[CPMD-list] variable cell optimization troubles

[radbalu at comcast.net]

Dear All,

I am trying a variable cell optimization of a molecular crystal using DCACP based pseudopotentials. when I used regular TM psuedos without the dispersion
the geometry converged. However when I used DCACP terms and the following input I have problems getting convergence:

&CPMD
 OPTIMIZE GEOMETRY
 STRESS TENSOR
     5
 PARRINELLO-RAHMAN
 STEEPEST DESCENT  CELL
 PRINT FORCES ON
  PCG MINIMIZE NOPRECONDITIONING
  TIMESTEP
   10
  EMASS
   600
 CONJUGATE GRADIENTS IONS
&END

&SYSTEM
 ANGSTROM
 SCALE
 CELL VECTORS

I have some difficulty in understanding the output. The optimization proceeds nicely up to a point and then suddenly the message
"**    STRESS| SIZE OF THE PROGRAM IS  257652/24861248 kBYTES ***" gets printed implying the stress tensor is calculated.
Then there is a discontinuity in the energy and the optimization proceeds in a different direction without ever converging.

163  3.637E-05   1.585E-08   -4186.607885   -1.273E-10     47.06
 LINE SEARCH : LAMBDA=.447E-01 PREDICTED ENERGY = -4186.607884694
 164  3.448E-05   1.494E-08   -4186.607885   -1.073E-10     47.02
 LINE SEARCH : LAMBDA=.447E-01 PREDICTED ENERGY = -4186.607884694
 165  3.268E-05   1.409E-08   -4186.607885   -9.277E-11     46.99
 LINE SEARCH : LAMBDA=.447E-01 PREDICTED ENERGY = -4186.607884695
 166  3.097E-05   1.329E-08   -4186.607885   -8.822E-11     47.03
 LINE SEARCH : LAMBDA=.447E-01 PREDICTED ENERGY = -4186.607884695
 167  2.936E-05   1.254E-08   -4186.607885   -7.458E-11     47.07
 ***    STRESS| SIZE OF THE PROGRAM IS  257652/24861248 kBYTES ***
 LINE SEARCH : LAMBDA=.937E-01 PREDICTED ENERGY = -1375.213616473
 168  4.534E-01   1.020E-02   -4186.607885   -6.639E-11     87.46
 LINE SEARCH : LAMBDA=.281     PREDICTED ENERGY = -1687.717952139
 169  1.061E+00   1.791E-02   -1687.594684    2.499E+03     87.31
 LINE SEARCH : LAMBDA=.422     PREDICTED ENERGY = -1687.815386406
 170  1.061E+00   1.791E-02   -1687.704256   -1.096E-01     87.89
 LINE SEARCH : LAMBDA=.633     PREDICTED ENERGY = -1687.920018408
 171  1.061E+00   1.791E-02   -1687.803039   -9.878E-02     87.39
 LINE SEARCH : LAMBDA=.949     PREDICTED ENERGY = -1688.030157672
 172  1.061E+00   1.791E-02   -1687.907021   -1.040E-01     87.23
 LINE SEARCH : LAMBDA=1.42     PREDICTED ENERGY = -1688.146094082
 173  1.061E+00   1.791E-02   -1688.016476   -1.095E-01     87.34


Could some one help understand what is going on and point out anything that is odd in the way I set up the input?

Thank you in advance for your suggestions.

Rad