[Haskell-cafe] Re: [Haskell] ANNOUNCE: An index-aware linear algebra library in Haskell

[Frederik Eaton]

Hi Alberto,

Those are good questions, I've added some examples which hopefully
clarify the situation. Input and output of vectors is not a strong
point of the library, but I don't think there is a good alternative to
the way I do it.

  http://ofb.net/~frederik/futility/src/Vector/read-example.hs

(also, your example exposed some missing functionality. I've added
three new functions; in addition to listMat, now there are listMatCol,
listMatRow, and listMatSquare. Hopefully these should cover almost all
use cases.

  http://ofb.net/~frederik/futility/src/Vector/Base.hs
)

By the way, here is how I would download and run the thing, although
you seem to have figured it out:

$ wget http://ofb.net/~frederik/futility/futility-devel.tar.gz
$ tar -xvzf futility-devel.tar.gz
$ cd futility-devel/
$ ghc -fth --make Vector/read-example.hs -o read-example
$ ./read-example
<# 11.0, 23.0; 14.0, 30.0; 15.0, 33.0; 18.0, 40.0 #>
<1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0>
<1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0>
<# 1.0, 5.0; 2.0, 6.0; 3.0, 6.0; 4.0, 7.0 #>
<# 11.0, 23.0; 14.0, 30.0; 15.0, 33.0; 18.0, 40.0 #>
<# -3.0, -5.0; -5.0, -7.0 #>
<# 7.0, 10.0; 15.0, 22.0 #>

As for your questions:

> Vector/examples.hs:35:11:
>     GHC stage restriction: `x'
>       is used in a top-level splice, and must be imported, not defined locally
>     In the first argument of `dAM', namely `x'
>     In the definition of `m1': m1 = $[splice](dAM x)

This is a shortcoming of Template Haskell - it will not let you call a
function from a splice if that function is defined in the same file as
the splice. It should be possible to remove this shortcoming, but I
don't know what is planned.

> m2 = listMat x

This is not how listMat is used, see the example file above.

listMat takes a list of lists and a function, and passes the matrix
version of the list of lists to the function.

> I would also like to create a matrix from a data file:

See 'v3' in the example.

Cheers,

Frederik

On Sun, Apr 16, 2006 at 05:06:55PM +0200, Alberto Ruiz wrote:
> It is really counterintuitive! I will study carefully your library and the 
> "Implicit Configurations" paper. Using static dimension checking we can write 
> very solid code for matrix computations...
> 
> However, I don't know how to write some definitions. For instance, this is ok:
> 
> m = $(dAM [[1,2,3]])
> 
> but with:
> 
> x = [[1,2,3]] :: [[Double]]
> m1 = $(dAM x)
> m2 = listMat x
> 
> main = do
>     print m1
>     print m2
> 
> I get:
> 
> Vector/examples.hs:35:11:
>     GHC stage restriction: `x'
>       is used in a top-level splice, and must be imported, not defined locally
>     In the first argument of `dAM', namely `x'
>     In the definition of `m1': m1 = $[splice](dAM x)
> 
> Vector/examples.hs:40:10:
>     Inferred type is less polymorphic than expected
>       Quantified type variable `m' escapes
>       Quantified type variable `n' escapes
>       Expected type: (v (L m, L n) -> w) -> t
>       Inferred type: (forall n1 m1.
>                       (ReflectNum n1, ReflectNum m1) =>
>                       v (L m1, L n1) -> w)
>                      -> w
>     In the first argument of `print', namely `m2'
>     In the result of a 'do' expression: print m2
> 
> 
> I would also like to create a matrix from a data file:
> 
> main = do
>     let m1 = $(dAM [[1,2],[3,4::Double]])
>     s <- readFile "data.txt"
>     let list = read s :: [[Double]]
>     --let m2 = $(dAM list)
>     let m2 = listMat list
>     print $ m2 *> trans m1
> 
> But I get a similar error. Perhaps I must provide information about the 
> expected dimensions, but I don't know how to do it.
> 
> --
> Alberto
> 
> On Saturday 15 April 2006 22:09, Frederik Eaton wrote:
> > Yes, certainly... Otherwise the library would not be much use! If it
> > seems counterintuitive, as it did to me at first, you should check out
> > the "Implicit Configurations" paper, which uses modular arithmetic as
> > an example. My version of their code is in
> >
> > http://ofb.net/~frederik/futility/src/Prepose.hs
> >
> > The function I mainly use is:
> >
> > reifyIntegral :: Integral a => a -> (forall s. ReflectNum s => s -> w) -> w
> >
> > which turns an integral value into a type of the ReflectNum class
> > which represents that value, and calls the provided polymorphic
> > function with a dummy value (actually 'undefined') of that type; then
> > returning the function's result.
> >
> > Frederik
> >
> > On Sat, Apr 15, 2006 at 06:14:44PM +0200, Alberto Ruiz wrote:
> > > On Friday 14 April 2006 17:02, Frederik Eaton wrote:
> > > > An index-aware linear algebra library in Haskell
> > >
> > > Excellent work!
> > >
> > > Is it possible to create a vector or matrix whose size is not known at
> > > compile time?
> > >
> 

-- 
http://ofb.net/~frederik/