Re: Add the Logarithms (error in integral)

• To: mathgroup at smc.vnet.net
• Subject: [mg25205] Re: Add the Logarithms (error in integral)
• From: Jens-Peer Kuska <kuska at informatik.uni-leipzig.de>
• Date: Fri, 15 Sep 2000 02:21:43 -0400 (EDT)
• Organization: Universitaet Leipzig
• References: <8p9j71\$1fl@smc.vnet.net> <KGVu5.30316\$Zh6.36755@ralph.vnet.net> <ozyv5.58213\$Zh6.61667@ralph.vnet.net>
• Sender: owner-wri-mathgroup at wolfram.com

```Jos R Bergervoet wrote:
>
> Jens-Peer Kuska wrote:
> >
> > h1 = h //. a_.*Log[b_] + c_.*Log[d_] :> Log[b^a*d^c];
> >
> > Chop[N[h1 /. {a -> 10, b -> 10}]]
> >
> > 0.000716695
>
> The trick seems interesting. But the answer is incorrect (for
> a=b=10 it must be ArcTan[50/Sqrt[101]], which is:

It is *not* a trick. The formulas can be found in Abramowitz/Stegun
4.1.7 -- 4.1.11

The condition for

Log[z1_]+Log[z2_] :> Log[z1*z2]  for -Pi < Arg[z1]+Arg[z2]<=Pi

and

n_.*Log[z_] :> Log[z^n]  for -Pi < Arg[n*z] <=Pi

it's clear that it is a bit complicated to fix these conditions
for a symbolic expression. The application is only save for real
arguments.

BTW: You ask for a *real* result and not for the correct real result ;-)

Regards
Jens

```

• Prev by Date: First derivative of interpolated spline
• Next by Date: Re: Mapping down two lists
• Previous by thread: Re: Add the Logarithms (error in integral)
• Next by thread: BSpline Curves