normal nvt step in CG simulation

  • Hongtham
  • Hongtham's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
3 years 6 months ago #5243 by Hongtham
nvt step in CG simulation was created by Hongtham
Hi everyone,
I have a question that I am not clear. When I run all-atom simulation, i usually heat the system under NVT ensemble (constant Number of particles, Volume, and Temperature) and then turn on the NPT ensemble in equilibration step. But in CG tutorial, the equilibration step (equilibration.mdp) includes heating system right away after energy minimization step. I wonder, why there is this difference in CG simulation. I hope some one can explain to me.
Thank you so much.
Best regards,
Hongtham

Please Log in or Create an account to join the conversation.

More
3 years 6 months ago #5251 by helgi
Replied by helgi on topic nvt step in CG simulation
Hello, hello,

The very short answer is that the CG Martini energy landscape is much smoother, so you don’t have to carefully massage the system into a correct starting configuration it will get there much easier.

Hope this helps,
Cheers,
- Helgi

Please Log in or Create an account to join the conversation.

  • Hongtham
  • Hongtham's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
3 years 6 months ago #5254 by Hongtham
Replied by Hongtham on topic nvt step in CG simulation
Hello,
Sorry, I don't get your meaning. Can you explain more about the smoother energy landscape and the "carefully massage" the system. I read again the Martini forcefield papers but I couldn't find or understand these terms.
Thank you for your kindness.
Hongtham

Please Log in or Create an account to join the conversation.

  • peterkroon
  • peterkroon's Avatar
  • Away
  • Gold Boarder
More
3 years 6 months ago #5255 by peterkroon
Replied by peterkroon on topic nvt step in CG simulation
Hi Hongtham,

Martini systems are much less likely to blow up/crash than atomistic system, so you don't have to treat them as careful.
Effectively this means you can do your equilibration starting further from equilibrium.

The smoother energy landscape means that all forces acting on beads are smaller, and blowing up the system due to an numerical instability is less likely. "Carefully massaging" a system means that you start far from equilibrium, and then very gently equilibrate it, by gradually increasing the timestep, or starting with NVT conditions, which are closer to your starting point.

HTH
Peter

Please Log in or Create an account to join the conversation.

  • Hongtham
  • Hongtham's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
3 years 6 months ago #5258 by Hongtham
Replied by Hongtham on topic nvt step in CG simulation
hi Peter,
now I understand. Thank you for your explanation. :D

Hongtham

Please Log in or Create an account to join the conversation.

More
7 months 4 days ago #7852 by riccardo
Replied by riccardo on topic nvt step in CG simulation
Martini polymers do not require any special mdp parameters (unless clearly stated in the publication where the particular polymer you are interested in was developed, e.g., I think the PEO model of 2009 required a time step lower than 20 fs - but it is an outdated model anyway..).
For production run - i.e., your MD.mdp - you can use one of the files you can find at:

cgmartini.nl/images/parameters/exampleMDP

It can be useful to check also the associated publication. For example, you could use the "new" set of mdp parameters:

cgmartini.nl/images/parameters/exampleMDP/martini_v2.x_new.mdp

as done in the latest Martini (polymers and not only) papers. Just pay attention that you may not need the semiisotropic pressure coupling but just a normal isotropic coupling (if you system is isotropic, e.g., one polymer chain in solution).

For the equilibrations, you can use Berendsen pressure and temperature couplings, as they are more robust (this is especially true for the pressure coupling, see also the note in the default MDP files). For different temperature, just adjust the "ref_t" parameter accordingly.

Please Log in or Create an account to join the conversation.

  • peterkroon
  • peterkroon's Avatar
  • Away
  • Gold Boarder
More
7 months 4 days ago #7855 by peterkroon
Replied by peterkroon on topic nvt step in CG simulation
In addition to what Riccardo said, please bear in mind that Martini was parametrized for ~300K. Results you obtain at 600K may not be physically realistic so validate carefully.

Please Log in or Create an account to join the conversation.

  • peterkroon
  • peterkroon's Avatar
  • Away
  • Gold Boarder
More
7 months 3 days ago #7862 by peterkroon
Replied by peterkroon on topic nvt step in CG simulation
In that case you're not really interested in the structure at 600K, so all should be fine :)

You may need to reduce the timestep at 600K though (try it as-is first), and don't expect phase behaviour to be correct.

Please Log in or Create an account to join the conversation.

Time to create page: 0.094 seconds