Molcas Forum

Support and discussions for Molcas and OpenMolcas users and developers

You are not logged in.

Announcement

Welcome to the Molcas forum.

Please note: The forum's URL has changed. The new URL is: https://molcasforum.univie.ac.at. Please update your bookmarks!

You can choose an avatar and change the default style by going to "Profile" → "Personality" or "Display".

#1 2019-09-17 09:30:29

moabe
Member
Registered: 2018-06-03
Posts: 23

Thresholds in SlapAf

Dear all,

I am doing some relaxed potential energy scans for a series of molecules at the CASSCF level, which require so many constrained optimizations. For many of them, I really have to spend so much time to get the converged structure (see following example). Apparently, the difficulty is in the "Cartesian Displacements".
In the manual, it has been suggested not to waste computer time by using the default thresholds in slapaf, which are typically too tight. The default is the convergence criteria as in the Gaussian program, which do not consider the energy change. There is also Baker criteria that consider the energy change for the convergence. I wonder changing the default thresholds to Baker or other user-defined ones is a good idea and safe. I appreciate if someone has any experience in this regard to share with me.

Many thanks 



**********************************************************************************************************************
*                                    Energy Statistics for Geometry Optimization                                     *
**********************************************************************************************************************
                       Energy     Grad      Grad              Step                 Estimated   Geom       Hessian     
Iter      Energy       Change     Norm      Max    Element    Max     Element     Final Energy Update Update   Index  
  1   -964.61686139  0.00000000 0.016498 -0.005207 nrc002  -0.008273  nrc054     -964.61695646 RS-RFO  None      0    
  2   -964.61694748 -0.00008609 0.012947  0.005365 nrc002  -0.013613  nrc065     -964.61701994 RS-RFO  BFGS      0    
  3   -964.61693767  0.00000981 0.010435  0.003577 nrc021   0.014592  nrc054     -964.61695634 RS-RFO  BFGS      0    
  4   -964.61699848 -0.00006081 0.002692  0.000905 nrc021   0.008245  nrc053     -964.61700623 RS-RFO  BFGS      0    
  5   -964.61700679 -0.00000831 0.001150 -0.000580 nrc065   0.032768  nrc053     -964.61702927 RS-RFO  BFGS      0    
  6   -964.61704109 -0.00003430 0.003432 -0.001133 nrc021   0.014635  nrc053     -964.61704778 RS-RFO  BFGS      0    
  7   -964.61705076 -0.00000967 0.003138 -0.001120 nrc007   0.080116  nrc053     -964.61708926 RS-RFO  BFGS      0    
  8   -964.61709327 -0.00004250 0.003347  0.001262 nrc007  -0.007692  nrc053     -964.61709910 RS-RFO  BFGS      0    
  9   -964.61710023 -0.00000697 0.001395 -0.000667 nrc066  -0.007541  nrc066     -964.61710463 RS-RFO  BFGS      0    
 10   -964.61710670 -0.00000646 0.001708  0.000485 nrc011  -0.004613  nrc066     -964.61710897 RS-RFO  BFGS      0    
 11   -964.61710577  0.00000093 0.001369  0.000591 nrc007  -0.007417  nrc063     -964.61710889 RS-RFO  BFGS      0    
 12   -964.61710675 -0.00000098 0.000996 -0.000310 nrc054  -0.011007  nrc065     -964.61710988 RS-RFO  BFGS      0    
 13   -964.61710671  0.00000004 0.000823 -0.000229 nrc007  -0.026707  nrc053     -964.61711333 RS-RFO  BFGS      0    
 14   -964.61711243 -0.00000572 0.001111 -0.000379 nrc005  -0.012624  nrc065     -964.61711516 RS-RFO  BFGS      0    
 15   -964.61711431 -0.00000188 0.001010 -0.000361 nrc065  -0.023032  nrc065     -964.61711811 RS-RFO  BFGS      0    
 16   -964.61712009 -0.00000579 0.001218  0.000291 nrc034  -0.002538  nrc053     -964.61712094 RS-RFO  BFGS      0    
 17   -964.61712444 -0.00000435 0.000508 -0.000269 nrc065  -0.011276  nrc065     -964.61712543 RS-RFO  BFGS      0    
 18   -964.61712149  0.00000296 0.001268 -0.000333 nrc021  -0.019446  nrc053     -964.61712622 RS-RFO  BFGS      0    
 19   -964.61712680 -0.00000531 0.000726 -0.000232 nrc065  -0.002746  nrc065     -964.61712748 RS-RFO  BFGS      0    
 20   -964.61712515  0.00000165 0.000530 -0.000210 nrc065   0.016530  nrc053     -964.61712865 RS-RFO  BFGS      0    
 21   -964.61711895  0.00000619 0.000402 -0.000146 nrc065  -0.020057  nrc053     -964.61712722 RS-RFO  BFGS      0    
 22   -964.61712187 -0.00000291 0.000851 -0.000344 nrc066   0.008633  nrc065     -964.61712378 RS-RFO  BFGS      0    
 23   -964.61712640 -0.00000454 0.000433  0.000119 nrc053   0.011038  nrc053     -964.61712727 RS-RFO  BFGS      0    
 24   -964.61712410  0.00000230 0.000226 -0.000068 nrc066  -0.010199  nrc053     -964.61712660 RS-RFO  BFGS      0    
 25   -964.61712444 -0.00000034 0.000372  0.000121 nrc053   0.004711  nrc053     -964.61712448 RS-RFO  BFGS      0    
 26   -964.61712397  0.00000047 0.000173 -0.000057 nrc066  -0.004006  nrc053     -964.61712459 RS-RFO  BFGS      0    
 27   -964.61712485 -0.00000088 0.000186 -0.000073 nrc066   0.000382  nrc063     -964.61712489 RS-RFO  BFGS      0

Offline

#2 2019-09-17 10:00:38

Ignacio
Administrator
From: Uppsala
Registered: 2015-11-03
Posts: 1,085

Re: Thresholds in SlapAf

I'd use, e.g. THRS = 1.0D-6 3.0D-4. This is the same as the normal criteria, but it will converge when the gradient is below the threshold and either of the following is satisfied:

1. The step size is below the threshold
2. The energy change is below the 1st threshold

The default is 0.0 3.0D-4, so #2 can never be satisfied, and #1 becomes required. By having a positive number you allow for #2.

Offline

Board footer

Powered by FluxBB 1.5.11

Last refresh: Today 00:48:07