Re: Differentiation problem/bug?

*To*: mathgroup at smc.vnet.net*Subject*: [mg69820] Re: [mg69734] Differentiation problem/bug?*From*: "Carl K. Woll" <carlw at wolfram.com>*Date*: Sat, 23 Sep 2006 23:45:24 -0400 (EDT)*References*: <200609220504.BAA16344@smc.vnet.net>

Arturas Acus wrote: > Dear group, > > the only way I can explain the rezults of simple differentiation command > D[] bellow is the dissapointing bug, which was absent in 5.0, but exist > in: > $Version > 5.2 for Linux (June 20, 2005) > > In[1]: inp = 1/4 + 3/(8*E^((2*I)*F)) + (3*E^((2*I)*F))/8 - > E^((-2*I)*F - I*\[Theta])/4 + E^((2*I)*F - I*\[Theta])/4 - > E^((-2*I)*F + I*\[Theta])/4 + E^((2*I)*F + I*\[Theta])/4 + > E^((-2*I)*F - (2*I)*\[Theta])/16 + E^((2*I)*F - (2*I)*\[Theta])/16+ > E^((-2*I)*F + (2*I)*\[Theta])/16 + E^((2*I)*F + (2*I)*\[Theta])/16- > 1/(8*E^((2*I)*\[Theta])) - E^((2*I)*\[Theta])/8 > > In[2]: D[inp, r, NonConstants -> {F}] > Out[2]: 0 > The bad news is that this unfortunately is a bug in version 5.2. The good news is that it has been corrected in the development version. Carl Woll Wolfram Research > how it was found: > > In[3]: D[#, r, NonConstants -> {F}] & /@ Expand[inp] > Out[3]: (((-3*I)/4)*D[F, r, NonConstants -> {F}])/E^((2*I)*F) + > ((3*I)/4)*E^((2*I)*F)*D[F, r, NonConstants -> {F}] > > check: > > In[4]: D[Evaluate[inp /. {F -> F[r]}], r] // FullSimplify > Out[4]: ((-I/8)*((-1 + E^(I*\[Theta]))^4 - E^((4*I)*F[r])*(1 + E^(I* > \[Theta]))^4)*Derivative[1][F][r])/E^((2*I)*(\[Theta] + F[r])) > > > I believe I can trust the Out[4] rezult. Most dissapointing is that now > I cannot trust my previous calculations, because somwhere I changed from > version 5.0 to 5.2, and 5.0 gives correct rezult. Please check this > behaviour for other versions/OS and be carefull using NonConstants > option. > > Sincerely, >

**References**:**Differentiation problem/bug?***From:*Arturas Acus <acus@itpa.lt>