MathSource Package Corrected

*To*: mathgroup at smc.vnet.net*Subject*: [mg26664] MathSource Package Corrected*From*: "Ersek, Ted R" <ErsekTR at navair.navy.mil>*Date*: Wed, 17 Jan 2001 00:47:18 -0500 (EST)*Sender*: owner-wri-mathgroup at wolfram.com

> In october 2000 I posted a package on MathSourec that allows one to easily > enable and disable the behavior provided by the RealOnly package. I called > the package SwitchableRealOnly and posted it at > http://www.mathsource.com/cgi-bin/msitem?0211-396 > > Besides allowing one to enable or disable the RealOnly behavior I tried to > correct a problem explained in the comments of the original RealOnly > package. Once the original RealOnly package is loaded Roots and Solve > will indicate that ( x^3+6x^2-9*0 ) > has non-real roots but the three roots are actually real. After loading my > earlier version of SwitchableRealOnly on my PC, Solve and Roots don't > indicate that the roots in the above case are non-real. However, I > recently learend that after installing the package on a Mac, Solve and > Roots would indicate the roots are non-real. It turns out the problem is > due to the way I tried to decide if a root is probably real. In my first > version of the package I used (N[Im[#]]==0 &) to determine if a root is > probably real. However that uses machine precesion arithmatic and the > results are different on various platforms, and is certainly not full > proof on any computer. I posted an improved version which avoids machine > precision arithmatic by using (Im[#]==0 &) to determine if a number is > probably real. > > The improved package can be down loaded from http://www.mathsource.com/cgi-bin/msitem?0211-396 -------------------- Regards, Ted Ersek Down load Mathematica tips, tricks from http://www.verbeia.com/mathematica/tips/Tricks.html