Molcas Forum

shuoshuo

Member

From: Beijing Normal University

Registered: 2017-10-18

Posts: 35

QMMM convergence problem

I found the convergence of QMMM calculations very difficult, especially when calculating intersections. The same QM is easy to converge in the gas phase.
Is there any good solution?
Here comes my input file and key associated with QMMM setup.

&ESPF
	external = tinker
	LAMOROKUMA
END OF INPUT

* The QM/MM electrostatics interactions are computed according to the ESPF method
* Currently only this method is available together with Molcas
*QMMM-ELECTROSTATICS ESPF 
* The MM microiteration scheme is enabled
* The other option in OFF
QMMM-MICROITERATION ON

                       Energy     Grad      Grad              Step                 Estimated   Geom       Hessian     
Iter      Energy       Change     Norm      Max    Element    Max     Element     Final Energy Update Update   Index  
  1  -2020.58389244  0.00000000 0.198798  0.089489 lnm186   0.077591  lnm034    -2020.60179847 RS-RFO  None      0    
  2  -2020.60742720 -0.02353475 0.117549  0.065210 lnm189  -0.063602  lnm038    -2020.61429036 RS-RFO  BFGS      0    
  3  -2020.61465479 -0.00722759 0.120438  0.067009 lnm187  -0.074994  lnm018    -2020.61796169 RS-RFO  BFGS      0    
  4  -2020.61345075  0.00120404 0.066719 -0.032284 lnm187  -0.073535  lnm003    -2020.61474311 RS-RFO  BFGS      0    
  5  -2020.61876628 -0.00531553 0.051496 -0.027902 lnm187  -0.120692  lnm017    -2020.62105378 RS-RFO  BFGS      0    
  6  -2020.61137482  0.00739146 0.085768  0.033120 lnm187   0.071167  lnm029    -2020.61564490 RS-RFO  BFGS      0    
  7  -2020.61853287 -0.00715805 0.050765  0.020879 lnm190   0.136195  lnm027    -2020.62344730 RS-RFO  BFGS      0    
  8  -2020.61090571  0.00762716 0.098328  0.051182 lnm187   0.072466  lnm064    -2020.61420561 RS-RFO  BFGS      0    
  9  -2020.61718610 -0.00628038 0.045825 -0.021443 lnm191  -0.183267  lnm030    -2020.62362902 RS-RFO  BFGS      0    
 10  -2020.60581608  0.01137001 0.075106 -0.027831 lnm149   0.106706  lnm063    -2020.61255726 RS-RFO  BFGS      0    
 11  -2020.61781297 -0.01199689 0.053534  0.019422 lnm190  -0.261346  lnm042    -2020.63026195 RS-RFO  BFGS      0    
 12  -2020.60086112  0.01695185 0.123546  0.062193 lnm187   0.129196  lnm065    -2020.60840338 RS-RFO  BFGS      0    
 13  -2020.61311702 -0.01225590 0.054188 -0.024069 lnm191   0.205911* lnm065    -2020.62083561 RS-RFO  BFGS      0    
 14  -2020.61175410  0.00136293 0.077820  0.034644 lnm190  -0.103623  lnm042    -2020.61821663 RS-RFO  BFGS      0    
 15  -2020.61902110 -0.00726700 0.039310 -0.020331 lnm187  -0.030601  lnm030    -2020.61982723 RS-RFO  BFGS      0    
 16  -2020.62060318 -0.00158208 0.049509 -0.030462 lnm187   0.087295  lnm018    -2020.62260748 RS-RFO  BFGS      0    
 17  -2020.60704755  0.01355563 0.165233  0.090734 lnm187   0.056966  lnm027    -2020.61370624 RS-RFO  BFGS      0    
 18  -2020.61861298 -0.01156544 0.040054  0.018983 lnm191   0.041265  lnm001    -2020.61918633 RS-RFO  BFGS      0    
 19  -2020.61864474 -0.00003176 0.115463  0.066551 lnm187   0.023325  lnm003    -2020.61926221 RS-RFO  BFGS      0    
 20  -2020.61851035  0.00013439 0.028511 -0.011222 lnm187  -0.030795  lnm001    -2020.61898581 RS-RFO  BFGS      0    
 21  -2020.62084231 -0.00233196 0.112813  0.066784 lnm187   0.086835  lnm003    -2020.62189536 RS-RFO  BFGS      0    
 22  -2020.61462922  0.00621309 0.058472  0.019927 lnm063   0.055428  lnm033    -2020.61689466 RS-RFO  BFGS      0    
 

niko

Member

From: Marseille

Registered: 2015-11-08

Posts: 59

Website

Re: QMMM convergence problem

Hi,
The QM/MM geometry optimization convergence may be improved by using both the Cartesian and the rHidden keywords in Slapaf, see https://molcas.gitlab.io/OpenMolcas/sph … iterations
Hope this helps

shuoshuo

Member

From: Beijing Normal University

Registered: 2017-10-18

Posts: 35

Re: QMMM convergence problem

Hi, niko. Thanks for your reply. I have already used  Cartesian and the rHidden keywords in Slapaf. I am wondering if there are hydrogen bonds between QM and MM regions that affect the convergence.

niko

Member

From: Marseille

Registered: 2015-11-08

Posts: 59

Website

Re: QMMM convergence problem

You may try to find which atoms carry the largest gradient components.

Also, if your molecular system includes one or several link atoms, be sure to include (at least) the corresponding frontier MM atoms in the Tinker QMMM list. If not, the Morokuma scheme is useless.