|Perl: the Markov chain saw|
I think there are a variety of improvements that could be made to the style of the code.
Firstly, initialising @a and @b can be done rather more simply:
.. which removes the need for the variable $n
Secondly, the @ready array is being built up an element at a time, starting from element zero. So it would be simpler to remove the index calculations, and just use push:
$kk is used to count down from the start value to 0, and I'd find it clearer to express that directly at the head of the loop:
It also seems odd that you declare @ready with my(), but none of the other variables. While writing everything with use strict is a good habit to get into, since there are no function calls in this code nor variables declared inside blocks there isn't much to be gained here. However consistency is always good, and you should either declare all your variables that way or none of them.
I don't understand the algorithm you are using at all: when I run the code for small values of $nn I get many warnings, and the results include many duplications, odd spacing, and some zeros. Also, this algorithm always produces exactly 3n results, which is too many for small values of n and too few for larger values. (The first example I noticed of a missing partition was for n = 6 = 2 + 2 + 2.)
If I were attempting to produce an algorithm for this, I'd be inclined to start off with the assumption that I'd want an iterator, so I'd consider first how to define a canonical ordering for a partition (eg with the numbers sorted in descending order), and then consider how given one partition I could generate the next one.
This is fairly easy to do if you define the iterator function to take an additional parameter, but I don't want to give too much away here in case you still want the fun of working it out for yourself.