From chemistry-request(+ at +)ccl.net Tue Jan 28 20:47:35 1992
Date: Tue, 28 Jan 1992 19:25 CST
From: Andy Holder <AHOLDER#* at *#VAX1.UMKC.EDU>
Subject: SADDLE in MOPAC/AMPAC
To: CHEMISTRY ^at^ ccl.net
Status: R

Hi Guys and Guyettes!
  A while back someone posted a messgae on using the SADDLE method to
  locate a transition state with one of the Dewar semiempirical pro-
  grams (MOPAC or AMPAC).  My experience is that this is not as reliable
  as some other techniques.  AMPAC has implemented the excellent CHAIN
  algorithm, installed by Daniel Liotard.  This requires input of a
  guess transition state, and left and right minima.  It has performed
  very well in locating reasonable TS's.  SADDLE has some problems
  that almost require one to specify everything in Cartesian coordinates
  for it to function properly (if that will do the trick).  I usually
  only resort to these automated procedures after the "shotgun" method
  has failed.  When beginning, I make an approximate guess at the trans-
  ition state and use a gradient minimization routine (POWELL, NLLSQ, or
  LTRD).  POWELL is generally the fastest.  After a POWELL computation,
  one MUST MUST MUST calculate the force constant matrix. The output
  file contains an "estimate" of the number of imaginary frequencies
  that is generally wrong.  (In fact, I have eliminated this part
  of the printout to avoid confusion to other users in my version.)
  LTRD is a very stable approach as it computes a full second deriv-
  ative matrix on each step.  It's frequency calculations are reliable.
  To sum up:

   1.  estimate TS geometry (Hammond's postulate is useful here)
   2.  gradient minimize with POWELL, compute force constants
   3.  if POWELL fails, try LTRD gradient minimzation/force const.
   4.  if this fails use CHAIN to locate approximate TS and go back
         to step 2
   5.  if this fails, find an easier reaction!

  Notes:

    Multiple bond breaking/forming reactions will be hard.  Grid
     searches along tow or more coordinates will be needed to get
     a good starting point for refinement. 

    Even if you have a CRAY-ZMP59000, do the calculation with a 
    semiempirical method before you waste time searching the pot-
    ential surface with a big, hairy ab initio method.


  Bye now.  Hope the sermon helps out some folks.

    Andy Holder

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
                              DR. ANDREW HOLDER
             Assistant Professor of Computational/Organic Chemistry

Department of Chemistry              ||  BITNET Addr:   AHOLDER -A_T- UMKCVAX1
University of Missouri - Kansas City ||  Internet Addr: aholder ^at^ vax1.umkc.edu
Spencer Chemistry, Room 502          ||  Phone Number:  (816) 235-2293
Kansas City, Missouri 64110          ||  FAX Number:    (816) 235-1717
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=