[CP2K-user] Steered Molecular Dynamics

[Marcella Iannuzzi]

Dear Devyesh Rana,

Setting the target  of the constraint imposes that the collective variable 
assumes that value immediately. 
Setting the target growth, one can determine a rate of change of the target 
value along the simulation. 
This method is not optimal to estimate free energy profiles reliably. I 
would rather suggest the thermodynamics integration. 
Anyway CP2K just runs the MD and if required can impose restraints or 
constraints. How you extract free energy surfaces from the data of the MD 
simulations is something you need to learn from the literature.  There are 
books and review articles that can help you. 
Kind regards
Marcella

On Monday, November 2, 2020 at 3:30:19 PM UTC+1 Dev Rana wrote:

> Hi Marcella,
>
> Thank you for your suggestions. Is there an example of the 3rd method 
> anywhere? I *think *this is the technique that I actually need. I have 
> some bond lengths (RDF) from ab initio MD between different species so I'd 
> like to place two atoms within the solvent. Subsequently, constraint 1 
> atom, and push the other atom toward a target distance (based on RDF). 
> However, I'm not sure how to do this correctly. The closest example I have 
> found is https://www.cp2k.org/exercises:2020_uzh_acpc2:ex03?s[]=target 
> and 
> https://www.cp2k.org/exercises:2017_ethz_mmm:nacl_free_energy?s[]=target. 
> However, they don't explain what CP2K is doing when you specify a target. 
> The first example seems to do what I think is the proper technique in which 
> initial coordinates differ from the TARGET setpoint. I'll play around with 
> this example. The 2nd example explicitly sets the distance of the starting 
> atom COORD the same as the TARGET, i.e. TARGET = ATOM COORD. So even though 
> this 2nd example is MD, the atom is at the target distance from the 
> beginning of the simulation. 
>
> So my question is for the 1st example, does this method, for example, 
> allow you to place two atoms 10A away and set TARGET to 1.72A? Will MD then 
> forcibly through some sort of steepest descent push these two atoms towards 
> 1.72A? Is it this simple?
>
> Thank you again!
>
> Best Regards,
> Devyesh Rana
>
> On Monday, November 2, 2020 at 3:23:24 AM UTC-5, Marcella Iannuzzi wrote:
>>
>>
>> Dear Devyesh Rana
>>
>> You can run a series of MD with restraints, i.e., following an Umbrella 
>> Sampling procedure.
>> You can run a series of MD with constraints, sampling the corresponding 
>> Lagrangian multiplier, i.e., following a Thermodynamic Integration 
>> procedure. 
>> You can run MD with constraints, specifying  a target, a target_growth, 
>> and a target_limit to steer the change along the MD.
>>
>> Kind regards
>> Marcella
>>
>>
>> On Friday, October 30, 2020 at 9:06:04 PM UTC+1 Dev Rana wrote:
>>
>>> Is there an easy way of preconfiguring a system for steered molecular 
>>> dynamics?
>>>
>>> Specifically, I'd like to understand a reaction in a solvent via AIMD. 
>>> However, instead of waiting for spontaneous reactions to occur, I'd like to 
>>> forcibly make particles interact. Using the potential of mean force to 
>>> obtain transition states etc. Is there an easy way to constrain atoms and 
>>> automatically have them incrementally move towards each other?
>>>
>>> I think the nudged elastic band method might help, but this may not be 
>>> the best avenue. 
>>>
>>> Any suggestions?
>>>
>>> Best Regards,
>>> Devyesh Rana
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20201102/e5f4e97d/attachment.htm>