[CPMD-list] Force on constraint.
Juerg Hutter
hutter at pci.unizh.ch
Mon Jul 28 13:47:39 CEST 2003
Hi
CPMD treats fixed coordinates and intrinsic constraints differently.
Fixed coordinates are easy to handle by just ignoring the forces
acting on them. With other words the constraint force is just the
actual force on the coordinate. Intrinsic constraints, like a bond length
constraint are more difficult and need a SHAKE/RATTLE algorithm
to calculate the constraint force. It's usually this type of constraints
that is used in thermodynamic integration and it's therefore the
only forces that are written to the CONSTRAINT file.
> Dear everyone,
>
> I am doing constrained CPMD. I want to get the free energy vs. reaction
> coordinate (RC) by integrating the average force along RC. To do that, I
> printed out forces on ions at every MD step and want to get mean forces
> after time average. As a result, I found instantaneous 'forces on
> constrained(fixed) ions' are all zero in every MD step. On the other hand,
> the forces on free ions have some values but the averaged forces are closed
> to zero. My questions are,
>
> (1) Does CPMD compute forces on constrained ion, which are fixed in MD run?
Yes, see above.
> (2) If answer of (1) is yes, does CPMD reset forces on constrained ion to
> zero after computing force?
Yes, see above.
> (3) How can I get forces on the constraint?
You can print the forces before they are set to zero in forces.F.
IF(PARENT) THEN
CALL SYMVEC(FION,SCR,LSCR)
CALL TAUCL(FION) <--- "clean" forces
CALL GSIZE(FION,GNMAX,GNORM)
ENDIF
For example in puttau.F in subroutine taucl
DO IS=1,NSP
DO IA=1,NA(IS)
IAT=IAT+1
DO L=1,3
IF(LSKCOR(L,IAT).EQ.0) WRITE(xx,*) L,iat,TAU(L,IA,IS)
IF(LSKCOR(L,IAT).EQ.0) TAU(L,IA,IS)=0.0D0
ENDDO
ENDDO
ENDDO
regards
Juerg Hutter
>
> Thanks for any comments in advance.
>
> Yi
> ----
> Dr. Yi Liu
> Department of Chemistry
> The University of Western Ontario
> London, Ontario, N6A 5B7, Canada
>
> Email: yliu252 at uwo.ca
> Phone: (519) 661-2111, ext. 88882
> Fax: (519) 661-3022
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://www.cpmd.org/mailman/listinfo/cpmd-list
>
More information about the CPMD-list
mailing list