Scribd Logo
Upload

Welcome to Scribd!

  • 25 views

    Topic 2: Molecular Dynamics of Lennard-Jones System (Continued)

    Uploaded by

    Muklis Latif
    This document discusses techniques for efficiently simulating molecular dynamics in large systems, specifically: 1) Using a cutoff radius (rc) beyond which interparticle forces are considered negligible, to reduce the scaling from O(N2) to O(N); 2) Maintaining neighbor lists of particle pairs within a maximum separation (rmax) that are updated periodically, to track pairs within the cutoff radius; 3) Dividing the system into cells smaller than the cutoff radius and only calculating forces between particles in neighboring cells, also achieving O(N) scaling.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Topic 2: Molecular Dynamics of Lennard-Jones System (Continued)

    Uploaded by

    Muklis Latif
    25 views3 pages
    This document discusses techniques for efficiently simulating molecular dynamics in large systems, specifically: 1) Using a cutoff radius (rc) beyond which interparticle forces are considered negligible, to reduce the scaling from O(N2) to O(N); 2) Maintaining neighbor lists of particle pairs within a maximum separation (rmax) that are updated periodically, to track pairs within the cutoff radius; 3) Dividing the system into cells smaller than the cutoff radius and only calculating forces between particles in neighboring cells, also achieving O(N) scaling.

    Original Description:

    simulasi dinamika molekul

    Original Title

    lec9

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses techniques for efficiently simulating molecular dynamics in large systems, specifically: 1) Using a cutoff radius (rc) beyond which interparticle forces are considered negligible, to reduce the scaling from O(N2) to O(N); 2) Maintaining neighbor lists of particle pairs within a maximum separation (rmax) that are updated periodically, to track pairs within the cutoff radius; 3) Dividing the system into cells smaller than the cutoff radius and only calculating forces between particles in neighboring cells, also achieving O(N) scaling.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    25 views3 pages

    Topic 2: Molecular Dynamics of Lennard-Jones System (Continued)

    Uploaded by

    Muklis Latif
    This document discusses techniques for efficiently simulating molecular dynamics in large systems, specifically: 1) Using a cutoff radius (rc) beyond which interparticle forces are considered negligible, to reduce the scaling from O(N2) to O(N); 2) Maintaining neighbor lists of particle pairs within a maximum separation (rmax) that are updated periodically, to track pairs within the cutoff radius; 3) Dividing the system into cells smaller than the cutoff radius and only calculating forces between particles in neighboring cells, also achieving O(N) scaling.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 3

    High Performance Computing II

    Lecture 9

    Topic 2: Molecular Dynamics of Lennard-Jones System (continued)


    Many molecular dynamics programs use spatial decomposition, a finite range cutoff,
    and neighbor lists. Dr. Nakanos program pmd.c and pmd.h is a nice implementation
    of these important concepts. He has a writeup describing this program.

    Finite Range Force Cutoff


    The evaluation of forces scales like O(N 2) if all pairs of particles are taken into account.
    However, the Lennard-Jones potential energy function
     
    
    12  6
    U (r) = 4

    r
    r
    falls off fairly rapidly with increasing r, so the forces between particles that are very
    far apart may be negligible.
    A reasonable approximation is to cut off the force for r > rc where rc is some suitably chosen cutoff radius. Unfortunately, doing this violates conservation of energy:
    because Newtons equations are discretized, the energy of a pair changes by a constant amount U (rc) rather than a quantity proportional to the time step h when the
    separation crosses the cutoff length. This can be corrected by shifting the potential
    function
    
    U (r) U (rc), for r rc
    Uc(r) =
    0,
    for r > rc
    Page 1

    February 13, 2002

    High Performance Computing II

    Lecture 9

    While this shifted potential is continuous at r = rc, the force, which has strength
    dUc/dr is discontinuous at r = rc. This discontinuity can cause instabilities in the
    integration algorithm. A simple remedy for this problem is to modify the potential
    further
    

    dU
    Uc(r) = U (r) U (rc) dr r=rc (r rc), for r rc
    0,
    for r > rc
    so that
    
    dUc
    F (r) F (rc), for r rc
    Fc(r) =
    =
    0,
    for r > rc
    dr
    is continuous at r = rc.
    Since the potential and force have now been changed, observables such as the pressure
    must also be corrected. For example, it can be shown that
    Z
    4N 2
    hU i = hUci +
    r drU (r)g(r) ,
    V
    rc
    where g(r) is the radial distribution function, which can be approximated in the integral
    by g(r) = 1. Similarly, the virial formula for the pressure can be corrected using
    *N
    +
    Z
    X
    1
    N
    dU (r)
    P V = N kB T
    ri Fci
    r 3dr
    g(r) .
    3 i=1
    3V rc
    dr
    Note that the averages in these formulas involve pairs with r rc and calculating the
    averages scales like O(N ).
    Page 2

    February 13, 2002

    High Performance Computing II

    Lecture 9

    Neighbor Lists and Cell Structures


    If only pairs with a separation r rc are taken into account, the total number of pairs
    scales like O(N ). However there is still a problem! How do we decided which pairs
    have separation r rc without actually measuring r for all pairs? Two useful ways of
    solving this problem are:
    Verlet Neighbor List: A list of all particle pairs with separation rmax > rc is maintained and updated every say 10 or 20 time steps. rmax is chosen large enough
    that it is unlikely that a particle pair not in the list will come closer than rc before
    the list is updated.
    It is also possible to decide automatically when the neighbor list needs to be
    updated. When the list is created, a vector for each particle is set to zero. At
    each time step, the vector is incremented by the particle displacement. The list
    is updated when the sum of the magnitudes of the two largest vectors exceeds
    rmax rc.
    The neighbor list method becomes inefficient for very large N because each update
    requires O(N 2) tests.
    Cell Structures and Linked Lists: The system volume V is divided into cubic cells
    of side ` > rc. At each time step, the particles in each cube are stored in a linked
    list. Particles in a given cell interact only with particles in the 26 neighboring
    cells. For given rc, the number of such pairs scales like O(N ), and so does the
    construction of the linked lists.
    Page 3

    February 13, 2002

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy