Month: August 2009

The Mersenne Twiser is a random number generator that has a lot of applications, particularly in finance. After discovering that there was no Clojure implementation at Wikipedia I decided to give it a try as my first attempt at something useful in Clojure. As it turns out it’s problably not a good candidate to be implemented in a functional language because the whole thing requires modifying a mutable array for every call to (genrand).

I’m not very excited by the solution because it doesn’t seem very lispy. It much more resembles the reference implementation, and gives the same results as it for all the tests I tried.

I’m sure it could be better though.

If it’s of any use to anyone you can find it here.

Yesterday I watched Dan Weinreb talking about ITA, Lisp, and well, really, really complex stuff 😉

The kind-of conclusion that he drew on the future of Lisp is that Clojure is the next Common Lisp. I’ve been dodging Clojure for about a year now. I bought the book from The Pragmatic Programmers when the book was in beta and eagerly followed the examples and decided it had promise. But, I figured that if I needed to really learn it (and by that I mean use it, not just talk about using it) it would have to come to me. It would do this, I reasoned, by being hard-to-ignore.

Well there seem to be more than enough smart-folks on-board now (like: Eric Normand, Bill Clementson to name only two) so I guess I better not miss the party. Not because I’m smart but because if I don’t make too much noise I can blend in and no one will notice I’m there.

So I came to the conclusion that I should try and use it for ad-hoc things that might cross-my-dome and are hard(er) to solve in non-functional languages.

For instance, this very afternoon I wanted to know how many possible hand distributions by suit there are in the card game Bridge. So four possible distributions might be:

• All 13 hearts
• 12 hearts, one diamond
• 11 hearts, one diamond, one club
• 10 hearts, one diamond, one club, one spade

The question is how many total distributions are possible?

I banged my head on the table-hard trying to figure out the answer to this. At first it seemed like a simple counting problem, but if it is I’m too simple to see it. Then I wondered if it could be an additive patitioning problem, but ordering is important so I don’t think it is. It didn’t feel NP complete. I know one thing though, at this late hour it might as well be.

1:57 bridge=> (count (for [spades (range 0 14) 
                                hearts (range 0 14) 
                                diamonds (range 0 14) 
                                clubs (range 0 14) 
                                :when (= 13 (+ spades hearts diamonds clubs))] 
                                [spades hearts diamonds clubs]))
560

Functional programming rocks.