MathGroup Archive 2012

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

Search the Archive

Re: Pattern Matching with ReplaceRepated Behaving Unexpectedly


The second rule is never seen since the first result matches the first
rule. Note:

{u, a, a, v} //. {
  {p___, x : Longest[(a_) ..], q___} :>
   (Print["Rule 1 applied"];
    {p, a^Length[{x}], q}),
  {x___, a^2, y___} :>
   (Print["Rule 2 applied"];
    {x, e, y})}

Rule 1 applied

Rule 1 applied

{u, a^2, v}

MatchQ[%, {p___, x : Longest[(a_) ..], q___}]

True

You can just reverse the order of the rules

{u, a, a, v} //. {
  {x___, a^2, y___} :>
   (Print["Rule 1 applied"];
    {x, e, y}),
  {p___, x : Longest[(a_) ..], q___} :>
   (Print["Rule 2 applied"];
    {p, a^Length[{x}], q})}

Rule 2 applied

Rule 1 applied

Rule 2 applied

{u, e, v}


Bob Hanlon


On Tue, Dec 4, 2012 at 4:13 AM, Brentt <brenttnewman at gmail.com> wrote:
>
> I'm having trouble understanding how replace repeated works.
>
> I have this expression:
>
> In[1]:= {u,a, a,v} //. {
>   {p___, x : Longest[(a_) ..], q___} :> {p, a^Length[{x}], q},
>   {x___, a^2, y___} :> {x, e, y}
>   }
>
>
> What I think should happen here is that {u,a,a,v} should become {u,a^2,v}
> and then, since ReplaceRepeated should make another pass, {u,a^2,v} should
> become {u,e,v}. And actually if I break up the replacement rules, and
> evaluate them one by one, this is precisely what happens. But when I try to
> put the replacement rules into a list of rules with replace repeated as
> above it just does the first rule and stops. I.e. it outputs
>
> Out[1]:= {u,a^2,v}
>
> Strangely if I do
>
> In[2]:= % //. {
>   {p___, x : Longest[(a_) ..], q___} :> {p, a^Length[{x}], q},
>   {x___, a^2, y___} :> {x, e, y}
>   }
>
> I get the result I expected ReplaceRepeated to give me before
>
> Out[2]:= {u,e,v}
>
>
> I actually thought Replace Repated exists for this very reason, to keep on
> applying the rules until the rules do not change the output any longer, but
> it is failing to do that in this case for some reason even though there
> still seems to be patterns that should be changed?
>
> Can anyone tell me what I am missing here?
>
> Thank you, any help will be greatly appreciated.
>
>



  • Prev by Date: Re: Mathematica 9: bug in merging cells
  • Next by Date: Re: Could someone help optimize this for Mathematica 9?
  • Previous by thread: Pattern Matching with ReplaceRepated Behaving Unexpectedly
  • Next by thread: Could someone help optimize this for Mathematica 9?