summaryrefslogtreecommitdiff
path: root/README
diff options
context:
space:
mode:
authorAlberto Ruiz <aruiz@um.es>2008-10-17 21:21:39 +0000
committerAlberto Ruiz <aruiz@um.es>2008-10-17 21:21:39 +0000
commit9d9b1274a522e1bf0c5dea210765a0368ebb74a5 (patch)
treef876023bb51d7419d855169b308ed0611e325b40 /README
parent212ba96a26b6590868d9724560cef8b2471697f5 (diff)
acknowledgement to Dylan Simon, updated readme and fixed mapCAux
Diffstat (limited to 'README')
-rw-r--r--README21
1 files changed, 10 insertions, 11 deletions
diff --git a/README b/README
index 8d7775b..3a260a4 100644
--- a/README
+++ b/README
@@ -10,15 +10,12 @@ REQUIREMENTS ----------------------------
102) BLAS/LAPACK (http://www.netlib.org/lapack). 102) BLAS/LAPACK (http://www.netlib.org/lapack).
11 An optimized implementation is recommended. I have tested: 11 An optimized implementation is recommended. I have tested:
12 12
13 - Intel's MKL (http://www.intel.com/cd/software/products).
14 There is a free noncommercial download of MKL for Linux.
15
16 - ATLAS (http://math-atlas.sourceforge.net). 13 - ATLAS (http://math-atlas.sourceforge.net).
17 In Ubuntu the required packages are "refblas3-dev", "lapack3-dev", 14 In Ubuntu the required packages are "refblas3-dev", "lapack3-dev",
18 and "atlas3-base-dev" (or a version tuned for your machine). 15 and "atlas3-base-dev" (or a version tuned for your machine).
19 Please note that ATLAS currently requires compilation -fviaC in 32bit 16
20 machines. Otherwise many functions fail, producing strange NaN's. 17 - Intel's MKL (http://www.intel.com/cd/software/products).
21 Even with -fvia-C we may get wrong behavior in some cases. 18 There is a free noncommercial download of MKL for Linux.
22 19
23For ghc-6.8.x you may also need: 20For ghc-6.8.x you may also need:
24 21
@@ -59,6 +56,7 @@ Using Intel's MKL:
59 $ runhaskell Setup.lhs build 56 $ runhaskell Setup.lhs build
60 $ runhaskell Setup.lhs install 57 $ runhaskell Setup.lhs install
61 58
59More information: http://www.hmatrix.googlepages.com/installation
62 60
63See below for installation on Windows. 61See below for installation on Windows.
64 62
@@ -70,7 +68,7 @@ Prelude> Numeric.LinearAlgebra.Tests.runTests 20
70 68
71Additional tests with big matrices (taking a few minutes): 69Additional tests with big matrices (taking a few minutes):
72 70
73$ runhaskell examples/experiments bigtests 71$ runhaskell examples/experiments/bigtests
74 72
75EXAMPLES ------------------------------------------------------ 73EXAMPLES ------------------------------------------------------
76 74
@@ -99,10 +97,8 @@ A number of illustrative programs are included in the examples folder.
99KNOWN PROBLEMS / BUGS ------------------------------- 97KNOWN PROBLEMS / BUGS -------------------------------
100 98
101- Compilation with -O -fasm on 32-bit machines produces strange 99- Compilation with -O -fasm on 32-bit machines produces strange
102 NaN's results on certain blas/lapack calls. In these machines 100 NaN's results on certain foreign calls. More info at
103 the library is automatically compiled -fvia-C, which apparently 101 http://www.hmatrix.googlepages.com/bugs
104 solves the problem.
105 On 64-bit, or using MKL, the default and faster -fasm seems to work well.
106 102
107- On 64-bit machines the example "minimize.hs", when run from ghci, 103- On 64-bit machines the example "minimize.hs", when run from ghci,
108 produces a segmentation fault. It happens in the call to 104 produces a segmentation fault. It happens in the call to
@@ -228,3 +224,6 @@ in the Haskell mailing lists for their help.
228- Don Stewart fixed the implementation of the internal data structures 224- Don Stewart fixed the implementation of the internal data structures
229 to achieve excellent, C-like performance in Haskell functions which 225 to achieve excellent, C-like performance in Haskell functions which
230 explicitly work with the elements of vectors and matrices. 226 explicitly work with the elements of vectors and matrices.
227
228- Dylan Alex Simon improved the numeric instances to allow optimized
229 implementations of signum and abs on Vectors.