of a normalized mixture of Gaussian overlap kernels. Mixture weights are normalized to sum to one.

Usage

gaze_gaussian_mixture(
  sigmas,
  weights = NULL,
  unit = c("native", "px", "deg", "normalized")
)

Arguments

sigmas: Positive Gaussian bandwidths.
weights: Optional non-negative mixture weights.
unit: Coordinate unit for `sigmas`.

Value

A `gaze_spatial_spec` object.

Examples

# Two spatial scales, in the same unit as the fixation coordinates.
gaze_gaussian_mixture(c(30, 60), unit = "px")
#> $sigmas
#> [1] 30 60
#> 
#> $weights
#> [1] 0.5 0.5
#> 
#> $unit
#> [1] "px"
#> 
#> attr(,"class")
#> [1] "gaze_spatial_spec" "list"