MathGroup Archive 2011

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Evaluation of limit

  • To: mathgroup at
  • Subject: [mg115742] Re: Evaluation of limit
  • From: Roland Franzius <roland.franzius at>
  • Date: Wed, 19 Jan 2011 05:30:41 -0500 (EST)
  • References: <ih3rc9$bfl$>

Am 18.01.2011 11:51, schrieb Dr. C. S. Jog:
> If I give the command
> In[1]:= Limit[Cosh[x]+Sinh[x]^2*Log[Tanh[x/2]],x->Infinity]
> I get
>                                   x          2
> Out[1]= Limit[Cosh[x] + Log[Tanh[-]] Sinh[x] , x ->  Infinity]
>                                   2
> instead of zero. This result can be verified numerically (of course, by
> using higher precision than the default).

The numerical evaluation is instable, oscillating for x >10 as seen by 
Plot, thanks to the general flaw of Mathematica to handle the algebra of 
Log and ArcTan functions properly.

Replace the hyperbolic functions by

Cosh[x]-> 1/2 (q+1/q), Sinh[x]->1/2(q-1/q), Tanh[x/2]-> (q-1)/(q+1)

and let q->oo.

Since one and a half decade I am waiting for easy to use standard 
commands like

LogExpand, LogTogether, LogAssumptions for Simplify, LogToTrig etc.


Roland Franzius

  • Prev by Date: Re: CUDADot bug?
  • Next by Date: Re: InterpolatingPolynomial error message
  • Previous by thread: Re: Evaluation of limit
  • Next by thread: Array of arrays of various sizes and compile