[gmx-users] Re: cutoff problem

[XAvier Periole]

Vitaly,

your are right in principle but for "real" systems long-range LJ
interactions do matter a lot.

In addition Justin reference the following is a good example of
the effect of treatment of long-range interaction:
Anezo et al. J. Phys. Chem. B 2003, 107, 9424-9433

XAvier.

On Mar 25, 2010, at 11:46 AM, Vitaly V. Chaban wrote:

> XAvier,
>
> You should probably know that E(LJ)(r)->0 at r->2.5*sigma. This is a
> result of pure mathematical analysis. If all the sigmas in your system
> are less than 0.4 nm, then after r=1.0 nm, the energy is very close to
> zero.
>
> All your below statement are correct but they do not address the
> problem if E(LJ)->0 at R(cutoff). This was the only hint I posted
> before.
>
> Vitaly
>
> On Thu, Mar 25, 2010 at 12:37 PM, XAvier Periole <x.periole at rug.nl>  
> wrote:
>>
>> the development of biomolecular force fields in general have tough
>> the community that cutoff values are important parameters for any
>> force field, this is true for both Elec and vdW interactions.
>>
>> The example of a bilayer was the most obvious to me when thinking
>> of the effect of the vdW cutoff. The balance between the Elec and vdW
>> is determinant to the properties of the bilayer. No particular atom- 
>> pair
>> in mind, though.
>>
>> It important not to give the wrong impression to people new in the
>> field. The literature is full of reports of cutoff artefacts ...
>>
>> On Mar 25, 2010, at 9:29 AM, Vitaly V. Chaban wrote:
>>
>>> On Thu, Mar 25, 2010 at 12:45 AM, XAvier Periole  
>>> <x.periole at rug.nl> wrote:
>>>>
>>>> On Mar 24, 2010, at 10:58 PM, Vitaly V. Chaban wrote:
>>>>
>>>>>> Hi,
>>>>>>
>>>>>> I want to test different values of cutoff for vdw. However, in  
>>>>>> the list
>>>>>> someone said cutoff = 0.9 for vdw was used for paramaterizing of
>>>>>>> Gromos
>>>>>> force field, other values would cause error, and also there was  
>>>>>> someone
>>>>>> saying that cutoff for vdw should be set to 1.4. Can >someone  
>>>>>> make this
>>>>>> question clear? Thanks a lot.
>>>>>
>>>>> The common rule for all FFs is to set the LJ-cutoff around
>>>>> 2.5*sigma(LJ12-6). In fact, you'll get about 0.9 nm for the most
>>>>> objects. I see nothing terrible if you use, say, 1.4 nm just  
>>>>> because
>>>>> the difference will be negligible.
>>>>
>>>> Where is that coming from? This is so wrong!
>>>
>>>> From the properties of LJ12-6 curve.
>>>
>>>>> Just test it. :)
>>>>
>>>> I've you ever tried that on a lipid bilayer?
>>>
>>> My interests are a bit shifted from bilayers. What concrete atom  
>>> pair
>>> will cause the problem in your example?
>>>
>>>> Who could ever say (whatever force field) a vdW cutoff of 0.9/1.4  
>>>> nm
>>>> is no difference?
>>>>>
>>>>> --
>>>>> Dr. Vitaly Chaban
>>>>> --
>>
>>
>
>
>
> -- 
> Dr. Vitaly Chaban
> http://chaban.at.ua

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100325/30fbfd84/attachment.html>