Re: Why does this pattern fail to match?

*To*: mathgroup at smc.vnet.net*Subject*: [mg114165] Re: Why does this pattern fail to match?*From*: Albert Retey <awnl at gmx-topmail.de>*Date*: Fri, 26 Nov 2010 05:26:03 -0500 (EST)*References*: <iclfej$lb7$1@smc.vnet.net>

Hi, > I'm tearing my hair out over this one. Could someone please explain > to me why all the following MatchQ expressions > > MatchQ[Times[ -1, Power[s, -1] ], > Times[___, Power[___], ___]] > > MatchQ[Times[ -1, Power[s, -1]], > Times[___, Power[___] ]] > > MatchQ[Times[ -1, Power[s, -1] ], > Times[___, _Power , ___]] > > return False? (In all cases, s is undefined). And yet, this succeeds: > > MatchQ[Times[ -1, Power[s, -1]], > Times[___, _Power ]] Usually it is the FullForm that is what you need to look at, but this is what you already do. Your examples are all showing another common problem when doing pattern matching: the patterns are evaluated before the pattern matching is done, so the pattern that is tried is not what you actually expect. you can see this from evaluating the pattern: Times[___, Power[___], ___] > Is there a systematic way to debug/troubleshoot such pattern matching > problems? Trace is useless here. actually I think by using Trace you should be able to spot that the second argument of MatchQ is evaluated, but it is something that is easy to overlook. This will prevent the evaluation of the pattern and work as you expect: MatchQ[Times[-1, Power[s, -1]], HoldPattern[Times[___, Power[___], ___]]] you might want to look also at Verbatim, which is another symbol which is very helpful in constructing nontrivial patterns. > Also, whatever the reason is for the failure of this pattern to > match, where is this reason documented? I've gone blind poring > over the documentation trying to find an answer, without success. hth, albert