MathGroup Archive 1999

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Sort

  • To: mathgroup at smc.vnet.net
  • Subject: [mg19176] Re: [mg19152] Sort
  • From: "Wolf, Hartmut" <hwolf at debis.com>
  • Date: Tue, 10 Aug 1999 02:52:40 -0400
  • Organization: debis Systemhaus
  • References: <199908060358.XAA08207@smc.vnet.net.>
  • Sender: owner-wri-mathgroup at wolfram.com

Hermann Meier schrieb:
> 
> Sort is not reliable. Try the following (Mathematica Version 4):
> 
> In[1]:=
> Sort[{0, -Sqrt[2], 5}]
> 
> Out[1]=
> {0, 5, -Sqrt[2]}
> 
> In[2]:=
> Sort[{0., -Sqrt[2], 5}]
> 
> Out[2]=
> {0., 5, -Sqrt[2]}
> 
> One has to apply N to the list to get the correct result:
> 
> In[3]:=
> Sort[N[{0., -Sqrt[2], 5}]]
> 
> Out[3]=
> {-1.414213562, 0., 5.}
> 
> HM

Hermann, of course Sort is reliable, but without applying N it doesn't
know anything of the value of Sqrt[2]. But you can make Sort to know
that if you just do

In[1]:= exactnumbers = {0, -Sqrt[2], 5};

In[4]:= Last /@ Sort[Transpose[{N[exactnumbers], exactnumbers}]]
Out[4]= {-Sqrt[2], 0, 5}

Sometimes perhaps, you have to be cautious with the numerical precision
you are using.

kind regards, hw



  • References:
    • Sort
      • From: "Hermann Meier" <hmeier@webshuttle.ch>
  • Prev by Date: Re: Sort
  • Next by Date: Re: Sort
  • Previous by thread: Sort
  • Next by thread: Re: Sort