MathGroup Archive 2007

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

Search the Archive

Re: ReplaceList and //.

  • To: mathgroup at smc.vnet.net
  • Subject: [mg73616] Re: ReplaceList and //.
  • From: Jean-Marc Gulliet <jeanmarc.gulliet at gmail.com>
  • Date: Fri, 23 Feb 2007 04:32:45 -0500 (EST)
  • Organization: The Open University, Milton Keynes, UK
  • References: <erjo1u$nn5$1@smc.vnet.net>

Mr Ajit Sen wrote:
> Dear Mathgroup,
> 
>  Could someone please enlighten me as to when I should
> use  ReplaceList and when to use //.  in pattern
> matching.
> 
>   Thus, if  Y = a x^2/b^3,  then
> 
>    ReplaceList[Y, a^m_. x^n_. b^p_. :> z 
>     b^(p + 1) a^(m - 1) x^(n - 1)][[1]]
> 
>  as well as 
>    
>   Y //. a^m_. x^n_. b^p_. :> z b^(p + 1) a^(m - 1)
> x^(n - 1)
>   
> give the same result.
> 
>  However,
> 
>      ReplaceList[a*b*c, {(x_)*(y_) -> x^y, (x_)*(y_)
> -> x + y}]
> 
>  and 
>       a*b*c //. {(x_)*(y_) :> x^y, (x_y)*_ :> x + y}
> 
>   yield different outputs.
> 
>  What am I missing here?

//. applies the transformation rule(s) once, as /. does. When done, //. 
applies the transformation rule(s) on the resulting expression. If any 
modifications have been made, //. applies again the rules the new 
expression; otherwise it stops.

Also, keep in mind that /. and //. attempt to match the most general 
patterns. For instance, a rule such as x_*y_, when applied to  a*b*c, 
will match only a*(b*c) (that is x -> a and y -> b*c) although there are 
some other interpretations.

The built-in function Trace should help you to understand what is going on:

In[1]:=
Trace[a*b*c /. {(x_)*(y_) -> x^y, (x_)*(y_) -> x + y}]

Out[1]=
                          y
{a b c /. {(x_) (y_) -> x , (x_) (y_) -> x + y},

          b c    b c
   Times[a   ], a   }

In[2]:=
Trace[a*b*c //. {(x_)*(y_) -> x^y, (x_y)*_ -> x + y}]

Out[2]=
{{{{(x_y) _, _ (x_y)}, _ (x_y) -> x + y,

     _ (x_y) -> x + y},

                   y
    {(x_) (y_) -> x , _ (x_y) -> x + y}},

                            y
   a b c //. {(x_) (y_) -> x , _ (x_y) -> x + y},

                          c
           b c    b c    b
   {Times[a   ], a   }, a  }

Regards,
Jean-Marc


  • Prev by Date: Re: split
  • Next by Date: RE: Re: Showing that ArcSinh[2]/ArcCsch[2] is 3?
  • Previous by thread: Re: ReplaceList and //.
  • Next by thread: NonLinearRegression Weights