Re: RealDigits bug?
- To: mathgroup at smc.vnet.net
- Subject: [mg45910] Re: RealDigits bug?
- From: drbob at bigfoot.com (Bobby R. Treat)
- Date: Thu, 29 Jan 2004 05:35:34 -0500 (EST)
- References: <bv00g1$je7$1@smc.vnet.net> <bv2g1t$68$1@smc.vnet.net> <bv82tq$i97$1@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
I'm using Version 5.0.1 under WinXP and OS X, and I just reproduced it
again on both machines. It's a rather glaring bug.
pisot =
N[((1/3)*((1/2)*(27 +
3*Sqrt[69]))^
(1/3) +
(2/(27 + 3*Sqrt[69]))^
(1/3))^27351, 6000];
Frequencies[First[RealDigits[
pisot]]]
{{2086, 0}, {422, 1},
{427, 2}, {434, 3},
{467, 4}, {444, 5},
{452, 6}, {447, 7},
{437, 8}, {381, 9}, {3, 10}}
Bobby
adam.smith at hillsdale.edu (Adam Smith) wrote in message news:<bv82tq$i97$1 at smc.vnet.net>...
> Bobby,
>
> What version of Mathematica are you using? I am running Version 4.1
> and it does not include 10 in the Frequencies list
>
> In[3]:=
> Frequencies[First[RealDigits[ pisot]]]
>
>
> Out[3]=
> {{2089,0},{422,1},{427,2},{434,3},{467,4},{444,5},{452,6},{447,7},{437,
> 8},{381,9}}
>
>
> In response to Enrique. I suspect that the failure is related to a
> misunderstanding of the N[] function. In particular look in the Help
> for N[] and study the $MaxMachineNumber and $MaxPrecision. Since your
> pisot is on the order of 10^3340, you may be running into truncation
> errors with N[]. On my machine $MaxMachineNumber is around 10^308.
> Since pisot is clearly much greater than this, I would expect
> truncation errors.
>
> Adam Smith
>
> drbob at bigfoot.com (Bobby R. Treat) wrote in message news:<bv2g1t$68$1 at smc.vnet.net>...
> > I don't know whether the equality should be true, but it's very clear
> > that 10 isn't a base-10 digit!
> >
> > Frequencies[First[RealDigits[
> > pisot]]]
> > {{2086, 0}, {422, 1},
> > {427, 2}, {434, 3},
> > {467, 4}, {444, 5},
> > {452, 6}, {447, 7},
> > {437, 8}, {381, 9}, {3, 10}}
> >
> > Bobby
> >
> > Enrique Zeleny <ezeleny at fismat1.fcfm.buap.mx> wrote in message news:<bv00g1$je7$1 at smc.vnet.net>...
> > > This is wrong:
> > >
> > > In[1]:= pisot = N[((1/3)*((1/2)*(27 + 3*Sqrt[69]))^(1/3) + (2/(27 +
> > > 3*Sqrt[69]))^(1/3))^27351, 6000];
> > >
> > > In[2]:= pisot == FromDigits[RealDigits[pisot]]
> > >
> > > Out[2]:= False
> > >
> > > If you look at the value returned by RealDigits[pisot] contains three
> > > 10's!!!
> > >
> > >
> > > Enrique Zeleny
> > > Universidad Autonoma de Puebla
> > > Mexico