Re: Maximum Timestep for NAMD

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Thu Feb 07 2019 - 15:57:46 CST

OK, thanks for clarifying. You could definitely separate out the MTS
schemes for bonds, angles and dihedrals. But you won't gain much, because
once you go in the ~5 fs range angles and dihedrals start counting as
well. Note that all this is assuming a canonical atomistic force field for
proteins and other organic molecules.

If you have a system where bonds are clearly separated from the rest, you
may want to try more flexible MD engines (e.g. LAMMPS), where the RESPA
levels can be defined for each force field term. NAMD is optimized really
for one type of physical models.

Giacomo

On Thu, Feb 7, 2019 at 4:47 PM Braden Kelly <bkelly08_at_uoguelph.ca> wrote:

> I don't believe so. That is for long range interactions. I mean evaluate
> all bonds at a very small time step, doing 10 to 50 timesteps based only on
> the evaluation of bond forces. Then do a long time step for lj and ewald
> and dihedral forces. Hopefully a very long time step. This is the really
> time consuming part so make it big...
>
> Braden
>
> Get Outlook for Android <https://aka.ms/ghei36>
>
> ------------------------------
> *From:* Giacomo Fiorin <giacomo.fiorin_at_gmail.com>
> *Sent:* Thursday, February 7, 2019 4:37:58 PM
> *To:* NAMD list; Braden Kelly
> *Subject:* Re: namd-l: Maximum Timestep for NAMD
>
> Is this what you mean?
>
> https://www.ks.uiuc.edu/Research/namd/2.13/ug/node35.html#SECTION000103400000000000000
>
> On Thu, Feb 7, 2019 at 4:36 PM Braden Kelly <bkelly08_at_uoguelph.ca> wrote:
>
>> Has any thought been put into using multiple time step on the bonds
>> rather than SHAKE?
>>
>> Braden
>>
>> Get Outlook for Android
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__aka.ms_ghei36&d=DwMFAg&c=OCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag&r=jUfnSyKkfkyVRBIUzlG1GSGGZAZGcznwr8YliSSCjPc&m=17kz1r5X0wCPBXbEw2XjffafwQDL11Em5F24AhnAHvU&s=qgRw_xV6uAxg-cgPrUui0b-pLSK6SaB_CONOO76uTQ0&e=>
>>
>> ------------------------------
>> *From:* owner-namd-l_at_ks.uiuc.edu <owner-namd-l_at_ks.uiuc.edu> on behalf of
>> Vermaas, Joshua <Joshua.Vermaas_at_nrel.gov>
>> *Sent:* Thursday, February 7, 2019 4:02:20 PM
>> *To:* namd-l_at_ks.uiuc.edu; alexsaadfalcon_at_gmail.com; bennion1_at_llnl.gov
>> *Subject:* RE: namd-l: Maximum Timestep for NAMD
>>
>> Yup. The 1-2 fs for atomic scale simulation comes from the fastest
>> vibrational modes in the system. Bonds to hydrogen have oscillatory periods
>> on the order of 10fs, with heavy atom bonds having periods that are only
>> somewhat larger. In order to have your system not routinely blow itself
>> apart, you need to keep your timestep somewhere between 8-10 times smaller
>> than the vibrational periods in your system. This limits classical atomic
>> simulations to 1fs. Unless you fix bond-lengths to hydrogen with
>> SHAKE/RATTLE/LINCS, which will get you to 2fs. Or you can also restrain
>> heavy-atom bond lengths and some angles, and get up to 4fs with virtual
>> sites in gromacs. Coarse grained systems allow for larger timesteps because
>> the particle masses are larger, and 20 fs timesteps aren't unheard of.
>> Basically, if we could take 100fs timesteps without a loss in accuracy, we
>> would already be doing it routinely. :)
>>
>> -Josh
>>
>>
>>
>> On 2019-02-07 13:49:27-07:00 owner-namd-l_at_ks.uiuc.edu wrote:
>>
>> Hello,
>>
>>
>>
>> The first thing to consider is why the 1-2 fs timestep is most often
>> used.
>>
>> You might also consider various accelerated MD methods and coarse grained
>> MD.
>>
>>
>>
>> Brian
>>
>>
>>
>>
>>
>> *From:* owner-namd-l_at_ks.uiuc.edu <owner-namd-l_at_ks.uiuc.edu> *On Behalf
>> Of *Alex Saad-Falcon
>> *Sent:* Thursday, February 7, 2019 12:25 PM
>> *To:* namd-l_at_ks.uiuc.edu
>> *Subject:* namd-l: Maximum Timestep for NAMD
>>
>>
>>
>> Hi all,
>>
>>
>>
>> What is the maximum acceptable timestep for NAMD? I would like to run
>> longer (1ns to 1ms) simulations, and I am wondering how much I can increase
>> the timestep (to 10fs or 100fs) without detracting terribly from accuracy.
>> What do I need to consider?
>>
>>
>>
>> Thanks,
>>
>>
>>
>> Alex
>>
>>
>
> --
> Giacomo Fiorin
> Associate Professor of Research, Temple University, Philadelphia, PA
> Contractor, National Institutes of Health, Bethesda, MD
> http://goo.gl/Q3TBQU
> https://github.com/giacomofiorin
>

-- 
Giacomo Fiorin
Associate Professor of Research, Temple University, Philadelphia, PA
Contractor, National Institutes of Health, Bethesda, MD
http://goo.gl/Q3TBQU
https://github.com/giacomofiorin

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:10 CST