OpenEuphoria

Shuffle, Shuffle, what's the Ruffle?

• Posted by Al Getz <Xaxo at aol.com> Mar 09, 2001
• 397 views

For Jiri and others interested in shuffle algorithms:

Hi there again Jiri,

  Im happy to hear you are so emphatic about the coding
of various algorithms, most people i meet dont care
one way or the other.  It is nice to find people who care
about what they write and modify and such.

  Sorry for the misunderstanding, but i wasnt trying
to correct you, but rather simply posting another
solution, as i often do.
I was a little surprised though, as ive found your particular
coding to be of value to me, both in application and theory.
Your algorithm 'remove()' was very efficient, and when i asked
you to elaborate on it, you posted exactly what i was looking
for!  Its mostly out of that kind of respect for your programming
technique that i decided to look into this a little further.

Also, i always felt this forum was a good way to hash out
different approaches to code anyway, hearing different opinions
and adding them up.  Several minds working on the same problem
is much like several cpu's in a computer, gets the work done 
N times as fast.

Well, as it turns out, wasnt that code you originally posted
exactly what i posted over two years ago in response to
questions regarding shuffling?  That was an
algorithm written in QBasic by a twelve year old kid if
i remember right.  I also remember you getting peeved
because of all the talk about shuffling?  

Granted i should have looked it up before posting, as i
remembered what i considered near optimum code for
a shuffle function incorrectly -- im too often in a hurry,
and couldnt even find time for one of my usual salutations.
The one i had considered best was actually the one you originally
posted as well, but certainly not because the original is so much
faster, but rather because its slightly more statistically balanced
for a great number of shuffles.  Its not much, but it is better
for apps that will shuffle a great number of times.

It's to that end that i can therefore only offer a very slight
improvement to the original code, which turns out to be
just as statistically balanced as the first:

function shuffle_Al(sequence s)
     object temp
     integer j

     for i = 2 to length(s) do
         j = rand(i)
         temp = s[j]
         s[j] = s[i]
         s[i] = temp
     end for
     return s
end function


Perhaps this is the last word on shuffle?  Perhaps not.  In any case,
im sure there will be more newcommers to the list who are looking
for a shuffle algorithm--perhaps that suggestion to start a
standard library could start right here?

Good luck with your future programming,
--Al