#haskell - Fri 6 Apr 2007 between 00:00 and 00:08

NY Lost Funds



sorear.
dmead: pong
narainis there any point defining something to be a monad if i may never need to use it as one?
dmeadping
chessguy@quote backwards
lambdabotNo quotes match. My pet ferret can type better than you!
procyon112chessguy: I'm just saying, that it may be possible to express the entire algorithm as a lazy list sorted by fitness.. at which point, it's more like fuzzy unification than GP.
monochrommonad is rather special - you will most likely want to use it as a monad in the near future. I would say define the monad.
chessguyprocyon112: ohhhhhh, now i get what you're saying. sorry
(light bulb goes off)
dmeadit's just a jump to the left
narainmonochrom: ok, now i'll have to be on the lookout for opportunities for using monadic operations
chessguyprocyon112: mmm, i don't think that's likely to be very useful
monochromheh, sounds like I've spewed a self-fulfilling prophecy!
dmeadis a forumla a valid parameter to pass to a predicate?
procyon112chessguy: so, being aware that we may be able to factor out primary tenets of what makes GP GP, we need to stop at that point, leave GP in place, and do the factorization on the side, as an option... otherwise our GP lib will be some other form of problem solving that is isomorphic to GP, but not useful to people who just want GP.
chessguyi don't think it is isomorphic
procyon112chessguy: don't think it's isomorphic to what?
chessguyto GP
procyon112chessguy: no.. my question was on the other side of the arrow :) You don't think GP is isomorphic to....
phoniqfoo a b = a ** b
foo :: forall a. (Floating a) => a -> a -> a
chessguyugh
phoniqhow can i restrict this so that the first arg is integral?
chessguyi don't think GP is isomorphic to your zipWith think
*thing
int-ephoniq: that will make them all integral
chessguy?hoogle Integral
lambdabotPrelude.Integral :: class (Real a, Enum a) => Integral a
Prelude.fromIntegral :: (Integral a, Num b) => a -> b
int-ephoniq: because the types are equal
monochromdmead: in most logics no. but in most logics there is an expression language (function symbols, constants, vars) and any expression is fair game in predicate arguments, e.g., forall x. isChick( father(x) ), isChick predicate, father function
phoniqah
the type signature is just what ghci tells me
dmeadahh
phoniqwhat ought it be is the question... old man C is getting his ass kicked
chessguyprocyon112: your zipWith thing basically says to generate all possible programs in random order and find the best one you can (correct me if i'm wrong). that's very different from what GP does
phoniqthat's what i've managed to boil the problem down to.
procyon112chessguy: I haven't invented the zipWith, that's just wild speculation, but I think parts of GP can be expressed lazily, to the point where anyone using the code would not recognize it as GP anymore.
chessguyperhaps

Page: 2 9 16 23 30 37 44 51 58 65 72 79 86 

IrcArchive

NY Lost Funds