Dear LAMMPS users,

I want calculated the chain end-to-end unit cell vector autocorrelation function <u(t)*u(0)>. I am wondering can I use the ** colvar** module to calculate it?

Kind regards,

MC

Dear LAMMPS users,

I want calculated the chain end-to-end unit cell vector autocorrelation function <u(t)*u(0)>. I am wondering can I use the ** colvar** module to calculate it?

Kind regards,

MC

Dear LAMMPS users,

I want calculated the chain end-to-end unit cell vector autocorrelation

function <u(t)*u(0)>. I am wondering can I use the *colvar* module to

calculate it?

http://lammps.sandia.gov/doc/PDF/colvars-refman-lammps.pdf

Mingchao, the syntax of defining an ACF of a variable of any kind is as explained in the link that Axel sent.

One issue that I see, however, is that currently the unit cell vectors are not available as a collective variable, unfortunately.

If you can define the vectors of interest as a function of certain atomic coordinates, it’s possible to do what you suggested.

Giacomo

I was having trouble parsing the requested quantity:

chain end-to-end unit cell vector autocorrelation function <u(t)*u(0)>

which seems to combine two unrelated concepts. But I see from the subject line that it should be:

chain end-to-end unit vector autocorrelation function <u(t)*u(0)>

This makes more sense. For chains with fixed bonding, u(t) for each molecule can be calculated using explicit variable formulas for each chain:

variable rfac equal “1.0/sqrt((x[{id0}] - x[{id1}])^2 + (y[[id0}] -
y[{id1}])^2 + (z[{id0}] - z[{id1}])^2)”

variable ux equal (x[{id0}] - x[{id1}])*rfac

variable uy equal (y[{id0}] - y[{id1}])*rfac

variable uz equal (z[{id0}] - z[{id1}])*rfac

This is no fun for lots of chains.

If you search the LAMMPS list archive you will find this comment from Steve in 2013:

“Doing it for lots of chains is not currently possible,

and wouldn’t be simple to implement, b/c a processor

that owns one end of the polymer will typically not know

anything about the atom at the other end.”

What has changed since then is fix chunk, which allows you to get things like center-of-mass vector of a molecule:

http://lammps.sandia.gov/doc/compute_com_chunk.html

It would not be hard (for someone) to modify that source code to do end-to-end distance of a 2-atom chunk.

Aidan

Hello Aidan, in that case there is no source code modification needed. Mingchao can define two groups, each made by one atom, and calculate the ACF of the distance vector variable.

Giacomo