Skip to contents

Reductions over array margins.

g-reductions.Rd

g_sum accumulates 8- and 16-bit integers in i32 (32-bit unsigned in i64), so a sum does not wrap at the input type's range.

Usage

g_sum(x, dims = NULL, nan_rm = FALSE)

g_mean(x, dims = NULL, nan_rm = FALSE)

g_min(x, dims = NULL, nan_rm = FALSE)

g_max(x, dims = NULL, nan_rm = FALSE)

g_median(x, dims = NULL, nan_rm = FALSE)

g_count(x, dims = NULL)

Arguments

x

Numeric array.

dims

Integer margins to reduce, or NULL for all.

nan_rm

Skip NaN (nodata)?

Value

Reduced array (margins in dims dropped) or scalar.

Details

With nan_rm = TRUE, a slice that is entirely NaN reduces to the reduction's identity value: g_sum gives 0, g_min gives Inf, g_max gives -Inf, and g_mean / g_median give NaN. g_count counts non-NaN values, so an all-NaN slice gives 0.