Re: The PM5 method?

[David Gallagher <dgallagher \\at// cachesoftware.com>]

 Dr James Stewart (author of MOPAC) gave a presentation on PM5 at the
 Chicago ACS on August 28, 2001. 

 He stated that PM5, now includes ALL main group elements in one method,
 adding Li, Be, B, Na, Mg, K, Ca, etc. 

 PM5 is also claimed to be significantly more accurate than PM3 or AM1 for
 heats of formation. However, the actual improvement in accuracy varies by
 element. The average unweighted errors in absolute heats of formation
 (kcal/mol) for each method compared to experiment is approximately as
 shown below. The averages are based on all elements previously published
 by Stewart for PM3 and AM1, and for all main group elements for PM5. Of
 course, relative errors are typically much smaller.

 AM1       16.4 kcals/mol
 PM3       20.4 kcals/mol
 PM5        5.6 kcals/mol (preliminary
 results)

 PM5 works with the COSMO solvation field and on systems of up to 20,000
 atoms when using the linear scaling option.

 Fujitsu is currently testing a beta version of PM5 at present. The PM5
 method will be available in CAChe at the end of this month (September),
 and also in a new release of MOPAC 2002 in the near future, when more
 information should become available.

 David Gallagher
 Fujitsu
 (www.cachesoftware.com/cache5/)

 

 At 06:36 AM 9/11/2001 -0700, Mark A. Thompson wrote:
>  
>  Has the new "PM5" method been
>  described in the literature anywhere?
>   
>  Mark
>   
>   
>   
>  http://www.planaria-software.com
>