Hello community,
here is the log from the commit of package ghc-math-functions for
openSUSE:Factory checked in at 2020-08-28 21:34:46
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Comparing /work/SRC/openSUSE:Factory/ghc-math-functions (Old)
and /work/SRC/openSUSE:Factory/.ghc-math-functions.new.3399 (New)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "ghc-math-functions"
Fri Aug 28 21:34:46 2020 rev:7 rq:829342 version:0.3.4.1
Changes:
--------
--- /work/SRC/openSUSE:Factory/ghc-math-functions/ghc-math-functions.changes
2020-06-19 17:15:10.306484439 +0200
+++
/work/SRC/openSUSE:Factory/.ghc-math-functions.new.3399/ghc-math-functions.changes
2020-08-28 21:34:47.748707251 +0200
@@ -1,0 +2,13 @@
+Tue Aug 18 10:45:24 UTC 2020 - Peter Simons <psim...@suse.com>
+
+- Replace %setup -q with the more modern %autosetup macro.
+
+-------------------------------------------------------------------
+Sat Aug 15 02:02:15 UTC 2020 - psim...@suse.com
+
+- Update math-functions to version 0.3.4.1.
+ ## Changes in 0.3.4.1
+
+ * Precision of `incompleteGamma` improved.
+
+-------------------------------------------------------------------
Old:
----
math-functions-0.3.4.0.tar.gz
math-functions.cabal
New:
----
math-functions-0.3.4.1.tar.gz
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Other differences:
------------------
++++++ ghc-math-functions.spec ++++++
--- /var/tmp/diff_new_pack.lNVydz/_old 2020-08-28 21:34:48.456707586 +0200
+++ /var/tmp/diff_new_pack.lNVydz/_new 2020-08-28 21:34:48.460707588 +0200
@@ -19,13 +19,12 @@
%global pkg_name math-functions
%bcond_with tests
Name: ghc-%{pkg_name}
-Version: 0.3.4.0
+Version: 0.3.4.1
Release: 0
Summary: Collection of tools for numeric computations
License: BSD-2-Clause
URL: https://hackage.haskell.org/package/%{pkg_name}
Source0:
https://hackage.haskell.org/package/%{pkg_name}-%{version}/%{pkg_name}-%{version}.tar.gz
-Source1:
https://hackage.haskell.org/package/%{pkg_name}-%{version}/revision/1.cabal#/%{pkg_name}.cabal
BuildRequires: ghc-Cabal-devel
BuildRequires: ghc-data-default-class-devel
BuildRequires: ghc-deepseq-devel
@@ -58,8 +57,7 @@
This package provides the Haskell %{pkg_name} library development files.
%prep
-%setup -q -n %{pkg_name}-%{version}
-cp -p %{SOURCE1} %{pkg_name}.cabal
+%autosetup -n %{pkg_name}-%{version}
%build
%ghc_lib_build
++++++ math-functions-0.3.4.0.tar.gz -> math-functions-0.3.4.1.tar.gz ++++++
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore'
old/math-functions-0.3.4.0/Numeric/SpecFunctions/Internal.hs
new/math-functions-0.3.4.1/Numeric/SpecFunctions/Internal.hs
--- old/math-functions-0.3.4.0/Numeric/SpecFunctions/Internal.hs
2001-09-09 03:46:40.000000000 +0200
+++ new/math-functions-0.3.4.1/Numeric/SpecFunctions/Internal.hs
2001-09-09 03:46:40.000000000 +0200
@@ -409,12 +409,21 @@
|| (abs mu < 0.4)
-- Gautschi's algorithm.
--
- -- Evaluate series for P(a,x). See [Temme1994] Eq. 5.5
- --
- -- FIXME: Term `exp (log x * z - x - logGamma (z+1))` doesn't give full
precision
+ -- Evaluate series for P(a,x). See [Temme1994] Eq. 5.5 and [NOTE:
+ -- incompleteGamma.taylorP]
+ factorP
+ | a < 10 = x ** a
+ / (exp x * exp (logGamma (a + 1)))
+ | a < 1182.5 = (x * exp 1 / a) ** a
+ / exp x
+ / sqrt (2*pi*a)
+ / exp (logGammaCorrection a)
+ | otherwise = (x * exp 1 / a * exp (-x/a)) ** a
+ / sqrt (2*pi*a)
+ / exp (logGammaCorrection a)
taylorSeriesP
= sumPowerSeries x (scanSequence (/) 1 $ enumSequenceFrom (a+1))
- * exp (log x * a - x - logGamma (a+1))
+ * factorP
-- Series for 1-Q(a,x). See [Temme1994] Eq. 5.5
taylorSeriesComplQ
= sumPowerSeries (-x) (scanSequence (/) 1 (enumSequenceFrom 1) /
enumSequenceFrom a)
@@ -1322,3 +1331,53 @@
, 4.269068009004705e304
, 7.257415615307998e306
]
+
+
+-- [NOTE: incompleteGamma.taylorP]
+--
+-- Incompltete gamma uses several algorithms for different parts of
+-- parameter space. Most troublesome is P(a,x) Taylor series
+-- [Temme1994,Eq.5.5] which requires to evaluate rather nasty
+-- expression:
+--
+-- x^a x^a
+-- ------------- = -------------
+-- exp(x)·Γ(a+1) exp(x)·a·Γ(a)
+--
+-- Conditions:
+-- | 0.5<x<1.1 = x < 4/3*a
+-- | otherwise = x < a
+--
+-- For small `a` computation could be performed directly. However for
+-- largish values of `a` it's possible some of factor in the
+-- expression overflow. Values below take into account ranges for
+-- Taylor P approximation:
+--
+-- · a > 155 - x^a could overflow
+-- · a > 1182.5 - exp(x) could overflow
+--
+-- Usual way to avoid overflow problem is to perform calculations in
+-- the log domain. It however doesn't work very well in this case
+-- since we encounter catastrophic cancellations and could easily lose
+-- up to 6(!) digits for large `a`.
+--
+-- So we take another approach and use Stirling approximation with
+-- correction (logGammaCorrection).
+--
+-- x^a / x·e \^a 1
+-- ≈ ------------------------- = | --- | · ----------------
+-- exp(x)·sqrt(2πa)·(a/e)^a) \ a / exp(x)·sqrt(2πa)
+--
+-- We're using this approach as soon as logGammaCorrection starts
+-- working (a>10) because we don't have implementation for gamma
+-- function and exp(logGamma z) results in errors for large a.
+--
+-- Once we get into region when exp(x) could overflow we rewrite
+-- expression above once more:
+--
+-- / x·e \^a 1
+-- | --- · e^(-x/a) | · ---------
+-- \ a / sqrt(2πa)
+--
+-- This approach doesn't work very well but it's still big improvement
+-- over calculations in the log domain.
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore' old/math-functions-0.3.4.0/bench/bench.hs
new/math-functions-0.3.4.1/bench/bench.hs
--- old/math-functions-0.3.4.0/bench/bench.hs 1970-01-01 01:00:00.000000000
+0100
+++ new/math-functions-0.3.4.1/bench/bench.hs 2001-09-09 03:46:40.000000000
+0200
@@ -0,0 +1,125 @@
+{-# LANGUAGE NumDecimals #-}
+import Gauge.Main
+import Data.Default.Class
+import qualified Data.Vector.Unboxed as U
+import Text.Printf
+import System.Random (randomIO)
+
+import qualified Numeric.Sum as Sum
+import Numeric.SpecFunctions
+import Numeric.Polynomial
+import Numeric.RootFinding
+
+
+
+-- Uniformly sample logGamma performance between 10^-6 to 10^6
+benchmarkLogGamma logG =
+ [ bench (printf "%.3g" x) $ nf logG x
+ | x <- [ m * 10**n | n <- [ -8 .. 8 ]
+ , m <- [ 10**(i / tics) | i <- [0 .. tics-1] ]
+ ]
+ ]
+ where tics = 3
+{-# INLINE benchmarkLogGamma #-}
+
+
+-- Power of polynomial to be evaluated (In other words length of coefficients
vector)
+coef_size :: [Int]
+coef_size = [ 1,2,3,4,5,6,7,8,9
+ , 10, 30
+ , 100, 300
+ , 1000, 3000
+ , 10000, 30000
+ ]
+{-# INLINE coef_size #-}
+
+-- Precalculated coefficients
+coef_list :: [U.Vector Double]
+coef_list = [ U.replicate n 1.2 | n <- coef_size]
+{-# NOINLINE coef_list #-}
+
+
+
+main :: IO ()
+main = do
+ v <- U.replicateM 1e6 randomIO :: IO (U.Vector Double)
+ defaultMain
+ [ bgroup "logGamma" $
+ benchmarkLogGamma logGamma
+ , bgroup "incompleteGamma" $
+ [ bench (show p) $ nf (incompleteGamma p) p
+ | p <- [ 0.1
+ , 1, 3
+ , 10, 30
+ , 100, 300
+ , 999, 1000
+ ]
+ ]
+ , bgroup "factorial"
+ [ bench (show n) $ nf factorial n
+ | n <- [ 0, 1, 3, 6, 9, 11, 15
+ , 20, 30, 40, 50, 60, 70, 80, 90, 100
+ ]
+ ]
+ , bgroup "incompleteBeta"
+ [ bench (show (p,q,x)) $ nf (incompleteBeta p q) x
+ | (p,q,x) <- [ (10, 10, 0.5)
+ , (101, 101, 0.5)
+ , (1010, 1010, 0.5)
+ , (10100, 10100, 0.5)
+ , (100100, 100100, 0.5)
+ , (1001000, 1001000, 0.5)
+ , (10010000,10010000,0.5)
+ ]
+ ]
+ , bgroup "log1p"
+ [ bench (show x) $ nf log1p x
+ | x <- [ -0.9
+ , -0.5
+ , -0.1
+ , 0.1
+ , 0.5
+ , 1
+ , 10
+ , 100
+ ] :: [Double]
+ ]
+ , bgroup "sinc" $
+ bench "sin" (nf sin (0.55 :: Double))
+ : [ bench (show x) $ nf sinc x
+ | x <- [0, 1e-6, 1e-3, 0.5]
+ ]
+ , bgroup "erf & Co"
+ [ bgroup "erf"
+ [ bench (show x) $ nf erf x
+ | x <- [0, 1.1, 100, 1000]
+ ]
+ , bgroup "erfc"
+ [ bench (show x) $ nf erfc x
+ | x <- [0, 1.1, 100, 1000]
+ ]
+ , bgroup "invErfc"
+ [ bench (show x) $ nf erfc x
+ | x <- [1e-9, 1e-6, 1e-3, 0.1, 1]
+ ]
+ ]
+ , bgroup "expm1"
+ [ bench (show x) $ nf expm1 (x :: Double)
+ | x <- [-0.1, 0, 1, 19]
+ ]
+ , bgroup "poly"
+ $ [ bench ("vector_"++show (U.length coefs)) $ nf (\x ->
evaluatePolynomial x coefs) (1 :: Double)
+ | coefs <- coef_list ]
+ ++ [ bench ("unpacked_"++show n) $ nf (\x -> evaluatePolynomialL x
(map fromIntegral [1..n])) (1 :: Double)
+ | n <- coef_size ]
+ , bgroup "RootFinding"
+ [ bench "ridders sin" $ nf (ridders def (0,pi/2)) (\x -> sin x
- 0.525)
+ , bench "newton sin" $ nf (newtonRaphson def (0,1.2,pi/2)) (\x -> (sin
x - 0.525,cos x))
+ ]
+ , bgroup "Sum"
+ [ bench "naive" $ whnf U.sum v
+ , bench "kahan" $ whnf (Sum.sumVector Sum.kahan) v
+ , bench "kbn" $ whnf (Sum.sumVector Sum.kbn) v
+ , bench "kb2" $ whnf (Sum.sumVector Sum.kb2) v
+ ]
+ ]
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore' old/math-functions-0.3.4.0/changelog.md
new/math-functions-0.3.4.1/changelog.md
--- old/math-functions-0.3.4.0/changelog.md 2001-09-09 03:46:40.000000000
+0200
+++ new/math-functions-0.3.4.1/changelog.md 2001-09-09 03:46:40.000000000
+0200
@@ -1,8 +1,14 @@
+## Changes in 0.3.4.1
+
+ * Precision of `incompleteGamma` improved.
+
+
## Changes in 0.3.4.0
* Dependency on `vector-th-unbox` is dropped. All instances are written by
hand now.
+
## Changes in 0.3.3.0
* New implementation for `logGamma`. Now it's precise within 2 ulps at full
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore' old/math-functions-0.3.4.0/math-functions.cabal
new/math-functions-0.3.4.1/math-functions.cabal
--- old/math-functions-0.3.4.0/math-functions.cabal 2001-09-09
03:46:40.000000000 +0200
+++ new/math-functions-0.3.4.1/math-functions.cabal 2001-09-09
03:46:40.000000000 +0200
@@ -1,5 +1,5 @@
name: math-functions
-version: 0.3.4.0
+version: 0.3.4.1
cabal-version: >= 1.10
license: BSD2
license-file: LICENSE
@@ -74,7 +74,7 @@
build-depends: base >= 4.5 && < 5
, deepseq
, data-default-class >= 0.1.2.0
- , vector >= 0.7
+ , vector >= 0.11
, primitive
if flag(system-expm1) && !os(windows)
cpp-options: -DUSE_SYSTEM_EXPM1
@@ -94,7 +94,7 @@
other-modules:
Numeric.SpecFunctions.Compat
-test-suite tests
+test-suite math-function-tests
default-language: Haskell2010
other-extensions: ViewPatterns
@@ -128,6 +128,31 @@
, tasty-hunit >= 0.10
, tasty-quickcheck >= 0.10
+benchmark math-functions-bench
+ type: exitcode-stdio-1.0
+ if impl(ghc <= 7.10 ) || impl(ghcjs)
+ buildable: False
+ default-language: Haskell2010
+ other-extensions:
+ BangPatterns
+ CPP
+ DeriveDataTypeable
+ FlexibleContexts
+ MultiParamTypeClasses
+ ScopedTypeVariables
+ TemplateHaskell
+ TypeFamilies
+ DeriveGeneric
+ ghc-options: -Wall -O2
+ hs-source-dirs: bench
+ Main-is: bench.hs
+ build-depends: base >= 4.5 && < 5
+ , math-functions
+ , data-default-class
+ , vector
+ , random
+ , gauge >=0.2.5
+
source-repository head
type: git
location: https://github.com/bos/math-functions
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore' old/math-functions-0.3.4.0/tests/Tests/SpecFunctions.hs
new/math-functions-0.3.4.1/tests/Tests/SpecFunctions.hs
--- old/math-functions-0.3.4.0/tests/Tests/SpecFunctions.hs 2001-09-09
03:46:40.000000000 +0200
+++ new/math-functions-0.3.4.1/tests/Tests/SpecFunctions.hs 2001-09-09
03:46:40.000000000 +0200
@@ -8,6 +8,7 @@
import Control.Monad
import Data.List
+import Data.Maybe
import qualified Data.Vector as V
import Data.Vector ((!))
import qualified Data.Vector.Unboxed as U
@@ -35,6 +36,13 @@
erfcLargeTol = 64
#endif
+isGHCJS :: Bool
+#if defined(__GHCJS__)
+isGHCJS = True
+#else
+isGHCJS = False
+#endif
+
tests :: TestTree
tests = testGroup "Special functions"
[ testGroup "erf"
@@ -42,14 +50,14 @@
-- large arguments
testCase "erfc table" $
forTable "tests/tables/erfc.dat" $ \[x, exact] ->
- checkTabular erfcTol (show x) exact (erfc x)
+ checkTabularPure erfcTol (show x) exact (erfc x)
, testCase "erfc table [large]" $
forTable "tests/tables/erfc-large.dat" $ \[x, exact] ->
- checkTabular erfcLargeTol (show x) exact (erfc x)
+ checkTabularPure erfcLargeTol (show x) exact (erfc x)
--
, testCase "erf table" $
forTable "tests/tables/erf.dat" $ \[x, exact] -> do
- checkTabular erfTol (show x) exact (erf x)
+ checkTabularPure erfTol (show x) exact (erf x)
, testProperty "id = erfc . invErfc" invErfcIsInverse
, testProperty "id = invErfc . erfc" invErfcIsInverse2
, testProperty "invErf = erf^-1" invErfIsInverse
@@ -58,16 +66,16 @@
, testGroup "log1p & Co"
[ testCase "expm1 table" $
forTable "tests/tables/expm1.dat" $ \[x, exact] ->
- checkTabular 2 (show x) exact (expm1 x)
+ checkTabularPure 2 (show x) exact (expm1 x)
, testCase "log1p table" $
forTable "tests/tables/log1p.dat" $ \[x, exact] ->
- checkTabular 1 (show x) exact (log1p x)
+ checkTabularPure 1 (show x) exact (log1p x)
]
----------------
, testGroup "gamma function"
[ testCase "logGamma table [fractional points" $
forTable "tests/tables/loggamma.dat" $ \[x, exact] -> do
- checkTabular 2 (show x) exact (logGamma x)
+ checkTabularPure 2 (show x) exact (logGamma x)
, testProperty "Gamma(x+1) = x*Gamma(x)" $ gammaReccurence
, testCase "logGamma is expected to be precise at 1e-15 level" $
forM_ [3..10000::Int] $ \n -> do
@@ -79,7 +87,11 @@
, testGroup "incomplete gamma"
[ testCase "incompleteGamma table" $
forTable "tests/tables/igamma.dat" $ \[a,x,exact] -> do
- checkTabular 16 (show (a,x)) exact (incompleteGamma a x)
+ let err | a < 10 = 16
+ | a <= 101 = if isGHCJS then 64 else 32
+ | a == 201 = 200
+ | otherwise = 32
+ checkTabularPure err (show (a,x)) exact (incompleteGamma a x)
, testProperty "incomplete gamma - increases" $
\(abs -> s) (abs -> x) (abs -> y) -> s > 0 ==> monotonicallyIncreases
(incompleteGamma s) x y
, testProperty "0 <= gamma <= 1" incompleteGammaInRange
@@ -89,7 +101,7 @@
----------------
, testGroup "beta function"
[ testCase "logBeta table" $
- forTable "tests/tables/logbeta.dat" $ \[p,q,exact] -> do
+ forTable "tests/tables/logbeta.dat" $ \[p,q,exact] ->
let errEst
-- For Stirling approx. errors are very good
| b > 10 = 2
@@ -109,9 +121,7 @@
est = ceiling
$ abs (logGamma a) + abs (logGamma b) + abs (logGamma (a
+ b))
/ abs (logBeta a b)
-
-
- checkTabular errEst (show (p,q)) exact (logBeta p q)
+ in checkTabularPure errEst (show (p,q)) exact (logBeta p q)
, testCase "logBeta factorial" betaFactorial
, testProperty "beta(1,p) = 1/p" beta1p
-- , testProperty "beta recurrence" betaRecurrence
@@ -137,7 +147,7 @@
-- Relative precision is lost when digamma(x) ≈ 0
, testCase "digamma is expected to be precise at 1e-12" $
forTable "tests/tables/digamma.dat" $ \[x, exact] ->
- checkTabular 2048
+ checkTabularPure 2048
(show x) (digamma x) exact
]
----------------
@@ -159,7 +169,7 @@
$ U.length factorialTable == 171
, testCase "Log factorial table" $
forTable "tests/tables/factorial.dat" $ \[i,exact] ->
- checkTabular 3
+ checkTabularPure 3
(show i) (logFactorial (round i :: Int)) exact
]
----------------
@@ -435,16 +445,26 @@
= fmap (fmap (fmap read . words) . lines)
. readFile
-forTable :: FilePath -> ([Double] -> IO ()) -> IO ()
+forTable :: FilePath -> ([Double] -> Maybe String) -> IO ()
forTable path fun = do
- mapM_ fun =<< readTable path
+ rows <- readTable path
+ case mapMaybe fun rows of
+ [] -> return ()
+ errs -> assertFailure $ intercalate "---\n" errs
checkTabular :: Int -> String -> Double -> Double -> IO ()
checkTabular prec x exact val =
- assertBool (unlines [ " x = " ++ x
- , " expected = " ++ show exact
- , " got = " ++ show val
- , " ulps diff = " ++ show (ulpDistance exact val)
- , " err.est. = " ++ show prec
- ])
- (within prec exact val)
+ case checkTabularPure prec x exact val of
+ Nothing -> return ()
+ Just s -> assertFailure s
+
+checkTabularPure :: Int -> String -> Double -> Double -> Maybe String
+checkTabularPure prec x exact val
+ | within prec exact val = Nothing
+ | otherwise = Just $ unlines
+ [ " x = " ++ x
+ , " expected = " ++ show exact
+ , " got = " ++ show val
+ , " ulps diff = " ++ show (ulpDistance exact val)
+ , " err.est. = " ++ show prec
+ ]
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore' old/math-functions-0.3.4.0/tests/tables/generate.py
new/math-functions-0.3.4.1/tests/tables/generate.py
--- old/math-functions-0.3.4.0/tests/tables/generate.py 2001-09-09
03:46:40.000000000 +0200
+++ new/math-functions-0.3.4.1/tests/tables/generate.py 2001-09-09
03:46:40.000000000 +0200
@@ -1,4 +1,4 @@
-#!/usr/bin/python
+#!/usr/bin/env python3
"""
"""
import itertools
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore' old/math-functions-0.3.4.0/tests/tables/igamma.dat
new/math-functions-0.3.4.1/tests/tables/igamma.dat
--- old/math-functions-0.3.4.0/tests/tables/igamma.dat 2001-09-09
03:46:40.000000000 +0200
+++ new/math-functions-0.3.4.1/tests/tables/igamma.dat 2001-09-09
03:46:40.000000000 +0200
@@ -7,6 +7,7 @@
0.000100000000000000005 10 0.999999999584179852012789127241
0.000100000000000000005 100 1.0
0.000100000000000000005 1000 1.0
+0.000100000000000000005 3301 1.0
0.00100000000000000002 9.99999999999999955e-07 0.986848133694076665339510700432
0.00100000000000000002 1.00000000000000008e-05 0.989123044695782668850940465364
0.00100000000000000002 0.00100000000000000002 0.993687646708860290096219637037
@@ -16,6 +17,7 @@
0.00100000000000000002 10 0.999999995830692182809738396874
0.00100000000000000002 100 1.0
0.00100000000000000002 1000 1.0
+0.00100000000000000002 3301 1.0
0.0100000000000000002 9.99999999999999955e-07 0.875933759832353305070970748339
0.0100000000000000002 1.00000000000000008e-05 0.896336798267197200971193210546
0.0100000000000000002 0.00100000000000000002 0.938570652526128985382985766694
@@ -25,6 +27,7 @@
0.0100000000000000002 10 0.99999995718295022590061122615
0.0100000000000000002 100 1.0
0.0100000000000000002 1000 1.0
+0.0100000000000000002 3301 1.0
0.100000000000000006 9.99999999999999955e-07 0.264033654327922324857187514752
0.100000000000000006 1.00000000000000008e-05 0.332398405040503295401903974218
0.100000000000000006 0.00100000000000000002 0.526768568392445111817869842601
@@ -34,6 +37,7 @@
0.100000000000000006 10 0.999999445201428209809392042838
0.100000000000000006 100 1.0
0.100000000000000006 1000 1.0
+0.100000000000000006 3301 1.0
0.200000000000000011 9.99999999999999955e-07
0.0687190937987684780583487585964
0.200000000000000011 1.00000000000000008e-05 0.108912260585591831357806184556
0.200000000000000011 0.00100000000000000002 0.273530102033034019134013607911
@@ -43,6 +47,7 @@
0.200000000000000011 10 0.999998540143010797930830697127
0.200000000000000011 100 1.0
0.200000000000000011 1000 1.0
+0.200000000000000011 3301 1.0
0.299999999999999989 9.99999999999999955e-07
0.0176595495901936665671778397068
0.299999999999999989 1.00000000000000008e-05
0.0352353606155625769158079986437
0.299999999999999989 0.00100000000000000002 0.140242458924867370902963425856
@@ -52,6 +57,7 @@
0.299999999999999989 10 0.99999715515533278951644372428
0.299999999999999989 100 1.0
0.299999999999999989 1000 1.0
+0.299999999999999989 3301 1.0
0.400000000000000022 9.99999999999999955e-07
0.00448690737698480048018769908146
0.400000000000000022 1.00000000000000008e-05
0.0112705727782256817005130809278
0.400000000000000022 0.00100000000000000002
0.0710923978953330760218610475958
@@ -61,6 +67,7 @@
0.400000000000000022 10 0.999995127544578487259167603836
0.400000000000000022 100 1.0
0.400000000000000022 1000 1.0
+0.400000000000000022 3301 1.0
0.5 9.99999999999999955e-07 0.00112837879096923635441785924383
0.5 1.00000000000000008e-05 0.00356823633818045042058077366094
0.5 0.00100000000000000002 0.0356705917296798854171108747554
@@ -70,6 +77,7 @@
0.5 10 0.999992255783568955916362323619
0.5 100 1.0
0.5 1000 1.0
+0.5 3301 1.0
0.599999999999999978 9.99999999999999955e-07
0.000281123932739927969642852394278
0.599999999999999978 1.00000000000000008e-05
0.00111917075717695837092315556803
0.599999999999999978 0.00100000000000000002
0.0177310780570833845378940622156
@@ -79,3 +87,94 @@
0.599999999999999978 10 0.999988293084421631090774556099
0.599999999999999978 100 1.0
0.599999999999999978 1000 1.0
+0.599999999999999978 3301 1.0
+2 9.99999999999999955e-07 4.9999966666679162138149041803e-13
+2 1.00000000000000008e-05 4.99996666679166715135638665241e-11
+2 0.00100000000000000002 0.000000499666791633340297383350611252
+2 0.0100000000000000002 0.000049667913340265892415918274953
+2 0.100000000000000006 0.00467884016044447002161170213187
+2 1 0.264241117657115356808952459677
+2 10 0.999500600772612666633108493329
+2 100 1.0
+2 1000 1.0
+2 3301 1.0
+3 9.99999999999999955e-07 1.6666654166671664402685631963e-19
+3 1.00000000000000008e-05 1.66665416671666693678925483422e-16
+3 0.00100000000000000002 1.66541716652780763845435502936e-10
+3 0.0100000000000000002 0.000000165421652807487686572525438025
+3 0.100000000000000006 0.000154653070264671678619072687835
+3 1 0.0803013970713941960111905745964
+3 10 0.997230604284488424056328917551
+3 100 1.0
+3 1000 1.0
+3 3301 1.0
+6 9.99999999999999955e-07 1.38888769841321886877873406008e-39
+6 1.00000000000000008e-05 1.38887698417906798768211894792e-33
+6 0.00100000000000000002 1.38769893337745993330072800382e-21
+6 0.0100000000000000002 1.37703605634306470721227911913e-15
+6 0.100000000000000006 0.00000000127489869222979188498133341591
+6 1 0.000594184817581692998827091061365
+6 10 0.932914037120968217714240937173
+6 100 1.0
+6 1000 1.0
+6 3301 1.0
+12 9.99999999999999955e-07 2.08767377170244306355144979982e-81
+12 1.00000000000000008e-05 2.08765642802367929860815380041e-69
+12 0.00100000000000000002 2.08574950796625691581696824468e-45
+12 0.0100000000000000002 2.06849404029269947208812650771e-33
+12 0.100000000000000006 1.90364240064062767847395955076e-21
+12 1 8.31610742688233390952391737168e-10
+12 10 0.30322385369689331180582927404
+12 100 0.99999999999999999999999999999
+12 1000 1.0
+12 3301 1.0
+18 9.99999999999999955e-07 1.56191921714497984569652172566e-124
+18 1.00000000000000008e-05 1.56190589978546723049068336087e-106
+18 0.00100000000000000002 1.56044168515546614690893456571e-70
+18 0.0100000000000000002 1.5471936172459858931960924774e-52
+18 0.100000000000000006 1.42075999849733529123506295102e-34
+18 1 6.06428067721557331946176715382e-17
+18 10 0.0142776135970496129089830965836
+18 100 0.999999999999999999999998742916
+18 1000 1.0
+18 3301 1.0
+101 9.99999999999999955e-07 1.06090022487198225461869933883e-766
+101 1.00000000000000008e-05 1.06089077042094832542309156458e-665
+101 0.00100000000000000002 1.05985129506822713235416044578e-463
+101 0.0100000000000000002 1.05044811631749169413217529395e-362
+101 0.100000000000000006 9.60885206142996276573449906708e-262
+101 1 3.94147589063752014615062291541e-161
+101 10 5.33940546071971052337450543217e-64
+101 100 0.473437801470001529623393607105
+101 1000 1.0
+101 3301 1.0
+201 9.99999999999999955e-07 6.30833677503352627325382205745e-1584
+201 1.00000000000000008e-05 6.30828028132019299365575041933e-1383
+201 0.00100000000000000002 6.30206906057329746426353105998e-981
+201 0.0100000000000000002 6.24588319207081699412164871293e-780
+201 0.100000000000000006 5.71085198693835010992991011704e-579
+201 1 2.33225525188187904433135775933e-378
+201 10 3.01310889066541622340100430602e-181
+201 100 4.62617947019577288096442044516e-19
+201 1000 1.0
+201 3301 1.0
+1000 9.99999999999999955e-07 2.4851656605824546988202041697e-8568
+1000 1.00000000000000008e-05 2.48514331653642003877116895294e-7568
+1000 0.00100000000000000002 2.48268669750003876617472467308e-5568
+1000 0.0100000000000000002 2.46046488714861784228553201292e-4568
+1000 0.100000000000000006 2.24889779123046964018322233315e-3568
+1000 1 9.15156509116491575720545785542e-2569
+1000 10 1.13964958763725134069338684487e-1572
+1000 100 1.0270971815582277877665615585e-611
+1000 1000 0.504205244180215508503777843602
+1000 3301 1.0
+3003 9.99999999999999955e-07 8.90812855529069811196158732723e-27160
+3003 1.00000000000000008e-05 8.90804840918643793907789460332e-24157
+3003 0.00100000000000000002 8.89923673804629728415411068459e-18151
+3003 0.0100000000000000002 8.81952937102869844156580901528e-15148
+3003 0.100000000000000006 8.0606844309353722733298755906e-12145
+3003 1 3.27821191297260784144201433956e-9142
+3003 10 4.05779611158376897690039719814e-6143
+3003 100 3.42800829838332693574242322961e-3179
+3003 1000 6.77752709164469153719156665409e-567
+3003 3301 0.999999933219976432069280011406
diff -urN '--exclude=CVS' '--exclude=.cvsignore' '--exclude=.svn'
'--exclude=.svnignore'
old/math-functions-0.3.4.0/tests/tables/inputs/igamma.dat
new/math-functions-0.3.4.1/tests/tables/inputs/igamma.dat
--- old/math-functions-0.3.4.0/tests/tables/inputs/igamma.dat 2001-09-09
03:46:40.000000000 +0200
+++ new/math-functions-0.3.4.1/tests/tables/inputs/igamma.dat 2001-09-09
03:46:40.000000000 +0200
@@ -8,6 +8,15 @@
0.4
0.5
0.6
+2
+3
+6
+12
+18
+101
+201
+1000
+3003
x =
1e-6
@@ -18,4 +27,5 @@
1
10
100
-1000
\ No newline at end of file
+1000
+3301
