[Pw_forum] forces NaNQ

Giacomo Saielli giacomo.saielli at unipd.it
Mon Mar 14 16:45:09 CET 2005


Dear users and developers,
I am trying an MD simulation using cp.x. The starting config. is taken 
from the final config. of a classical MD.
At first I minimize the electronic wave function for some timesteps 
(e.g. 50) using steepest descent, then
I switch to damp, calculating the damping factor as indicated in the 
manual. I noted that, as soon as I used the damp option,
forces on some atoms  become NaNQ (at CINECA IBM SP4, aix 5.2). I also 
noted that the same forces calculated with the option steepest descent 
in the preceeding run are very high during the minimization of the wave 
function, of the order of 10+3 or more. I suspect this has to do with 
some inappropriate internal (stretching and/or bending) force field 
parameters used in the classical MD run.

However, surprisingly, the 2nd minimization using damp actually 
converged! (see the output below)

The first question is if this behaviour is expected, I mean if in case 
of out-of-equilibrium ionic positions one might have problems in 
minimizing the electronic wave function using 'damp' option.
If it is not expected I would like to know your opinion about what other 
sources of problems I might have in my system.

Since I would like to minimize the wave function, then  the ionic 
position, and, eventually start the MD cp.x run all with espresso, if 
this behaviour is expected, which is the best way  to minimize the wave 
function in order to avoid NaNQ forces? Continuing with the option 
steepest descent up to the end or there is some other possibility?

Thank you very much for your for your help.

I include here the input file and an excerpt of the output
best wishes,
Giacomo

----- INPUT------

 &control
    title = '',
    calculation = 'cp',
    restart_mode = 'from_scratch',
    ndr = 51,
    ndw = 51,
    nstep  = 50,
    iprint = 1,
    isave  = 1,
    tstress = .TRUE.,
    tprnfor = .TRUE.,
    dt    = 5.0d0,
    etot_conv_thr = 1.d-9,
    ekin_conv_thr = 1.d-5,
    prefix = '',
    pseudo_dir='',
    outdir='',
    wf_collect=.TRUE.,
/
 &system
    ibrav = 14,
    celldm(1) = 22.69808,
    celldm(2) = 1.0,
    celldm(3) = 1.0,
    celldm(4) = 0.0,
    celldm(5) = 0.0,
    celldm(6) = 0.0,
    nat = 160,
    ntyp = 4,
    nbnd = 208,
    nelec = 416,
    ecutwfc = 25.0,
    nr1b= 10, nr2b = 10, nr3b = 10,
    xc_type = 'PBE',
 /
 &electrons
    emass = 400.d0,
    emass_cutoff = 2.5d0,
    electron_dynamics = 'sd',
% this line becomes
%  electron_dynamics = 'damp',
%  electron_damping = 0.12,
%  in the successive run that produce NaNQ forces.
 /
 &ions
    ion_dynamics = 'none',
 /
ATOMIC_SPECIES
 Cl 35.45d0  Cl.pbe-n-van.UPF
 N  14.01d0  N.pbe-van_bm.UPF
 C  12.01d0  C.pbe-van_bm.UPF
 H  1.0079d0 H.pbe-van_bm.UPF
ATOMIC_POSITIONS (bohr)
....
....
....
....

------ END OF OUTPUT  USING DAMP FOR WAVE-FUNC-----
.....
.....
   Atomic Forces (AU)

   STEP:    380   0.045959
 -0.148009E-02 -0.428108E-02  0.729338E-03
  0.505482E-03  0.148271E-02 -0.201681E-02
 -0.231406E-02 -0.873777E-03  0.298936E-03
 -0.673673E-03 -0.141326E-02 -0.154087E-02
 -0.301043E-03 -0.258251E-02  0.260788E-02
  0.755572E-03  0.261011E-02  0.342261E-02
  0.160367E-02  0.955647E-03 -0.274878E-02
  0.296522E-02  0.204936E-02  0.345995E-02
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
...
...
...

          -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ
         -NaNQ         -NaNQ         -NaNQ


 subroutine rhoofr: total integrated electronic density
 in g-space =416.000000     in r-space =416.000000


                total energy =     -612.41136 a.u.
              kinetic energy =      342.32906 a.u.
        electrostatic energy =     -836.84425 a.u.
                         esr =        0.00694 a.u.
                       eself =     1314.91376 a.u.
      pseudopotential energy =      -37.29305 a.u.
  n-l pseudopotential energy =      108.17482 a.u.
 exchange-correlation energy =     -188.77794 a.u.
           average potential =       -0.28148 a.u.


 cell parameters h
     22.69808      0.00000      0.00000
      0.00000     22.69808      0.00000
      0.00000      0.00000     22.69808


 
 derivative of e(tot)
     -0.24068     -0.02804     -0.00763
     -0.02804     -0.23236     -0.01792
     -0.00763     -0.01792     -0.19040


 


 eigenvalues at k-point:  0.000 0.000 0.000
  -23.10  -23.08  -22.86  -22.72  -22.66  -22.45  -22.33  -22.17  
-20.08  -19.90
  -19.76  -19.57  -19.50  -19.49  -19.24  -19.19  -15.98  -15.93  
-15.79  -15.69
  -15.65  -15.62  -15.55  -15.52  -15.16  -15.04  -14.83  -14.79  
-14.76  -14.69
  -14.65  -14.57  -13.97  -13.96  -13.87  -13.79  -13.78  -13.71  
-13.59  -13.51
  -12.57  -12.53  -12.49  -12.47  -12.47  -12.36  -12.33  -12.32  
-12.21  -12.17
  -11.92  -11.90  -11.88  -11.85  -11.81  -11.79  -11.69  -11.66  
-11.42  -11.38
  -11.37  -11.22  -11.17  -11.10  -10.86  -10.62  -10.56  -10.54  
-10.52  -10.34
  -10.21  -10.17   -9.70   -9.58   -9.40   -9.22   -9.19   -9.17   
-9.15   -9.09
   -8.52   -8.46   -8.23   -8.21   -8.17   -8.06   -7.92   -7.88   
-7.84   -7.78
   -7.61   -7.55   -7.48   -7.46   -7.39   -7.34   -7.29   -7.26   
-7.23   -7.20
   -7.19   -7.13   -7.11   -7.08   -7.05   -7.01   -6.98   -6.97   
-6.94   -6.93
   -6.91   -6.89   -6.88   -6.79   -6.75   -6.73   -6.71   -6.68   
-6.64   -6.63
   -6.53   -6.46   -6.45   -6.42   -6.35   -6.20   -6.18   -6.09   
-6.06   -6.04
   -5.94   -5.86   -5.85   -5.79   -5.75   -5.64   -5.54   -5.51   
-5.43   -5.36
   -5.32   -5.30   -5.29   -5.23   -5.19   -5.17   -5.12   -5.09   
-5.05   -5.03
   -4.94   -4.83   -4.69   -4.60   -4.59   -4.55   -4.51   -4.48   
-4.43   -4.38
   -4.26   -4.21   -4.14   -4.07   -3.99   -3.95   -3.87   -3.84   
-3.18   -3.16
   -3.06   -2.91   -2.81   -2.70   -2.67   -2.57   -2.34   -2.28   
-2.18   -2.15
   -2.06   -2.02   -1.99   -1.72   -0.91   -0.87   -0.86   -0.86   
-0.84   -0.83
   -0.81   -0.81   -0.79   -0.78   -0.78   -0.76   -0.72   -0.69   
-0.67   -0.64
   -0.63   -0.62   -0.61   -0.56   -0.55   -0.54   -0.50   -0.48
 
 
 
 nfi    ekinc  temph  tempp        etot      enthal       econs       
econt    vnhh   xnhh0    vnhp   xnhp0
  381  0.00000    0.0    0.0  -612.41136  -612.41136  -612.41136  
-612.41136  0.0000  0.0000  0.0000  0.0000

   MAIN:          EKINC   (thr)          DETOT   (thr)       MAXFORCE   
(thr)
   MAIN:   0.699383D-10  0.1D-04  0.920522D-09  0.1D-08  0.000000D+00  
0.1D+12
   MAIN: convergence achieved for system relaxation


              averaged quantities :
         ekinc          ekin          epot          etot     tempp
       0.89934     337.48442   -1036.50923    -593.36580       0.0
     initialize   :    23.84s CPU
     total_time   :  7212.45s CPU (     371 calls,  19.441 s avg)
     formf        :     2.41s CPU
     rhoofr       :   703.04s CPU (     371 calls,   1.895 s avg)
     vofrho       :   176.18s CPU (     371 calls,   0.475 s avg)
     dforce       :  2663.18s CPU (   38584 calls,   0.069 s avg)
     calphi       :    67.31s CPU (     371 calls,   0.181 s avg)
     ortho        :   820.81s CPU (     371 calls,   2.212 s avg)
     updatc       :    34.18s CPU (     371 calls,   0.092 s avg)
     newd         :    28.59s CPU (     371 calls,   0.077 s avg)
     calbec       :     0.76s CPU (     372 calls,   0.002 s avg)
     prefor       :     5.66s CPU (     371 calls,   0.015 s avg)
     strucf       :     4.27s CPU (     372 calls,   0.011 s avg)
     nlfl         :   195.21s CPU (     371 calls,   0.526 s avg)
     nlfq         :   349.83s CPU (     371 calls,   0.943 s avg)
     set_cc       :     0.44s CPU (     371 calls,   0.001 s avg)
     rhov         :     8.34s CPU (     371 calls,   0.022 s avg)
     nlsm1        :   628.09s CPU (    1114 calls,   0.564 s avg)
     nlsm2        :   322.03s CPU (     371 calls,   0.868 s avg)
     forcecc      :     2.73s CPU (     371 calls,   0.007 s avg)
     fft          :    84.25s CPU (    7049 calls,   0.012 s avg)
     ffts         :     5.69s CPU (     742 calls,   0.008 s avg)
     fftw         :  2000.10s CPU (  115752 calls,   0.017 s avg)
     rsg          :    31.04s CPU (     371 calls,   0.084 s avg)
     setfftpara   :     0.06s CPU
     reduce       :  1610.64s CPU ( 3408619 calls,   0.000 s avg)


====================== end cprvan ======================


Estimated size (kB) of each process: 120816

   PE(1) MEMORY STATISTICS
   total space in arena :370935904
     CP           :     2h 0m CPU time

-- 
Giacomo Saielli
Istituto per la Tecnologia delle Membrane del CNR, Sezione di Padova;
Dipartimento di Scienze Chimiche, Via Marzolo, 1 - 35131 Padova, Italy; 
Tel: +39-049-8275279; Fax: -5239;
http://www.chfi.unipd.it/home/g.saielli/




More information about the Pw_forum mailing list