Re: Mathematica bugs?
- To: mathgroup at smc.vnet.net
- Subject: [mg67321] Re: Mathematica bugs?
- From: Peter Pein <petsie at dordos.net>
- Date: Sun, 18 Jun 2006 05:13:13 -0400 (EDT)
- References: <e70f29$rf4$1@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Yaroslav Bulatov schrieb:
> When I run the following line
> x = Pi/4; For[i = 0, i < 56, i += 1, x = 2*Abs[x - 1/2]]; N[x]
> I get
> 3.
>
> But x should always stay between 0 and 1, why do I get 3?
>
> Also
> x = Pi/4; For[i = 0, i < 50, i += 1, x = 2*Abs[x - 1/2]]; N[Log[x]]
> gives me
> Indeterminate
>
> How can I get an indeterminate here?
>
Hi Yaroslav,
each multiplication which 2 is a binary left shift of the multiplicand.
For iter56 = Nest[2*Abs[#1 - 1/2] & , Pi/4, 56];
Simplify[iter56]
56593902016227523 - 18014398509481984*Pi
you'll need about 56 bit to represent each of the summands:
Ceiling[Log[2, Abs /@ List @@ %]]
{56, 56}
compare this to
Log[2, 10]*$MachinePrecision
53.
Therefore you get an unreliable result using machine arithmetic:
N[iter56]
3.
Using multiple prec. arithmetic, you'll get
N[iter56, 10]
0.7938724527
Using insufficient precision leads also to:
logiter50 = Log[Nest[2*Abs[#1 - 1/2] & , Pi/4, 50]];
Simplify[logiter50]
Log[-884279719003555 + 281474976710656*Pi]
N[logiter50[[1]]]
0.
and therefore
N[logiter50]
Indeterminate
but:
N[logiter50, 16]
-3.367644345831603
hth,
Peter