A normalized kernel is one that, if given a data matrix that is all 1s, will result in 1. This function applies a scalier multiple to the kernel to make it normalized, if possible. Otherwise, it optionally warns and returns the input kernel.
normalize_kernel(k, warning_enabled = TRUE)[matrix] The kernel.
[logical] Whether to unable warnings, default
to TRUE.
The normalized kernel, a matrix.
kernel <- distance_kernel(vertical_radius = 2, horizontal_radius = 2)
kernel_norm <- normalize_kernel(kernel)
sum(kernel_norm)
#> [1] 1