VMD-L Mailing List

From: Laura Joana (laurajoanalopes_at_gmail.com)
Date: Tue Apr 19 2016 - 04:17:09 CDT

Hello Giacomo,

That's true, it's always going to be differences. But, calculating the
dihedral angles with this coordinates they give the same values, with
slightly differences, evidently. But as I calculate then with VMD they are
different. That's the problem!
Best,

Laura

2016-04-19 0:20 GMT+02:00 Giacomo Fiorin <giacomo.fiorin_at_gmail.com>:

> Hi, VMD reads all files and stores them in memory in single precision,
> thus atomic coordinates cannot be compared to their NAMD counterparts with
> full digits.
>
> Giacomo
>
> On Mon, Apr 18, 2016 at 10:05 AM, Laura Joana <laurajoanalopes_at_gmail.com>
> wrote:
>
>> I obtain close values, but they are not exactly equal:
>> NAMD
>> -7.75349182128906e+02 -7.74708642578125e+03 2.05269702148438e+03
>> -7.74688653758182e+02 -7.74826807202096e+03 2.05239478592227e+03
>> -7.74603393652299e+02 -7.74890428783011e+03 2.05105687624320e+03
>> -7.75804016113281e+02 -7.74978466796875e+03 2.05070825195312e+03
>> -7.77042909993649e+02 -7.74924461376854e+03 2.05071753160505e+03
>>
>> VMD
>> {-775.3491821289063 -7747.08642578125 2052.697021484375}
>> {-774.6886596679688 -7748.26806640625 2052.394775390625}
>> {-774.6033935546875 -7748.904296875 2051.056884765625}
>> {-775.8040161132813 -7749.78466796875 2050.708251953125}
>> {-777.0429077148438 -7749.24462890625 2050.717529296875}
>>
>> But it's strange to thing that this slightly difference will cause such a
>> big difference in the dihedral values... Do you think the problem is here?
>>
>> 2016-04-18 15:40 GMT+02:00 Jérôme Hénin <jerome.henin_at_ibpc.fr>:
>>
>>> Can you add the following colvar to your config, and then check its
>>> value against the Cartesian coordinates of those atoms (you can get those
>>> in VMD):
>>>
>>> colvar {
>>> name xyz
>>> cartesian {
>>> atoms {atomNumbers 5 7 9 15 17}
>>> }
>>> }
>>>
>>> Jerome
>>>
>>> On 18 April 2016 at 15:31, Laura Joana <laurajoanalopes_at_gmail.com>
>>> wrote:
>>>
>>>> Jerome,
>>>> Does it help if I send you the output I'm obtaining with NAMD?
>>>> Best,
>>>> Laura
>>>>
>>>> 2016-04-18 15:24 GMT+02:00 Laura Joana <laurajoanalopes_at_gmail.com>:
>>>>
>>>>> I just got the compiled version (linux x86_64) and I got the same
>>>>> wrong values as before. I definitely cannot understand why I'm having this
>>>>> error! It makes no sense....
>>>>>
>>>>> 2016-04-18 14:53 GMT+02:00 Laura Joana <laurajoanalopes_at_gmail.com>:
>>>>>
>>>>>> Actually I can try to use the compiled version for linux, the same as
>>>>>> you, and see if my problem goes away...
>>>>>>
>>>>>> 2016-04-18 14:52 GMT+02:00 Laura Joana <laurajoanalopes_at_gmail.com>:
>>>>>>
>>>>>>> I'm using the same version as you, NAMD 2.11 compiled from source in
>>>>>>> a linux system, ethernet version! I cannot use the development version
>>>>>>> because I got some problems with rattle in previous simulations....
>>>>>>> Now this is so strange! Certainly there is something different in my
>>>>>>> configuration that gives me that problem.... Any clue on how to find out
>>>>>>> what?
>>>>>>>
>>>>>>> 2016-04-18 14:40 GMT+02:00 Jérôme Hénin <jerome.henin_at_ibpc.fr>:
>>>>>>>
>>>>>>>> Yes, the VMD values are correct. I get the very same values with
>>>>>>>> NAMD 2.11 and the current development version in Linux x86_64. What version
>>>>>>>> and system are you using?
>>>>>>>>
>>>>>>>> Jerome
>>>>>>>>
>>>>>>>> On 18 April 2016 at 14:36, Laura Joana <laurajoanalopes_at_gmail.com>
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>>> Hi Jerome,
>>>>>>>>>
>>>>>>>>> Both commands gave me the same values and it seems the values I'm
>>>>>>>>> obtaining in VMD are correct because I'm also calculation then with measure
>>>>>>>>> and they are all the same! Now the problem is with NAMD...
>>>>>>>>> Are you obtaining -8.10195434545963e+01 and 7.02575414592844e+01
>>>>>>>>> when running the file init.conf that I sent you? What version of NAMD are
>>>>>>>>> you using?
>>>>>>>>> Best,
>>>>>>>>>
>>>>>>>>> Laura
>>>>>>>>>
>>>>>>>>> 2016-04-18 14:20 GMT+02:00 Jérôme Hénin <jerome.henin_at_ibpc.fr>:
>>>>>>>>>
>>>>>>>>>> Hi Laura,
>>>>>>>>>>
>>>>>>>>>> Not sure what is happening, I cannot reproduce this problem. Can
>>>>>>>>>> you compare with large values of the box sizes? eg.
>>>>>>>>>>
>>>>>>>>>> molinfo top set {a b c} {0 0 0}; cv update; cv printframe
>>>>>>>>>> molinfo top set {a b c} {100 100 100}; cv update; cv printframe
>>>>>>>>>>
>>>>>>>>>> Best,
>>>>>>>>>> Jerome
>>>>>>>>>>
>>>>>>>>>> On 16 April 2016 at 14:34, Laura Joana <laurajoanalopes_at_gmail.com
>>>>>>>>>> > wrote:
>>>>>>>>>>
>>>>>>>>>>> Hello Jerome!
>>>>>>>>>>>
>>>>>>>>>>> I tried to use the command line you gave me, but I still have
>>>>>>>>>>> different values for VMD and NAMD. With VMD I get:
>>>>>>>>>>> colvar -8.10195434545963e+01 7.02575414592844e+01
>>>>>>>>>>> and with NAMD I get:
>>>>>>>>>>> colvar -8.42715396480484e+01 5.74396799966552e+01
>>>>>>>>>>> Do you have any clue? And did you obtained the same values with
>>>>>>>>>>> the files I sent you? Because if you did it may be a problem with the
>>>>>>>>>>> versions of VMD and/or NAMD I'm using.
>>>>>>>>>>> Best,
>>>>>>>>>>>
>>>>>>>>>>> Laura
>>>>>>>>>>>
>>>>>>>>>>> 2016-04-16 12:16 GMT+02:00 Laura Joana <
>>>>>>>>>>> laurajoanalopes_at_gmail.com>:
>>>>>>>>>>>
>>>>>>>>>>>> Thanks Jerome! If that solves my problem I'll always going to
>>>>>>>>>>>> include that command line in my script when no periodic box information is
>>>>>>>>>>>> given.
>>>>>>>>>>>> Best,
>>>>>>>>>>>> Laura
>>>>>>>>>>>>
>>>>>>>>>>>> 2016-04-15 19:22 GMT+02:00 John Stone <johns_at_ks.uiuc.edu>:
>>>>>>>>>>>>
>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>> A very minor correction. Current VMD versions do set the
>>>>>>>>>>>>> periodic cell
>>>>>>>>>>>>> sizes to zero by default. That being said, there likely may
>>>>>>>>>>>>> still exist a
>>>>>>>>>>>>> molfile plugin or two that may be setting the periodic cell
>>>>>>>>>>>>> size to
>>>>>>>>>>>>> a non-zero number by default.
>>>>>>>>>>>>> Another possibility is that you may have loaded a file that
>>>>>>>>>>>>> had been processed
>>>>>>>>>>>>> by an old version of VMD that indeed did exactly what Jerome
>>>>>>>>>>>>> is warning about.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Cheers,
>>>>>>>>>>>>> John
>>>>>>>>>>>>>
>>>>>>>>>>>>> On Fri, Apr 15, 2016 at 06:50:32PM +0200, Jérôme Hénin wrote:
>>>>>>>>>>>>> > Hi Laura,
>>>>>>>>>>>>> > The problem comes from a very unfortunate default
>>>>>>>>>>>>> behavior of VMD: when no
>>>>>>>>>>>>> > periodic box information is provided in a DCD file, it
>>>>>>>>>>>>> sets the box sizes
>>>>>>>>>>>>> > to 1, which breaks PBC-aware codes like colvars.
>>>>>>>>>>>>> > You can set them to zero with
>>>>>>>>>>>>> > molinfo top set {a b c} {0 0 0}
>>>>>>>>>>>>> > then the dihedral values are correct.
>>>>>>>>>>>>> > Best,
>>>>>>>>>>>>> > Jerome
>>>>>>>>>>>>> > On 15 April 2016 at 16:50, Karteek Kumar <[1]
>>>>>>>>>>>>> karthik3327_at_gmail.com> wrote:
>>>>>>>>>>>>> >
>>>>>>>>>>>>> > Hi,
>>>>>>>>>>>>> > The dihedral value of same four atoms should be equal
>>>>>>>>>>>>> either you
>>>>>>>>>>>>> > calculate using colvars or VMD.
>>>>>>>>>>>>> > How much different the values are?
>>>>>>>>>>>>> > I am afraid, A if you are measuring the dihedral of
>>>>>>>>>>>>> different atoms.A
>>>>>>>>>>>>> > Can you double check the atoms?A
>>>>>>>>>>>>> > Remember, the colvarsA atom numbers start with 1.
>>>>>>>>>>>>> > In VMD, A serial starts with 1 and index starts with 0--001a1147b670d8d0dc0530d2f2c4--