Re: [Haskell-cafe] Error with Float

19 Jul 2005


      Here's what i got


writeln x = putStr (x++ "\n")


f:: Double -> IO Double
f x = do
        let t = 2*x
        if (t<1)
	then return t
	else return (t-1)


gen :: Double -> IO()
gen x = do c<-f x
	   writeln ("Value  is: " ++ show c)
	   if (c /= 0.0)
	   	then  gen c
		else  return ()


Main> gen 0.1
Value  is: 0.2
Value  is: 0.4
Value  is: 0.8
Value  is: 0.6
Value  is: 0.2
Value  is: 0.4
Value  is: 0.800000000000001
Value  is: 0.600000000000001
Value  is: 0.200000000000003
Value  is: 0.400000000000006
Value  is: 0.800000000000011
Value  is: 0.600000000000023
Value  is: 0.200000000000045
Value  is: 0.400000000000091
Value  is: 0.800000000000182
Value  is: 0.600000000000364
Value  is: 0.200000000000728
Value  is: 0.400000000001455
Value  is: 0.80000000000291
Value  is: 0.600000000005821
Value  is: 0.200000000011642
Value  is: 0.400000000023283
Value  is: 0.800000000046566
Value  is: 0.600000000093132
Value  is: 0.200000000186265
Value  is: 0.400000000372529
Value  is: 0.800000000745058
Value  is: 0.600000001490116
Value  is: 0.200000002980232
Value  is: 0.400000005960464
Value  is: 0.800000011920929
Value  is: 0.600000023841858
Value  is: 0.200000047683716
Value  is: 0.400000095367432
Value  is: 0.800000190734863
Value  is: 0.600000381469727
Value  is: 0.200000762939453
Value  is: 0.400001525878906
Value  is: 0.800003051757813
Value  is: 0.600006103515625
Value  is: 0.20001220703125
Value  is: 0.4000244140625
Value  is: 0.800048828125
Value  is: 0.60009765625
Value  is: 0.2001953125
Value  is: 0.400390625
Value  is: 0.80078125
Value  is: 0.6015625
Value  is: 0.203125
Value  is: 0.40625
Value  is: 0.8125
Value  is: 0.625
Value  is: 0.25
Value  is: 0.5
Value  is: 0.0
...
From: Cale Gibbard 
Reply-To: Cale Gibbard 
To: Dinh Tien Tuan Anh 
CC: haskell-cafe@haskell.org
Subject: Re: [Haskell-cafe] Error with Float
Date: Tue, 19 Jul 2005 11:00:34 -0400
Perhaps you mean:
f x
    | x < 1     = 0 : f (2*x)
    | otherwise = 1 : f (2*(x-1))
Note that in the second case, the 1 is subtracted before multiplication by 
2.
If you were referring to the problem that this eventually gives
constantly 0 for values like 0.6, try importing the Ratio module and
applying it to 6%10, which is the exact rational value rather than a
floating point representation.
- Cale
_________________________________________________________________
Winks & nudges are here - download MSN Messenger 7.0 today! 
http://messenger.msn.co.uk