Re: Dihedral definition in NAMD

From: Kevin C Chan (
Date: Tue Mar 10 2015 - 22:22:34 CDT

Thanks! A simple "puts" works! So it loses its screen-out function inside a

I found weird things when I look at my outputs. When I feed coor and vel
from a previous simulation (simulation-1) to a simulation-2, for which both
simulations were turned on with the same colvar setting (a harmonic
potential to dihedral value). I supposed the last dihedral value for the
simulation-1 should be the same as the first value for the simulation-2 but
it is not the case.

I checked the logs and they should be reading the correct input coor and
vel files. Using cv in VMD reading the coor file gives the first value of
simulation-2 (again saying that simulation-2 is reading the correct coor
file). And this value deviates a lot from the value I set with the center
for the harmonic. However, inside the colvar.traj of simulation-1, it says
my dihedral value is harmonically restrained to my desired value.

Thanks in advance,

On Wed, Mar 11, 2015 at 7:46 AM, Giacomo Fiorin <>

> Hi Kevin, I think you will probably need:
> puts [cv colvar dih value]
> to see the value printed on the screen. Think of that command between
> brackets as a measure command.
> You would need to recompile VMD if you load in multiple configurations
> files. What you're using there should be fine with 1.9.2
> On Tue, Mar 10, 2015 at 7:13 PM, Kevin C Chan <
> > wrote:
>> Thanks to all of you and your replies. I got vmd1.9.2 and cv works
>> horribly well.
>> ***Except*** that I found my “cv” (shipped with
>> vmd1.9.2-linux-opengl-cuda) just cannot run inside a for-loop:
>> # I have loaded a number of frames and wanna compare their colvar values #
>> cv molid top
>> cv configfile dih.colvar
>> for { set i 0 } { $i <= 9 } { incr i } {
>> cv frame $i
>> cv colvar dih update
>> cv colvar dih value
>> }
>> By "cannot run" it means the vmd did not either give error messages or
>> quit, instead just nothing happens.
>> I noticed that Giacomo had given an update of colvars from Github but it
>> seems I have to build my VMD from the source files especially all those
>> libraries. I am just asking whether the update solves my problem of
>> not-working-in-a-loop. If yes, I will go back and compile it. If no, what
>> is it about actually?
>> Thanks in advance,
>> Kevin
>> On 10 Mar, 2015, at 22:33, Giacomo Fiorin <>
>> wrote:
>> Kevin, you may want to try colvars inside VMD, calculating the variables
>> without starting NAMD.
>> I would recommend that you download and compile the source code of VMD,
>> updating it with the colvars code from Github using the provided
>> script:
>> Giacomo
>> On Tue, Mar 10, 2015 at 3:24 PM, Kevin C Chan <
>>> wrote:
>>> Thanks Josh!
>>> I am also wondering whether we can calculate such dihedral of 4 groups
>>> simply using pdb+psf. Because right now I am doing with colvars turned on
>>> and running few steps of a random MD and look at the colvars.traj and
>>> honestly it is as stupid as it sounds.
>>> Thanks in advance,
>>> Kevin
>>> On 10 Mar, 2015, at 22:14, Josh Vermaas <> wrote:
>>> Hi Kevin,
>>> I believe the way it is written is pretty standard, using the ordering
>>> that folks use to describe dihedral potentials within molecules, and just
>>> generalizing it to groups of atoms. So group1 would be first atom, group2
>>> would be the second, and so on. The "dihedral" angle is measuring rotation
>>> around the group2-group3 "bond" by measuring the angle between group1 and
>>> group4 if you are looking down the group2-group3 bond (Fig. 3 from the
>>> Wikipedia article).
>>> -Josh Vermaas
>>> On 03/10/2015 06:52 AM, Kevin C Chan wrote:
>>> Thank you so much for the hint from "distance".
>>> Concerning the dihedral definition, I am not sure whether I get what you
>>> are saying. In the Wikipedia, under the section "Dihedral angles of four
>>> atoms
>>> ​"
>>> , it uses words like first atom, second atom and so on, so is this the
>>> sequence that NAMD colvars are using for dihedral calculation?
>>> ​Thanks, in advance,
>>> Kevin​
>>> On Tue, Mar 10, 2015 at 7:28 PM, Giacomo Fiorin <
>>>> wrote:
>>>> The center of mass is not mentioned explicitly for all components,
>>>> but these are all consistent with the distance:
>>>> i.e. the groups' centers-of-mass are used to define individual points--047d7bb048764706fd0510fac79f--

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:42 CST