Andrew Nikitin wrote:
> fn2 =: {. (|[EMAIL PROTECTED] , }.) ]
> 6!:2 't10=.>./ <:([EMAIL PROTECTED] #@(fn2^:(<_))@:>:@A. i.) 10'
> 83.2398
> t10
> 38
>
Andrew:
I somehow missed your solution in my inbox: I've just see it in the
archives. Mine is obviously very similar, but was separately conceived.
> Also, they say "first 30 permutatuons" and in test file those permuataions
> are listed in some strange order which I do not have any motivation to
> decode. It's not lexicographic, that is for sure. Probably, some of the
> various "least change".
>
The issue is that you are supposed to implement a particular algorithm,
and the first 30 permutations is an artifact, presumably to check that you
really used the algorithm. I ignored this.
The problem is that the site administrators regard the algorithm as
language-neutral, when it is not. In particular, J will lose out if it is
forced to apply functions to small arguments.
Let me give an example of the non-neutrality of algorithms. Consider the
language I will euphemistically call BF as compared with C. Suppose you
had to implement an algorithm for calculating the sum of n numbers. The
Shootout people would regard the obvious algorithms as O(n). However, in
BF, adding a+b is O(a+b), since it stores a by incrementing 0 a times. If
you think this is bogus, consider arbitary precision integers in a
sensible language stored as an array of machine integers, where a+b is
O(log ab).
A more sensible contest would allow programmers to use algorithms
appropriate to their language. Specific algorithms are never
language-neutral.
Best wishes,
John
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
