Re: TraditionalForm bug?

*To*: mathgroup at smc.vnet.net*Subject*: [mg66069] Re: TraditionalForm bug?*From*: Andrzej Kozlowski <akoz at mimuw.edu.pl>*Date*: Fri, 28 Apr 2006 06:33:17 -0400 (EDT)*References*: <7B896DFE-0B74-4E2B-A5C0-D6FCBDE2B69F@mimuw.edu.pl>*Sender*: owner-wri-mathgroup at wolfram.com

On 27 Apr 2006, at 22:06, Andrzej Kozlowski wrote: > I just noticed something in my version of Mathematica: > > $Version > > 5.1 for Mac OS X (October 25, 2004) > > which certainly looks like a bug that should have been spotted a > long time ago. The "bug" is simply this. Evaluate > > In[14]:= > Sin' > > Out[14]= > cos(#1)& > > This is fine. But now simple convert the above input to > TraditionalForm and evaluate it again: > > In[15]:= > \!\(\*FormBox[ > SuperscriptBox["sin", "â?²", > MultilineFunction->None], TraditionalForm]\) > > Out[15]= > \!\(\*FormBox[ > SuperscriptBox["sin", "â?²", > MultilineFunction->None], TraditionalForm]\) > > What happens is that Mathematica converts Sin` to sin` and then > forgets what sin means. This does not happen if you do the same > with an argument: > > In[17]:= > Sin'[x] > > Out[17]= > cos(x) > > In[18]:= > \!\(\*FormBox[ > RowBox[{ > SuperscriptBox["sin", "â?²", > MultilineFunction->None], "(", "x", ")"}], TraditionalForm]\) > > Out[18]= > cos(x) > > The embarrassing thing is that I discovered this while trying to > show my students the advantages of functional notation, and in > particular the fact that one can refer to functions and their > derivatives without introducing unnecessary variables. Of course > this is true, but if you sue TraditionalForm form input! This > reminds me that there was recently a discussion of whether > TraditionalForm is or is not suitable for this purpose. I recall > that Paul Abbott and Daniel Lichtblau expressed opposing views and > I was tempted to enter the fray but decided to stay out, because > while my heart sided with Paul my brain did not. As is often the > case the brain seems to have been right. > > Andrzej Kozlowski Perhaps I was a bit too harsh, since this works: (sin(#)&)' cos(#1)& So things are not quite as bad as I wrote above, we can still use functional notation and derivatives as long as we remember to write sin(#)& instead of sin etc. Still, this means that TraditionalForm is not quite as "traditional" as one would wish it to be. Andrzej Kozlowski