Thread Subject: Queue hung again....

The queue is hung again on a code with some NaNs. The contest team
should consider giving us some way to easily contact them about these
hangs.....

So my question is:

Are the solutions that are being posted / winning the contest
actually the best solutions? I ran the best solution (at the time)
against the test suite that was provided and it actually increased
the score of submitting a solution of all zeroes. That blows my
mind. Shouldn't this contest be done for the sake of creative
algorithm design instead of tweaking to obtain a non-meaningful
score??

Not to say that I've come up with anything .. :) But I would like to
think that the winning solution would work the best for ALL test
cases.

Luigi

 the cyclist wrote:
>
>
> Luigi Sorbara wrote:
>>
>>
>> This may be a naive question. But all this mention of probe,
> etc,
>> does this mean that the test suite being used it static? Or am
I
>> missing what's going on here?
>>
>> Thanks
>
> The test suite used for scoring entries is indeed static; it is
> distinct from the test suite sent out in the file exchange for
> "home
> use".
>
> That is the reason it is possible to optimize certain constants in
> nonintuitive ways, and why certain sequences of pseudorandom
> numbers
> are better than others.

Nick Howe wrote:
>
>
> Sergey Yurgenson (SY) wrote:
>>
>>
>> At least this way you can get you name out as the “most active
>> participant” :-)
>>
>> On another topic:
>> What do people think about using random number generators in
> code?
>> Seems to me one always can “improve” code by inserting rand and
>> submitting it
>> zillion time hoping for randomly better score.
>> I just want to know if people consider it be inside boundaries
of
>> “gentlemen rules”
>>
>
> Assuming this contest works like past ones, every entry is run with
> a
> fixed pseudorandom sequence. So unless you change the code, you
> get
> the same result every time even using rand.
  

True, although you can 'seed' the random number generator with
different starting values. I already tried this in a series of tests
and it did indeed change the score. Not sure I want to go through
the trouble of finding the right seed though....

Alan Chalker wrote:
>
>

>
> Don Antonio did this last contest as well, so his automated
> submitter
> clearly works well. Perhaps this is just a way of trying to
> increase
> his name's page rank in google or something like that?
  

I vote for change, typing in the letters like in the newsgroup might
help.
I'd like to see a way to limit submissions per hour but I don't know
a good way to do it.

I laughed out loud when one of the queue fillers complained about the
stability of the server in one of his submission titles.