There are time/space tradeoffs in sampling without replacement. The version in
monte-carlo takes space O(n) and time O(k), for all k. I chose this algorithm
instead of a streaming algorithm, with takes space O(k) and time O(n).
If k is much less than n, you can improve a bit. Here is a
Correction: the version in monte-carlo takes time O(n), but it only consumes k
random numbers. The streaming algorithm consumes n random numbers.
On Jun 14, 2011, at 12:04 PM, Patrick Perry wrote:
There are time/space tradeoffs in sampling without replacement. The version
in monte-carlo
Upon popular request, I've renamed the ieee package to ieee754. The new
hackage page is http://hackage.haskell.org/package/ieee754 .
On Sep 19, 2010, at 11:16 PM, Conrad Parker wrote:
On 20 September 2010 11:18, Patrick Perry patpe...@gmail.com wrote:
Given that IEEE is actually
ieee is a Haskell library for dealing with IEEE floating numbers. It
was originally written to make testing with floating point values less
painful. The library provides an approximate equality type class,
AEq, with approximate equality operator ~==. One property of ~==
is that nearby floating
Given that IEEE is actually a standards body and they have many
standards, wouldn't it be more appropriate to call this library
ieee754?
If it seems important to people, I'd be happy to change the name. I'm
not religious about these things. Will it clutter up hackage, though?
Patrick
con...@metadecks.org wrote:
On 20 September 2010 11:18, Patrick Perry patpe...@gmail.com wrote:
Given that IEEE is actually a standards body and they have many
standards, wouldn't it be more appropriate to call this library
ieee754?
If it seems important to people, I'd be happy to change
I needed real IEEE754 binary support for round-trip parsing, where
(for example) maintaining the particular bit pattern of a NaN is
important. For 99% of people, the unsafe method will work fine.
How does a C-style cast not preserve the bit pattern of a NaN? Again,
sorry if this is a stupid
monte-carlo is a haskell library providing a monad and associated monad
transformer for computing with quasi-random numbers. It provides a high-level
interface to the distributions supported by the GNU Scientific Library
(currently just uniform, normal, exponential, Poisson, Levy, but others
I'm having some trouble with QuickCheck2 and Control.Applicative.
Specifically, I have the following functions (slightly simplified):
copyVector :: IOVector - Vector - IO ()
freezeVector :: IOVector - IO (Vector)
I have a test, part of which looks like
monadicIO $ do
run $
Hi Kenneth,
I wrote a benchmark similar to yours using the haskell blas library I
wrote (latest version is on github at http://github.com/patperry/blas/tree/master
, an older version is on hackage).
The pure Haskell implementation is not very good, and the problem
seems to be repeated
Hi Cetin,
This is probably the easiest way:
(m - es)^2/es
listMatrix (2,2)
[0.6163270413689804,0.600918865334756,0.33217626255600896,0.3238718559921087
]
This will create 2 temporary arrays. Alternatively,
let res = listMatrix (shape m) [ (o-e)^2 / e | o - colElems m | e
- colElems
Well, I guess I am not the only one!
This blog show exactly what I am looking for!
http://quantile95.com/2008/10/31/ann-blas-bindings-for-haskell-version-06/
If you're looking to implement other linear algebra algorithms in
Haskell, there's enough at
http://github.com/patperry/lapack
In C, if you try to alloca too much memory, then the stack gets
overwritten and bad things happen. Does GHC exhibit the same behavior
with allocaArray and the like? Is there a way to find out how much is
safe to allocate?
Thanks in advance for the help,
Patrick
My two favorite functions in Haskell are unsafeInterleaveIO and
unsafeInterleaveST. I'm surprised there isn't a type class for
them. I just fixed this by adding the monad-interleave package to
hackage:
http://hackage.haskell.org/cgi-bin/hackage-scripts/package/monad-interleave
The
Thanks for your help, Duncan.
On Jan 12, 2009, at 6:10 AM, Duncan Coutts wrote:
Do the (ForeignPtr e) and the (Ptr e) point to the same thing? They
appear to be related because you dereference p but touch f.
It used to be the ForeignPtr was slower to dereference than a Ptr
and so
caching
I have the following code:
IOVector n e = IOVector !ConjEnum !Int (ForeignPtr e)! (Ptr e)! Int!
newtype Vector n e = IOVector n e
unsafeAtVector :: Vector n e - Int - e
unsafeAtVector (Vector (IOVector c _ f p inc)) i =
let g = if c == Conj then conjugate else id
in inlinePerformIO $
New version!
The blas package is a set of bindings to the BLAS (Basic Linear Algebra
Subprograms) library.
On Hackage:
http://hackage.haskell.org/cgi-bin/hackage-scripts/package/blas-0.7
What's new:
* Get rid of most functional dependencies in favor of type families.
There is one
Here's the haddock documentation; I'm not sure if Hackage honors {-#
OPTIONS_HADDOCK hide #-} when it displays the exposed modules:
http://quantile95.com/blas/
Patrick
___
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
Hi David,
The problem is with Hackage, not with haddock.
Patrick
On Jan 10, 2009, at 3:38 AM, David Waern wrote:
2009/1/10 Patrick Perry patpe...@stanford.edu:
Here's the haddock documentation; I'm not sure if Hackage honors {-#
OPTIONS_HADDOCK hide #-} when it displays the exposed modules
The reason I want to expose modules but hide the documentation is
because there are a lot of unsafeX functions I want to provide
access to, but which 99% of users don't care about. The array library
does the same thing.
Patrick
___
(for whom you are providing access)
will still have a possibility to do so.
2009/1/10 Patrick Perry patpe...@stanford.edu:
The reason I want to expose modules but hide the documentation is
because
there are a lot of unsafeX functions I want to provide access to,
but
which 99% of users don't care
I've released a new version of the monte-carlo packages for haskell.
Here are the highlights for monte-carlo:
Changes in 0.2:
* More general type class, MonadMC, which allows all the functions to
work
in both MC and MCT monads.
* Functions to sample from discrete distributions.
*
I agree with everyone else who has said that the better solution is
Lemmih's. It is simple, fast, and does not use much memory.
On the other hand, here is a more faithful implementation of what you
were trying to write. To use mutable arrays, you need to work in
either the IO or the ST
I've noticed that there is no longer an unsafeThawByteArray#
function in GHC.Prim.
Looking through the darcs history, I found a patch with description:
[project @ 2000-03-13 12:11:43 by simonmar]
simonmar**2313121144
Remove unsafeThawByteArray# primop (which was a no-op), and use
Hi Everyone,
I've uploaded a new version of the permutation library to hackage.
Here is the package description:
This library includes data types for storing permutations. It
implements pure and impure types, the latter which can be modified
in-place. The main utility of the
What is the situation regarding statically dimension-checked linear
algebra libraries? It seems that Frederik Eaton was working on one in
2006 (see the paper Statically typed linear algebra in Haskell), and
he produced the Vectro library from this, based on GSLHaskell.
Are there any more
New version of BLAS bindings out. Now with support for the ST monad!
This breaks backwards compatibility, unfortunately.
More info (and some sample code) here:
http://quantile95.com/2008/10/31/ann-blas-bindings-for-haskell-version-06/
Also, if you want to help, please let me know. There
Hi everyone,
I've started on bindings for the random number generators and random
distributions provided by the gsl. The package is available here:
http://hackage.haskell.org/cgi-bin/hackage-scripts/package/gsl-random
I've also written a monad and transformer for doing monte carlo
Thanks for the heads up, Don. I fixed the file in version 0.1.1.
I also changed the default CBLAS to the one that comes with the GSL.
Anyone who cares at all about performance will want to configure the
package to use the best CBLAS on their system. If this is ATLAS, make
sure you pass
'
_ - rands r2 n'
Patrick
On Aug 14, 2008, at 2:17 AM, Johan Tibell wrote:
On Thu, Aug 14, 2008 at 1:58 AM, Patrick Perry
[EMAIL PROTECTED] wrote:
variant :: Int - Gen a - Gen a
variant v (Gen m) = Gen (\n r - m n (rands r !! v'))
where
v' = abs (v+1) `mod` 1
rands r0 = r1
in r4'
Patrick
On Aug 14, 2008, at 4:57 AM, Johan Tibell wrote:
On Thu, Aug 14, 2008 at 12:15 PM, Patrick Perry
[EMAIL PROTECTED] wrote:
Actually, a much better solution is:
variant :: Int - Gen a - Gen a
variant v (Gen m) = Gen (\n r - m n (rands r v))
where
rands r0 0 = r0
I'm running into problems with generating an arbitrary function of
type Double - Double. Specifically, the following code:
{-# LANGUAGE PatternSignatures #-}
import Test.QuickCheck
import Text.Show.Functions
prop_ok (f :: Double - Double) =
f (-5.5) `seq` True
prop_bug (f :: Double -
The bug is in the variant function in QuickCheck. I replaced
variant :: Int - Gen a - Gen a
variant v (Gen m) = Gen (\n r - m n (rands r !! (v+1))
where
rands r0 = r1 : rands r2 where (r1, r2) = split r0
with
variant :: Int - Gen a - Gen a
variant v (Gen m) = Gen (\n r - m n (rands r !!
Hey everyone,
I've put together a new release of the Haskell BLAS bindings, now
available on hackage.
Here are the new features:
* Add Banded matrix data type, as well as Tri Banded and Herm Banded.
* Add support for trapezoidal dense matrices (Tri Matrix (m,n) e, where
m is not the same
Last month Anatoly Yakovenko published some disturbing numbers about
the Haskell BLAS bindings I wrote being significantly slower than
using plain C. I wanted to let everyone know that I've closed the
performance gap, and now for doing ten million dot products, the
overhead for using
Yeah, I think that's where most of the performance gains came from. I
also added a re-write rule for unsafeGet dot (since it doesn't matter
if the arguments are conjugated or not if the vectors are real) that
shaved off about a tenth of a second.
Patrick
On Jul 24, 2008, at 4:26 PM,
Sorry Darrin, the BLAS library only includes matrix multiplication and
solving triangular systems. To solve a general system, you would need
to use LAPACK, but there aren't any bindings for that library yet. I
would suggest you take a look at the hmatrix package, which includes a
lot
There were a few hiccups in the release of 0.4, so this version is to
make installation go a little smoother for people. There are now
installation instructions (the INSTALL file) and configuration
settings for some common CBLAS (ATLAS, MKL, vecLib, GSL). The default
is now to assume
ieee is a library that provides approximate comparison of floating
point numbers based, NaN-aware minimum and maximum, and a type class
for approximate comparisons.
This version fixes a bug in the comparison implementation for Maybe.
Wow, thanks for noticing, Alberto! For anyone interested, I put up a
formal announcement describing the bindings a little bit here:
http://quantile95.com/
I just registered the domain yesterday, so it may take a few days to
resolve the DNS magic. Here's the text of the announcement:
I’m
40 matches
Mail list logo
