Skip to content
Prev 33081 / 398506 Next

Ordering long vectors

On Sun, 8 Jun 2003, Thomas Lumley wrote:

            
Thanks. This is really surprising: it is *much* faster to break ties by a 
second condition than not breaking them. I think it should be mentioned 
in the help. And could 'order/sort' be modified to check for 'tieness'? 
But I guess the the overhead would be too heavy.

(if (length(unique(x)) < alpha * length(x)) then .... else ....)

G?ran