[CPMD-list] Segmentation fault for a molecular dynamics run
Axel Kohlmeyer
axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Fri Apr 9 16:22:51 CEST 2004
On Fri, 9 Apr 2004 21:50:35 +0800 "Dai Ling" wrote:
> Dear Axel:
>
> Thanks for your kind suggestions. I have tried the program with smaller timestep but it results
the same problem, here is my input file. Is there any improvements to be made?
>
> Thanks & Best Regards!
> Scott
>
> &CPMD
> MOLECULAR DYNAMICS
you obviously did not do a wavefunction optimization first
(i already suspected that from your initial output, as it is a typical
'beginners error', but it is not the only possible source).
in order for the CP-method to work, you want to stay close to
the BO-surface, therefore you first have to get there.
thus you first calculate the (ground state) electronic structureof your
system, i.e. you have to do a WAVEFUNCTION OPTIMZATION job first.
then you have to add the line
RESTART WAVEFUNCTION COORDINATES LATEST
to the md-job input to read in the electronic structure
to start you md from there. if you want to continue the md
seamlessly later, you have to add VELOCITIES for the
successive runs as well.
in your example you start from the initial guess
for based on pseudo atomic orbitals. therefore you
are hardly close to the BO-surface _and_ realistic results.
> SPLINE POINTS
> 500
this is too low for your type of pseudopotential.
use 3000.
> TEMPERATURE
> 300
note that this sets only the inital temperature.
for your test system you may want to use something
like
TEMPCONTROL IONS
300 50
as well, until the system has somewhat equilibrated
and you can use a 'real' thermostat or run without
temperature control.
> TIMESTEP
> 1
use 4. (you can raise this to 6 when you use deuterium
instead of hydrogen, i.e. add the lines
ISOTOPES
12.0
2.0
to the &ATOMS section.
this should get you a little farther.
good luck,
axel kohlmeyer.
> &END
> &SYSTEM
> ANGSTROM
> CELL ABSOLUTE DEGREE
> 10.3651 10.3651 33.6851 90.0000 90.0000 90.0000
> SYMMETRY
> 1
> CUTOFF
> 29.4009555
> 400eV
this is far too low for your type of pseudopotential.
to determine a good cutoff value, do repeated wavefunction
optimizations with the option PRINT FORCES ON and check
for convergence of the forces (you can save time when you
use RESTART WAVEFUNTION LATEST after you have converged the
first wavefunction.) try cutoff values from 50 to 150,
about 80 should be the minimum usable value, 100 quite good,
and after 150 it will rarely get much better.
if you just want to do a test md, 120 should be a safe bet,
if you want to do a fast test, try 80, but you risk unphysical
results.
> &END
> &DFT
> FUNCTIONAL BLYP
> &END
> &ATOMS
> Carbon
> *C_SG_BLYP
> LMAX=P
> 64
> -7.388012732 -1.642168534 25.01440235
> -6.382609051 -0.978175454 24.63538093
> -5.323723058 -0.089043005 24.05093371
> -4.854726663 1.088428581 24.69780765
> -4.719722672 -0.45536938 22.77058871
> -3.853670574 1.878791233 24.05481075
> -3.727203589 0.322337745 22.12793105
> -3.285603018 1.529112936 22.75270539
> -2.31179763 2.370312667 22.06837233
> -1.657811935 3.083353134 21.29678098
> -0.969260023 3.919725664 20.31394199
> -1.245222179 5.332024449 20.32965785
> -0.794927554 6.149140856 19.26327506
> -0.133808243 3.346577987 19.2650768
> 0.066483028 5.554478915 18.21924126
> 0.358692971 4.168632108 18.25181672
> 0.760322484 6.096840284 17.0399977
> 1.330943039 5.969193527 15.93724099
> 2.064764037 4.551925203 13.40019082
> 2.110723841 5.519109311 14.75412533
> 2.844470922 2.580936109 9.150681578
> 3.206511944 4.244947858 12.30471085
> 3.721941516 6.077936036 14.60025391
> 3.480826739 3.50950027 8.637025095
> 4.333974732 5.834360573 9.068761058
> 4.201365474 4.691822097 8.207886525
> 4.6564382 4.794709024 12.38031206
> 4.84559896 5.741309672 13.58730304
> 5.217790181 6.854258569 8.696382427
> 4.86446849 4.668075611 6.964101537
> 5.69689119 4.469174387 11.4296495
> 5.6736301 5.740027018 6.581398159
> 5.87675123 6.862179387 7.443676852
> 6.498456315 4.162042335 10.54699918
> 6.911204616 7.838693848 7.08030164
> 7.133496501 2.42842985 8.798213476
> 7.161538395 3.753563367 9.319849702
> 7.820688397 8.377438701 6.438704627
> 7.783175783 4.742433501 8.539547392
> 7.73090042 2.056853041 7.570379483
> 8.252167403 7.725651567 4.102205182
> 8.344226381 4.402685273 7.304779751
> 8.617559259 8.549669206 5.227281327
> 8.391879435 3.069415265 6.803032604
> 8.859026526 7.896469794 2.855611781
> 9.705998505 9.45431933 4.991295902
> 9.22919318 2.972069614 5.601124026
> 9.918410068 8.839326375 2.643068958
> 10.37954295 9.617247186 3.776823625
> 10.06424046 3.441634566 4.792302618
> 10.35035758 5.416632762 2.945596858
> 10.53537128 9.082036274 1.318533568
> 10.86952412 4.345838673 3.966050179
> 11.09091158 9.230083247 0.217174145
> 11.19845421 6.349572904 2.050379437
> 11.37106703 9.383173339 -2.273832771
> 11.90967153 9.455959601 -0.974837543
> 12.15396279 9.686972916 -3.387359715
> 12.38778786 4.206898028 4.217927143
> 12.75866426 6.191201085 2.410354645
> 13.28457026 9.743101186 -0.78930858
> 13.30994919 5.261226317 3.528976273
> 13.51585246 9.97209375 -3.1888067
> 14.09264366 9.965891298 -1.895177182
> Hydrogen
> *H_SG_BLYP
> LMAX=S
> 36
> -8.28627065 -2.221067521 25.32458106
> -5.314762155 1.404516996 25.69734891
> -5.037422356 -1.408474281 22.27937223
> -3.542887875 2.826165676 24.5564517
> -3.267304898 0.013662949 21.13173595
> -1.936650946 5.76726208 21.12205615
> -1.079491558 7.223681278 19.25296379
> -0.002696172 2.250159556 19.16440929
> 1.318563979 3.936515068 12.7808767
> 0.979181639 3.704700398 17.416767
> 2.20623476 1.82672308 9.624491821
> 2.932778866 3.53733391 11.50000647
> 3.833335676 5.913336627 10.06093243
> 4.150794282 6.838793694 15.3040995
> 5.370120666 7.642366536 9.468855023
> 4.773184306 3.763289962 6.320925188
> 5.813356723 6.260950236 13.76031594
> 6.222505306 5.670580942 5.624376089
> 6.625950492 1.635336454 9.391389515
> 7.480756175 6.93225646 4.291961097
> 7.82778424 5.790369108 8.898001536
> 7.62979369 0.942071996 7.390266623
> 8.829068316 5.215307381 6.730866222
> 8.523895065 7.234066969 2.019013533
> 10.09118356 10.1193112 5.810743664
> 9.257100872 5.391549447 2.984442268
> 11.21018477 10.35863759 3.71808614
> 10.29944541 9.165648173 -2.407440086
> 11.06682354 7.128203412 1.154317173
> 11.70110255 9.732722955 -4.399371766
> 12.95767729 3.414805613 4.779787108
> 13.2792451 6.833931151 1.628962234
> 13.45640309 9.701370504 0.330786893
> 14.14605727 10.1441285 -4.080386464
> 14.41749211 5.126797661 4.075302533
> 15.18151391 10.09264052 -1.760186938
> &END
> &CONSTRAINTS
> FIX COORDINATES
> 2
> 98 0 0 0
> 99 0 0 0
>
> &END CONSTRAINTS
>
>
> -----Original Message-----
> From: Axel Kohlmeyer [mailto:axel.kohlmeyer at theochem.ruhr-uni-bochum.de]
> Sent: Wed 4/7/2004 12:48 PM
> To: cpmd-list at cpmd.org; Dai Ling
> Cc:
> Subject: Re: [CPMD-list] Segmentation fault for a molecular dynamics run
>
>
>
>
> On Wed, 7 Apr 2004, Dai Ling wrote:
>
> DL> I met a problem when runing a molecular dynamics (CP) run with some benzene ring struc
tures. There is some odd marks as "NaNQ" in the output file as shown below. Then the running fail
ed with "Segmentation fault". Who can help me to point out what goes wrong?
> DL>
> DL> Thanks in advance!
> DL>
> DL> Scott
> DL>
> DL> Outpupt:
> DL>
> DL> ....
> DL> ....
> DL> TOTAL INTEGRATED ELECTRONIC DENSITY
> DL> IN G-SPACE = 292.000000
> DL> IN R-SPACE = 292.000000
> DL> (K+E1+L+N+X) TOTAL ENERGY = -370.10939613 A.U.
> DL> (K) KINETIC ENERGY = 247.71353333 A.U.
> DL> (E1=A-S+R) ELECTROSTATIC ENERGY = -328.45005481 A.U.
> DL> (S) ESELF = 352.39900035 A.U.
> DL> (R) ESR = 17.54351703 A.U.
> DL> (L) LOCAL PSEUDOPOTENTIAL ENERGY = -214.10105210 A.U.
> DL> (N) N-L PSEUDOPOTENTIAL ENERGY = 41.28274138 A.U.
> DL> (X) EXCHANGE-CORRELATION ENERGY = -116.55456393 A.U.
> DL> GRADIENT CORRECTION ENERGY = -5.18331838 A.U.
> DL>
> DL> NFI EKINC TEMPP EKS ECLASSIC EHAM DIS TCPU
> DL> 1 3.11966 309.0 -370.10940 -369.96407 -366.84441 .000 48.41
>
> DL> FILE ./TRAJECTORY EXISTS, NEW DATA WILL BE APPENDED
> DL> FILE ./MOVIE EXISTS, NEW DATA WILL BE APPENDED
> DL> 2 4.88578 320.2 -371.88325 -371.73269 -366.84691 .000 49.49
>
> hmmmm, EKINC seems a bit high, and is fluctaing quite a bit.
> also EHAM ist not the conserved quantity that is should be.
> seems like are not very close to the BO-surface anymore
> (if ever) as you should be.
>
> could you please post your input file, so that we can have
> a better guess at the problem. my guess would be that you
> have a too large timestep or starte from a non-converged
> wavefunction.
>
> DL> 3 3.92472 319.0 -370.49747 -370.34748 -366.42276 .000 48.96
> DL> 4 4.22976 305.3 -371.13380 -370.99025 -366.76049 .000 52.20
> DL> 5 3.52748 288.6 -370.19907 -370.06334 -366.53586 .001 55.75
> DL> 6 3.28840 276.3 -369.96661 -369.83670 -366.54830 .001 54.76
> DL> 7 3.79151 269.8 -370.58147 -370.45460 -366.66309 .001 43.70
> DL> 8 3.77854 268.6 -370.52295 -370.39664 -366.61810 .001 43.22
> DL> 9 3.59004 273.2 -370.32378 -370.19530 -366.60527 .002 44.05
> DL> 10 3.48053 283.8 -370.23346 -370.09998 -366.61946 .002 43.19
> DL> .......................
> DL> .......................
> DL> 248 3.57992 1050.5 -370.66655 -370.17254 -366.59262 .270 42.99
> DL> 249 3.57719 1046.8 -370.66159 -370.16935 -366.59216 .269 41.47
> DL> 250 3.57912 1042.9 -370.66254 -370.17212 -366.59300 .269 38.90
> DL> 251 3.57836 1039.3 -370.65993 -370.17121 -366.59285 .268 43.86
> DL> 252 3.57595 1036.0 -370.65579 -370.16860 -366.59265 .267 42.58
> DL> 253 3.57593 1033.0 -370.65463 -370.16889 -366.59296 .267 40.59
> DL> 254 3.57641 1029.8 -370.65402 -370.16977 -366.59335 .266 44.35
> DL> 255 3.57504 1026.4 -370.65056 -370.16789 -366.59285 .266 44.05
> DL> 256 3.57670 1022.9 -370.65037 -370.16934 -366.59264 .265 43.99
> DL> 257 3.58117 1019.5 -370.65368 -370.17426 -366.59308 .265 44.41
>
> ok, your temperature has risen to >1000K (is that intentional?) and your
> Kohn-Sham energy has dropped. it seems that your starting structure was
> not in equilibrium. OTOH, if you are too far away from the BO-surface,
> it could mean anything or nothing.
>
> DL> 258 -NaNQ 1016.4**************************** -NaNQ .265 42.87
> DL> 259 NaNQ -NaNQ NaNQ -NaNQ -NaNQ .264 43.87
> DL> 260 NaNQ -NaNQ NaNQ -NaNQ -NaNQ NaNQ 42.20
> DL> 261 NaNQ -NaNQ NaNQ -NaNQ -NaNQ NaNQ 37.84
> DL> 262 NaNQ -NaNQ NaNQ -NaNQ -NaNQ NaNQ 38.56
>
> well seems like the fictitious dynamic of the electronic system
> could not keep up with atoms/ions anymore. again, it could be from a too
> large time step.
>
> hope this helps,
> axel kohlmeyer.
>
> DL> _______________________________________________
> DL> CPMD-list mailing list
> DL> CPMD-list at cpmd.org
> DL> http://www.cpmd.org/mailman/listinfo/cpmd-list
> DL>
> DL>
>
> --
>
>
> =======================================================================
> Dr. Axel Kohlmeyer e-mail: axel.kohlmeyer at rub.de
> Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673
> Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045
> D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
> =======================================================================
>
>
>
>
--
=======================================================================
Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045
D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.
More information about the CPMD-list
mailing list