[CPMD-list] how to do constraint optimization?

[Lan-Feng Yuan (lyuan@Princeton.EDU)]

----- Original Message -----
From: Salomon Billeter <ZRLSRB at ch.ibm.com>
Date: Thursday, December 8, 2005 4:10 pm
Subject: Re: [CPMD-list] how to do constraint optimization?
To: "Lan-Feng Yuan (lyuan at Princeton.EDU)" <lyuan at Princeton.EDU>
Cc: cpmd-list at cpmd.org

> Dear Lan-Feng,
> 
> which optimizer and which version of CPMD do you use? Depending on the
> version of code and the type of optimizer, you could have hit a 
> bug in the
> constraints code (which was fixed once) which caused the 
> optimization not
> to converge.
> Also, playing around with the PENALTY parameters could help. For 
> geometryoptimization (except with RFO), CPMD does not calculate 
> Lagrangemultipliers to satisfy the constraints but rather 
> harmonically penalizes
> the deviation from the target value.
> Could you please send the input?
> 
> Best regards,
> Salomon
> 
  Dear Salomon,

  Thank you. My version is 3.10 (yes, higher than the open source, I got it from a colleague who is a developer of CPMD). An example of my input is:

&INFO
(H2O)2 constraint optimization, O-O fixed at 3.25 A
&END

&CPMD
 CENTER MOLECULE
 ISOLATED MOLECULE
 OPTIMIZE GEOMETRY
 MAXSTEP
  100000
 CONJUGATE GRADIENTS IONS
 CONVERGENCE ORBITALS
  1.D-8
 CONVERGENCE GEOMETRY
  5.D-5 
&END

&SYSTEM
 SYMMETRY 
  1
 CELL
 18.89725988578923203102406997093 1. 1. 0. 0. 0.
 CUTOFF
  80.
&END

&DFT
 NEWCODE
 FUNCTIONAL PBE
&END


&ATOMS
*O_MT_PBE KLEINMAN-BYLANDER
LMAX=P
2
6.7455  9.6133  9.5180
12.2607  9.3529  9.2784
*H_MT_PBE.psp KLEINMAN-BYLANDER
LMAX=S
4
6.1602 10.9878  8.4572
6.0122  8.0957  8.8010
10.3669  9.4438  9.4475
12.8311  9.4411 11.0165
CONSTRAINTS
FIX STRUCTURE
1
DIST 1 2 6.1416094628815004100828227405521
END CONSTRAINTS
&END

  Without constraint, I did get convergence and all 12 positive frequencies, so the fundamental feature of optimization is correct. With constraint, I never get convergence.

  Thank you very much! Best regards,

  Lan-Feng