Re: Generalization of Variables ?

*To*: mathgroup at smc.vnet.net*Subject*: [mg109274] Re: Generalization of Variables ?*From*: Rui <rui.rojo at gmail.com>*Date*: Tue, 20 Apr 2010 05:50:22 -0400 (EDT)*References*: <hqc14l$g2v$1@smc.vnet.net>

On Apr 17, 7:05 am, Jack L Goldberg 1 <jackg... at umich.edu> wrote: > Hi Folks, > > The built-in command "Variables" does not give the (naively expected) > answer when it is called with functions such as x + Sin[x]. This > behavior is mentioned in the description of "Variables". > I have constructed a generalization of "Variables" which works as > expected - I think? I would appreciated a critique of this code which > addresses these issues: > > a) Is there a situation in which it fails? > > b) If the code works, can it be improved. > > Question b) is asked because I am not fully acquainted with all of > Mathematica's commands. > > Here is the code. > > variables[f_]:= Flatten[ > Union[ > Cases[ > M= ap[N,Level[f,{-1},z_Symbol]]/.True->{} > = ] > > I have tried "variables" on functions coded in the obvious manner, say, > Sin[x]+Exp[x^2], Sin[x+y]-z^2, etc. and a few esoteric examples. > > Jack variables[expr_] := DeleteDuplicates[ Cases[ expr, _Symbol?(! NumericQ[#] &), -1]] Or maybe, if you have lots of more duplicates than Pi, E, and numeric symbols, variables[expr_] := Cases[DeleteDuplicates[ Cases[expr, _Symbol, -1]], _?(! NumericQ[#] &)]