[racket] Knuth's algorithm S

From: Neil Van Dyke (neil at neilvandyke.org)
Date: Sun Mar 9 14:34:05 EDT 2014

Manfred Lotz wrote at 03/09/2014 01:13 PM:
>  There is a certain way the
> algorithm should be implemented (given at the top of the page).

I haven't traced through the algorithm completely (and I have to run out 
the door now), but looks like you might be able to replace the 
"(vector-copy vec)" with "vec".  (At least for purposes of this 
benchmark of 100,000 iterations; for a reusable library, you might not 
do it that way, nor the way that rosettacode.org prescribes.)

On my computer, eliminating that one "vector-copy" from the code drops 
run time down to 0.6 of what it is with your change.  The output 
distribution looks the same to me.

Neil V.

Posted on the users mailing list.