in reply to Re^3: Picking a random item through probability
in thread Picking a random item through probability

Woaw, impressive :) So concise, yet so clear. Not to mention that it actually works too!

I read up about List::Util::first and I wish I knew about it earlier, that's a hell of an awesome subroutine (in fact the whole module is pretty cool, too).

At first I was a bit puzzled about the (L::U::f)->[0] thing, but of course - all elements are in fact two elements each, and at this point of the progress we only care about the first. Nifty how you put all that in just one line without becoming too obscure.

So much for flattering around, because there's still one part I don't quite understand. They do it in Weighted random numbers generator too. What does $pick dropping below zero have to do with anything? How does it work? I tried to figure it out by running the code mentally, but I just don't understand...

Update: also, let it be known it's no homework :) I write Perl just as an hobby. I'm not really into computer science or whatsoever - I mean, I understand something like this can't be hard and if it would've been a homework assignment I would be ashamed I couldn't figure it out by myself :)

Replies are listed 'Best First'.
Re^5: Picking a random item through probability
by Tanktalus (Canon) on Nov 23, 2006 at 22:14 UTC

    List::Util - yeah, it was other monks who introduced me to this module. I've made good use of it since ;-)

    Back to your question ... Let's say that the number we get is "5". Looking at this list, that should result in picking "baz". How do we do that? We look at the numbers: the first weight is one. So that would be a pick of 0. That's not it. So then the next option, 'bar', has a weight of 3. So that will cover picks 1-3. But we have 5, so that's not in range, either. Next option has a weight of 4 - numbers 4-7. We have 5. Got it.

    That's the straight-forward approach. The convoluted approach reverses this. We start with the first option, and discover that it has a weight of 1. So, now we look at the number - if the number were 0, this would be a match. If the number were 1, it isn't. So we could simply check if $pick < $weight and return 1, otherwise we just subtract the weight from our pick, and check the next number. However, since we're in a block, not a sub, we need to be a bit more tricky. So, we do a bit of algebraic magic, and we come up with $pick - $weight < 0, and we still want to subtract the weight before going on to the next one. Finally, we just do the subtraction first, and let $pick < 0 be our boolean return from the block.

[reply]
      While we're plugging modules, in addition to List::Util I recommend looking at List::MoreUtils (note the irregular naming, plural "Utils"). This is pretty useful, if only for the "any": 
       print "At least one value undefined"
        if any { !defined($_) } @list;

      [download]

      Which is not as clean as the syntax of the "any" in Damien Conway's Quantum::Superpositions , but it has the virtue of not requiring the use of a deranged Damien Conway module...

      I'm also a fan of: Hash::Util. Being able to lock the keys on a hash is a great trick to catch typos.

      And I see there's also a Hash::MoreUtils but that's one I haven't played with yet.

[reply]
[d/l]
      That does it for me. Thank you for your simple, straight-forward solution and explanation (or convoluted, then, but still... I was worrying it would take dozens of lines of code etc)
[reply]

In Section Seekers of Perl Wisdom