[CPMD-list] About temperature and pressure
Axel Kohlmeyer
akohlmey at vitae.cmm.upenn.edu
Wed Nov 16 19:42:06 CET 2005
On Wed, 16 Nov 2005, Yuan Pengfei wrote:
YP> DEAR all:
YP>
dear yuan,
YP> (I am sorry to sent this mail for several times, the first
YP> several mails cannot be read easy.)
please have a look at: http://expita.com/nomime.html
that page nicely explains how to send readable
_plain ascii_ mails with many popular mail clients.
YP> Dear all:
YP> I have run a parrinello-rahman NPT calculation in the HPUX_IA64
YP> machine , the final result of the temperature and the pressure are
YP> convergenced to the give one. But the vibration of the two
YP> parameters are very big , for temperature it is about 26K and for
YP> pressure about 35kbar. Do someone know how todecrease the
YP> vibrations.I want the vibration can be lessen than 1K and 1kbar.
sorry to disappoint you, but this is just the way it is
for a system of your size (60 atoms). there are bound to
be fluctuations. the easiest (for me) way to explain this,
is to consider the system as a bunch of harmonic oscillators.
the instantaneous temperature, i.e. the number you see is
computed from the kinetic energy energy per degree of freedom.
for a single oscillator (1 DOF) the kinetic energy will be
maximal at r=r_0 and zero at r=r_max. now, if you have multiple
isolated oscillators with the same parameters and in-phase, this
will add up. BUT in a CP-MD simulation you have different force
constants (that constantly change), different masses for many
atoms and they are all coupled. as a result the oscillations
will decouple, but there will still be significant fluctuations.
from a simple statistical consideration, you can see that those
fluctuations depend on the size of the system. you would need a
pretty large system (> 100000 atoms, depending on the kind of
system) to keep the fluctuations below 1K).
with pressure the situation is even more extreme, due to the
comparatively low compressibility of liquids and solids.
solids and liquids the pressure fluctuates very much with only small
changes in the local density (i.e. movement of the atoms or
the cell walls).
now, provided that your system was in equilibrium during the
simulation (after you have cut away the equilibration part,
that is), then you can take advantage of the ergodicity
hypothesis and use the time avarage over the trajectory
instead of using a larger ensemble. in any larger system,
the _local_ fluctuations with be the _same_.
there still are some, sometimes severe, finite size
effect to be considered. please have a look into a statistical
mechanics textbook for more details on this whole issue.
best regards,
axel.
YP> Thank you very much.
YP> Yuan Pengfei
YP>
YP> the input file is list here :
YP> &CPMD
YP> MOLECULAR DYNAMICS CP
YP> RESTART WAVEFUNCTION COORDINATES
YP> TIMESTEP
YP> 0.5
YP> MAXSTEP
YP> 40000
YP> SPLINE RANGE
YP> 3.0
YP> NOSE CELL
YP> 30.0D0 120000
YP> NOSE IONS
YP> 300.0D0 1500
YP> PARRINELLO-RAHMAN NPT
YP> STRUCTURE BONDS
YP> STRUCTURE ANGLES
YP>
YP> &END
YP>
YP> &DFT
YP> FUNCTIONAL BLYP
YP> &END
YP>
YP> &SYSTEM
YP> ANGSTROM
YP> PRESSURE
YP> 100.0
YP> SYMMETRY
YP> 14
YP> CELL
YP> 7.037 1.0 1.5583 0.0 0.0 -0.5
YP> CUTOFF
YP> 80.0
YP> &END
YP>
YP> &ATOMS
YP> *C_MT_BLYP_1.psp KLEINMAN-BYLANDER
YP> LMAX=P
YP> 12
YP> 1.954747 -0.359930 2.170488
YP> 2.890668 0.627950 2.817350
YP> -1.507792 1.702786 0.881039
YP> -1.301370 2.223288 2.297061
YP> -2.149606 -2.205727 -0.785117
YP> -1.797468 -2.127230 0.681563
YP> 1.954747 -0.359930 7.653488
YP> 2.890668 0.627950 8.300350
YP> -1.507792 1.702786 6.364039
YP> -1.301370 2.223288 7.780061
YP> -2.149606 -2.205727 4.697883
YP> -1.797468 -2.127230 6.164563
YP>
YP> *O_SPRIK_BLYP.psp KLEINMAN-BYLANDER
YP> LMAX=P
YP> 12
YP> 0.947259 -0.723091 2.812488
YP> 2.199202 -0.773115 1.017326
YP> -0.650672 0.913879 0.431077
YP> -2.529066 2.045850 0.235963
YP> -2.056298 -1.158235 -1.457910
YP> -2.439865 -3.326691 -1.248190
YP> 0.947259 -0.723091 8.295488
YP> 2.199202 -0.773115 6.500326
YP> -0.650672 0.913879 5.914077
YP> -2.529066 2.045850 5.718963
YP> -2.056298 -1.158235 4.025090
YP> -2.439865 -3.326691 4.234810
YP>
YP> *N_MT_BLYP.psp KLEINMAN-BYLANDER
YP> LMAX=P
YP> 6
YP> 2.685116 0.720450 4.269928
YP> -0.873421 1.138715 3.195304
YP> -0.390811 -1.769375 0.728740
YP> 2.685116 0.720450 9.752928
YP> -0.873421 1.138715 8.678304
YP> -0.390811 -1.769375 6.211740
YP>
YP> *H_MT_BLYP.psp KLEINMAN-BYLANDER
YP> LMAX=S
YP> 30
YP> 3.476329 1.135486 4.768900
YP> 1.890056 1.354134 4.395797
YP> 2.468625 -0.171189 4.707649
YP> 2.756603 1.609063 2.371401
YP> 3.878756 0.287630 2.566958
YP> -0.541399 1.516423 4.082387
YP> -1.559845 0.420835 3.424845
YP> -0.109696 0.602641 2.786094
YP> -2.200211 2.697251 2.687188
YP> -0.531017 2.989539 2.268505
YP> -0.031957 -1.535664 1.655901
YP> 0.195362 -2.499005 0.322463
YP> -0.307369 -0.906107 0.193699
YP> -1.981258 -3.073606 1.183995
YP> -2.383603 -1.356955 1.172659
YP> 3.476329 1.135486 10.251900
YP> 1.890056 1.354134 9.878797
YP> 2.468625 -0.171189 10.190649
YP> 2.756603 1.609063 7.854401
YP> 3.878756 0.287630 8.049958
YP> -0.541399 1.516423 9.565387
YP> -1.559845 0.420835 8.907845
YP> -0.109696 0.602641 8.269094
YP> -2.200211 2.697251 8.170188
YP> -0.531017 2.989539 7.751505
YP> -0.031957 -1.535664 7.138901
YP> 0.195362 -2.499005 5.805463
YP> -0.307369 -0.906107 5.676699
YP> -1.981258 -3.073606 6.666995
YP> -2.383603 -1.356955 6.655659
YP> &END
YP>
YP> Thank You
YP> ¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
YP>
YP>
YP> ¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Yuan Pengfei
YP> ¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡pfyuan at ustc.edu
YP> ¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2005-11-16
YP> _______________________________________________
YP> CPMD-list mailing list
YP> CPMD-list at cpmd.org
YP> http://cpmd.org/mailman/listinfo/cpmd-list
YP>
--
=======================================================================
Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu http://www.cmm.upenn.edu
Center for Molecular Modeling -- University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.
More information about the CPMD-list
mailing list