Thanks for the clear explanation. I suppose the zerob / zeroa mechanism works OK for
specifying the limit direction, but if the general simplifier did zeroa - zeroa --> 0
in the middle of a limit computation, the limit direction could be lost. Maybe the
limit code is careful to never do that.
--Barton
-----drdieterkaiser at web.de wrote: -----
>I?know?two?different?mechanism?to?get?the?direction?of?a?one-sided?limit
>in?a?simplimit%function:
>
>The?direction?is?passed?as?an?argument.?The?problem?is?transformed?by
>limit?to?a?problem?with?the?limit?value?$zeroa,?when?we?come?from?above
>and?$zerob,?when?we?come?from?below:
>
>(%i20)?limit(n!,n,-1,plus);
>??0:?(SIMPLIMFACT?((MFACTORIAL?SIMP)?((MPLUS?SIMP)?-1?$N))?$N?$ZEROA)
>??0:?SIMPLIMFACT?returned?$INF
>Dieter?Kaiser