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

Re: ReplaceAll bug

• To: mathgroup at yoda.physics.unc.edu
• Subject: Re: ReplaceAll bug
• From: Silvio Levy <levy at math.berkeley.edu>
• Date: Fri, 31 Jan 92 20:06:34 PST

> Sqrt[240] /. Sqrt[x_] -> 2 Sqrt[x/4]
> 4 Sqrt[15]
> 
> Sqrt[480] /. Sqrt[x_] -> 2 Sqrt[x/4]
> Sqrt[480]

This isn't a problem with ReplaceAll, but with the seemingly random
normalization adopted by Sqrt.  Sqrt[240] is automatically converted
to 4 Sqrt[15], then the substitution rule transforms that into
8 Sqrt[15/4], which is automatically normalized again to 4 Sqrt[15]
(try it).  In the bottom example, the substitution rule acts first,
giving 2 Sqrt[120], and this is normalized to Sqrt[480].  Does anybody
know if there is any rhyme or reason to these normal forms?

Silvio