From: Steven Mike (stevenm.phys_at_gmail.com)
Date: Thu Aug 29 2013 - 13:24:41 CDT
Hi Giacomo,
Thanks a lot for such a quick response.
I'm performing Umbrella Sampling and will only apply forces to that colvar.
Tough I'm not using ABF right now, it's a great reminder that one needs to
modify the calculation of system forces.
Yet I've searched the src directory, and it seems colvarmodule.C is the
only file that has the specific function to compute the eigenvector colvar.
The part that computes the total force from the system using inverse atomic
gradient looks OK.
with best regards,
Steven
On Thu, Aug 29, 2013 at 9:00 AM, Giacomo Fiorin <giacomo.fiorin_at_gmail.com>wrote:
> Hello Steven, the modification you're planning looks OK.
>
> Regarding the side effects, what method are you planning on using later?
> If you only apply forces to that variable, it should be OK. However, if I
> remember correctly the calculation of system forces, e.g. such as in ABF,
> should be modified accordingly.
>
> Giacomo
>
>
> On Thu, Aug 29, 2013 at 12:14 AM, Steven Mike <stevenm.phys_at_gmail.com>wrote:
>
>> Dear NAMD users,
>>
>>
>>
>> I would like to have some colvar function similar to the eigenvector
>> colvar within NAMD. By default, eigenvector will guarantee the sum of vi
>> (components of the vector for each atom) equal zero. For cases where the
>> sum is not zero, vi will be changed to make the above condition hold. (That
>> means the direction of vi will also be changed in the same time.)
>>
>>
>>
>> Now I'm using vectors but they are not eigenvectors of a normal mode of
>> the system. The sum of my vi is not zero. I would like to use the exact
>> values I provide in "vector". Is that possible to turn on some option in
>> eigenvector or create a new colvar that does not require the sum is zero?
>>
>> I could, in principle, comment out the following line in
>> "colvarcomp_distances.C":
>>
>> 01031 # cvm::log<http://www.ks.uiuc.edu/Research/namd/doxygen/classcolvarmodule.html#e13>("Centering
>> the provided vector to zero.\n");
>>
>> 01032 # for (size_t i = 0; i < atoms<http://www.ks.uiuc.edu/Research/namd/doxygen/classcolvar_1_1eigenvector.html#p0>.size();
>> i++) {
>>
>> 01033 # eigenvec<http://www.ks.uiuc.edu/Research/namd/doxygen/classcolvar_1_1eigenvector.html#p3>[i]
>> -= eig_center;
>>
>> 01034 # }
>>
>> But i don't know if there're some side effects of doing this.
>>
>>
>>
>>
>>
>> Best,
>>
>> Steven
>>
>
>
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:38 CST