November 2010 - Haskell-Cafe

hmatrix's fitModel function crashes ghc(i)
by Vivian McPhail 08 Nov '10

08 Nov '10

> Message: 29 > Date: Sat, 6 Nov 2010 13:22:10 +0100 > From: Roel van Dijk <vandijk.roel(a)gmail.com> > Subject: [Haskell-cafe] hmatrix's fitModel function crashes ghc(i) > To: Haskell Caf? <haskell-cafe(a)haskell.org> > Message-ID: > <AANLkTim5eGSL_BZ+RUv-=d-Z3DB65Sc8o=cKQLTcyKT4(a)mail.gmail.com> > Content-Type: text/plain; charset=UTF-8 > > Hello, > > I would like to use hmatrix to do some function fitting with the > Levenberg Marquardt algorithm. As an example I would like to fit the > very simple function "f x = a*x + b" on some data points. The problem > is that executing the 'fitModel' function crashes GHC(i) with a > segmentation fault. This makes debugging difficult. Can anyone spot > what I am doing wrong? Given all the lists of Double's it seems very > easy to make an error regarding the number of arguments with the model > function or the derivative. > > Try to evaluate the 'test' function in the small program listed below. > I would expect an output of [1, 0] (y = 1*x + 0) instead of a > segmentation fault. > > Relevant versions: > - hmatrix-0.10.0.0 > - gsl-1.14 > - ghc-6.12.3 (64 bit) > > Is that the 64 bit Linux ghc? I think the problem is with the GSL random number generation through GHCi. Try: > module Test where > > Import Numeric.Container(RandDist,randomVector) > > seed = 0 > size = 100 > > main = putStrLn $ show $ randomVector seed Gaussian size This should work when compiled with `ghc --make` and crash when invoked in `ghci`. I think the problem is with linking static data in GHCi on x86_64. Hope this helps. I seem to recall there might be a ghc trac ticket related to this but a quick search turned up nothing. Cheers, Vivian

2 2

ANNOUNCE: arrow-list. List arrows for Haskell.
by Sebastiaan Visser 07 Nov '10

07 Nov '10

Hi all, Live from the Hackaton in Ghent, Belgium, I present the first release of the arrow-list[1,2] package. List arrows are a powerful tool when processing XML, building query languages and lots of other domains that build on functions that might return more than one value as their output. This package is inspired by the arrow combinators provided by the HXT package, but in my opinion list arrows deserve to be on Hackage on their own. Cheers, Sebastiaan [1] http://hackage.haskell.org/package/arrow-list [2] https://github.com/sebastiaanvisser/arrow-list ------------------------------------------------------------------------ (package description) List arrows for Haskell. This small Haskell library provides some type class, types and functions to work with list arrows. List arrows represent computations that may return multiple outputs. Making functions that return lists an instance of both the `Category` and `Arrow` type classes allow you to easily compose multiple computations into one with standard building blocks. This package provides: - A type class `ArrowList` for embedding functions that produce a list of outputs into _some_ list arrow. - A list of utility functions for working with list-arrows, these functions are based on the `ArrowList` type class so they are not tied one specific instance. - A concrete list arrow type that is implemented as a `Kleisli` arrow over the `ListT` list monad transformer. In short, you can both build pure list arrows and list arrows that produce tributary effects. - Not list arrow specific: A type class `ArrowKleisli` for embedding monadic computations into an arrow.

2 4

hmatrix's fitModel function crashes ghc(i)
by Roel van Dijk 07 Nov '10

07 Nov '10

Hello, I would like to use hmatrix to do some function fitting with the Levenberg Marquardt algorithm. As an example I would like to fit the very simple function "f x = a*x + b" on some data points. The problem is that executing the 'fitModel' function crashes GHC(i) with a segmentation fault. This makes debugging difficult. Can anyone spot what I am doing wrong? Given all the lists of Double's it seems very easy to make an error regarding the number of arguments with the model function or the derivative. Try to evaluate the 'test' function in the small program listed below. I would expect an output of [1, 0] (y = 1*x + 0) instead of a segmentation fault. Relevant versions: - hmatrix-0.10.0.0 - gsl-1.14 - ghc-6.12.3 (64 bit) Small program: module Test where -- from base: import Control.Arrow ( second ) import Control.Applicative ( pure ) -- from hmatrix: import Data.Packed.Matrix ( Matrix ) import Numeric.GSL.Fitting ( FittingMethod(LevenbergMarquardt), fitModel ) -- input list of (x, y) pairs, output coefficients of "f x = a x * b" fitLinear :: [(Double, Double)] -> ([Double], Matrix Double) fitLinear samples = fitModel 1 1 10 (linearModel, linearDer) (map (second pure) samples) [0, 0] linearModel :: [Double] -> Double -> [Double] linearModel [a, b] x = [a*x + b, 0] linearModel _ x = error "wrong arguments" linearDer :: [Double] -> Double -> [[Double]] linearDer [_, _] x = [[x, 0]] linearDer _ _ = error "wrong arguments" test = fitLinear [(0,0), (1,1), (2,2)]

2 2

Fwd: [Haskell-cafe] DSL libraries
by Dupont Corentin 07 Nov '10

07 Nov '10

Nobody had the compilation messages I had? ---------- Forwarded message ---------- From: Dupont Corentin <corentin.dupont(a)gmail.com> Date: Tue, Nov 2, 2010 at 2:30 PM Subject: [Haskell-cafe] DSL libraries (Was: Map constructor in a DSL) To: steffen <steffen.siering(a)googlemail.com>, haskell-cafe(a)haskell.org Hello Steffen, can you compile AwesomePrelude? I've got error messages (see below). By the way, I've looked at some DSLs made in Haskell, if I don't mistake there are lots of similarities between them. There similarities could be put in a library to help the implementors of a DSL, more or less like AwesomePrelude. Is there already packages on Hackage dealing with that? Cheers, Corentin AwesomePrelude compilation error: I'm using GHC 6.12.1. > cd tomlokhorst-AwesomePrelude-9819315 > cabal install (...) Warning: Lang.Haskell: Instances of type and data families are not yet supported. Instances of the following families will be filtered out: H Then, when trying with a very simple GATD in GHCI, I've got: <interactive>: HSAwesomePrelude-0.1.0.o: unknown symbol `AwesomePreludezm0zi1zi0_ CompilerziLiftDefinitions_inline_closure' ghc: unable to load package `AwesomePrelude-0.1.0' On Thu, Oct 28, 2010 at 2:02 PM, steffen <steffen.siering(a)googlemail.com>wrote: > I think you would love to have a look at AwesomePrelude[1] or a fork > of AwesomePrelude using associated types[2] > Some more background information by Tom Lokhorst [3][4]. > > [1] http://github.com/tomlokhorst/AwesomePrelude > [2] http://github.com/urso/AwesomePrelude > [3] http://tom.lokhorst.eu/2009/09/deeply-embedded-dsls > [4] http://tom.lokhorst.eu/2010/02/awesomeprelude-presentation-video > > On 28 Okt., 12:09, Dupont Corentin <corentin.dup...(a)gmail.com> wrote: > > Thank you for your rich responses. > > > > Indeed I think I miss some thinks in my DSL, that would make things > easier > > to deal with lists and first class functions. > > I don't really know what for now. > > Perhaps a List Constructor? Or a constructor on functions like yours > Ryan? > > EAp :: Exp ref (a -> b) -> Exp ref a -> Exp ref b > > It's from which DSL? It is accessible on the net? > > > > Chris suggested me that I can only define the Foldr constructor and > deduce > > Map from it. > > But maybe I have to add a List constructor for that. > > > > But in the suggestions from Ryan and Brandon I don't understand why I > should > > add an extra type parameter and what it is! > > > > Steffen: Wow nice. I'll integrate that ;) > > > > I'm also looking at the Atom's DSL to get inspiration. > > Something I don't understand in it is that it has two languages, on > typed: > > > > data E a where > > VRef :: V a -> E a > > Const :: a -> E a > > Cast :: (NumE a, NumE b) => E a -> E b > > Add :: NumE a => E a -> E a -> E a > > etc. > > > > And, along with it, an untyped counterpart: > > > > -- | An untyped term. > > data UE > > = UVRef UV > > | UConst Const > > | UCast Type UE > > | UAdd UE UE > > etc. > > > > What that for? What's the use of having beautiful GADT if you have to > > maintain an untyped ADT aside?? > > > > Corentin > > > > _______________________________________________ > > Haskell-Cafe mailing list > > Haskell-C...@haskell.orghttp:// > www.haskell.org/mailman/listinfo/haskell-cafe > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

3 2

ANNOUNCE: language-cil
by Tom Lokhorst 07 Nov '10

07 Nov '10

Hi all, I've just uploaded language-cil [1], a library for manipulating a Common Intermediate Language (CIL) [2]. CIL is the lowest level language used by Microsoft .NET and Mono on Mac an Linux. With language-cil, you can construct an AST and pretty print it to concrete syntax. The assembler `ilasm`, can create a .exe file which can be run by Mono or .NET. The package is not finished, it only supports a subset of the full language, but it's already usable. For more information, as well as an example to quickly get started, see the github page [3]. Cheers, Tom Lokhorst [1]: http://hackage.haskell.org/package/language-cil [2]: http://en.wikipedia.org/wiki/Common_Intermediate_Language [3]: https://github.com/tomlokhorst/language-cil#readme

1 0

Review request for my baby steps towards a "platform independent interactive graphics" using VNC
by C K Kashyap 06 Nov '10

06 Nov '10

Hi, I started with the implementation of a VNC server library intended to be used as a library for rendering graphics and interacting with the user(mouse/keyboard). I'd appreciate it very much if I could get some feedback on my approach to binary parsing and Haskellism. Also, any reference/suggestion on how I could go about using a state machine to deal with the RFB protocol. http://hpaste.org/41131/vnc_server It's really early - but just wanted to get some advice on the approach. -- Regards, Kashyap

2 2

internship opportunities in France
by Lorenzo Fundaró 06 Nov '10

06 Nov '10

Hello folks ! I am a Computer Science student looking for an internship of 6 months here in France. Does anybody know of any company working with Haskell ? Thanks in advance :D, Lorenzo Fundaró García

4 4

Most popular haskell applications
by Bulat Ziganshin 06 Nov '10

06 Nov '10

Hello haskell, people, are you know haskell apps that has more than 50k downloads per month (or more than 25k users) ? -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

4 6

What do you call Applicative Functor Morphism?
by roconnor＠theorem.ca 06 Nov '10

06 Nov '10

An applicative functor morphism is a polymorphic function, eta : forall a. A1 a -> A2 a between two applicative functors A1 and A2 that preserve pure and <*>: eta (pure c) = pure c eta (f <*> x) = eta f <*> eta x What do you guys call such a thing? My leading candidate is "idomatic transformation". -- Russell O'Connor <http://r6.ca/> ``All talk about `theft,''' the general counsel of the American Graphophone Company wrote, ``is the merest claptrap, for there exists no property in ideas musical, literary or artistic, except as defined by statute.''

7 7

Flagstone problem
by michael rice 06 Nov '10

06 Nov '10

Hi, I've been looking at a "flagstone problem," where no two adjacent n-tuples can be identical. I solved the problem with Icon using an array of stacks and was going to explore how to do it in Haskell when I saw another way to do it explained in the same text. Just count the ones between the zeros in a(n) to get b(i), a non-repeating sequence of zeros, ones, and twos. An a(n) is 0 if the number of one bits in the binary representation of n is even, otherwise odd, and the initial a(n) must be even. n a(n) b(i) 20 = 010100 0 21 = 010101 1 2 22 = 010110 1 23 = 010111 0 0 24 = 011000 0 25 = 011001 1 2 26 = 011010 1 27 = 011011 0 1 28 = 011100 1 29 = 011101 0 0 30 = 011110 0 31 = 011111 1 2 32 = 100000 1 33 = 100001 0 0 34 = 100010 0 1 35 = 100011 1 36 = 100100 0 ========= Here's my Haskell code ======== import Data.Bits import Data.List flagstone n = foldl (++) "" (take n (map show (f $ group [if even y then 0 else 1 | y <- [bitcount x | x <- [20..]]]))) bitcount :: (Integral t) => t -> t bitcount 0 = 0 bitcount n = let (q,r) = quotRem n 2 in bitcount q + r f [] = [] f ([0]:xs) = f xs f ([0,0]:xs) = 0 : f xs f ([1]:xs) = 1 : f xs f ([1,1]:xs) = 2 : f xs ========= My question ======== A further exercise in the text: "Exercise 5.-(a) Define a(n) as the sum of the binary digits in the binary representation of n. Define b(i) as the number of a's between successive zeros as before. Then T = b(1) b(2) b(3) b(4) ... gives an infinite sequence of *seven* symbols with no repeats. (b) Write a routine to generate a sequence for seven colors of beads on a string with no repeats." I may be misreading, but does this make any sense? There's a reference to The American Mathematical Monthly, June-July 1963, p. 675, by C. H. Braunholtz. Michael

3 4