[gmx-users] Re: angle constraints

[Thomas Schlesier]

Thanks for the clarification.
Forgot that virtual sites have no mass. With this info it is clear why a 
setup with 2 'normal' and one dummy atom would not work.


Am 26.07.2012 01:42, schrieb gmx-users-request at gromacs.org:
> On 26/07/2012 4:12 AM, Broadbent, Richard wrote:
>> >Virtual sites are by definition have no mass.
>> >
>> >If you simply ignore the mass of the carbon the molecule will be too light
>> >and its translational momentum will therefore be too small meaning it will
>> >move too quickly.
>> >
>> >If you place half the mass of the carbon on each oxygen the moment of
>> >inertia will be wrong and the molecule will spin too slowly.
> All correct so far.
>
>> >
>> >In practice you have to decide what you want to loose or if a balance
>> >between the two is better.
> Not true, as illustrated by the link I gave earlier in the thread, which
> nobody seems to have read/understood.
>
> One needs at least two massive particles to describe the available
> degrees of freedom of a linear molecule, and using exactly two
> side-steps the angle constraint issue. Each must have half the total
> mass of CO2 and the distance between them is chosen to reproduce the
> moment of inertia. These will not be in suitable positions to have
> non-bonded interactions, of course. Then three (massless) virtual sites
> are constructed from those two, and these are the only ones that have
> the non-bonded interactions.
>
>> >
>> >Richard
>> >
>> >
>> >On 25/07/2012 14:44, "Thomas Schlesier"<schlesi at uni-mainz.de>  wrote:
>> >
>>> >>Ok, read the topic about the acetonitril. But i'm somewhat clueless:
>>> >>
>>> >>Why is the following setup wrong:
>>> >>Use 2 particles as normal atoms. Put the third as a dummy in between.
>>> >>Give each particle its 'normal' mass?
>>> >>I would assume that this system should have the right mass and moment of
>>> >>inertia, due to the fact the all individual masses and the positions one
>>> >>the particles would be correct.
> The virtual site so constructed cannot have mass, so this cannot be an
> accurate model.
>
>>> >>
>>> >>Only idea i have, why this setup could be flawed, would be that the
>>> >>third particle does only interact indirectly through the other two
>>> >>particles (i mean, virtual site interacts normally with all othe
>>> >>particles, but the force which would act on the dummy get redistributed
>>> >>to the other particles)... and then it's mass does not come into play,
>>> >>since it new position is determined only by the other two particles. so
>>> >>the complete molecule would move with a reduced mass?!?
> Still not an accurate model - you'd have a CO2 with three sites and mass
> only at two of the sites, so either the mass or moment of intertia must
> be wrong.
>
> Mark
>
>>> >>
>>> >>Can anyone comment on this?
>>> >>
>>> >>greetings
>>> >>thomas
>>> >>
>>> >>
>>> >>On 25/07/2012 10:08 PM, Thomas Schlesier wrote:
>>>> >>>What you have done there looks very strange...
>>>> >>>easiest wy would be:
>>>> >>>define the two oxygens as normal atoms (1,2), give them a bondlength
>>>> >>>twotimes the C-O bond length
>>>> >>>define the carbon as a dummy (3), while you construct it's position
>>>> >>>from both oxygens with a=0.5
>>>> >>>one thing i don't know is how to handle the mass:
>>>> >>>1) give both oxygen half of the system mass
>>>> >>>2) give all atoms their normal mass
>>>> >>>would tend to (2)
>>> >>One should want to get both the total mass and the moment of inertia
>>> >>correct...
>>> >>http://lists.gromacs.org/pipermail/gmx-users/2003-September/007095.html.
>>> >>
>>> >>Mark
>>> >>
>>>> >>>greetings
>>>> >>>thomas
>>>> >>>
>>>> >>>Am 25.07.2012 13:15, schrieb gmx-users-request at gromacs.org:
>>>>> >>>>How to choose the positions of the dummy atoms while constraining the
>>>>> >>>>angle for a linear triatomic molecule?
>>>>> >>>>The topology for a such molecule ( af for example CO2 ) is as follows
>>>>> >>>>
>>>>> >>>>[ moleculetype ]
>>>>> >>>>; Name            nrexcl
>>>>> >>>>CO2      2
>>>>> >>>>
>>>>> >>>>[ atoms ]
>>>>> >>>>;   nr       type  resnr residue  atom   cgnr     charge mass
>>>>> >>>>typeB    chargeB      massB
>>>>> >>>>; residue 503 CO  rtp CO   q  0.0
>>>>> >>>>        1         D1    503     CO     D1      1          0 21.90158
>>>>> >>>>     ; qtot 0
>>>>> >>>>        2         D2    503     CO     D2      2          0 21.90158
>>>>> >>>>     ; qtot 0
>>>>> >>>>        3         CE    503     CO     CO      3        0.7 0.00000
>>>>> >>>>     ; qtot 0.7
>>>>> >>>>        4         OE    503     CO    OC1      4      -0.35 0.00000
>>>>> >>>>     ; qtot 0.35
>>>>> >>>>        5         OE    503     CO    OC2      5      -0.35 0.00000
>>>>> >>>>     ; qtot 0.35
>>>>> >>>>[ constraints ]
>>>>> >>>>;  ai  aj funct           b0
>>>>> >>>>1 2 1   0.2000
>>>>> >>>>
>>>>> >>>>[ dummies2 ]
>>>>> >>>>;  ai    aj    ak       funct   a
>>>>> >>>>       3     1     2       1       0.0170
>>>>> >>>>       4     1     2       1       0.1000
>>>>> >>>>       5     1     2       1       0.2170
>>>>> >>>>
>>>>> >>>>
>>>>> >>>>[ exclusions ]
>>>>> >>>>3 4 5
>>>>> >>>>4 5 3
>>>>> >>>>5 4 3
>>>>> >>>>
>>>>> >>>>
>>>>> >>>>
>>>>> >>>>The .rtp file for CO2
>>>>> >>>>
>>>>> >>>>[ CO ]
>>>>> >>>>    [ atoms ]
>>>>> >>>>       D1     D1          0.0000           1
>>>>> >>>>       D2     D2          0.0000           2
>>>>> >>>>       CO    CE          0.7000        3
>>>>> >>>>       OC1  OE         -0.3500       4
>>>>> >>>>       OC2  OE         -0.3500        5
>>>>> >>>>    [ bonds ]
>>>>> >>>>       CO      OC1
>>>>> >>>>       CO      OC2
>>>>> >>>>
>>>>> >>>>
>>>>> >>>>Can anyone please check above file parts whether I'm doing correct or
>>>>> >>>>not ?