/* ----------------------------------------------------------------------
   LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator
   http://lammps.sandia.gov, Sandia National Laboratories
   Steve Plimpton, sjplimp@sandia.gov

   Copyright (2003) Sandia Corporation.  Under the terms of Contract
   DE-AC04-94AL85000 with Sandia Corporation, the U.S. Government retains
   certain rights in this software.  This software is distributed under
   the GNU General Public License.

   See the README file in the top-level LAMMPS directory.
------------------------------------------------------------------------- */

/* ----------------------------------------------------------------------
   Contributing author: Axel Kohlmeyer (Temple U)
------------------------------------------------------------------------- */

#include "angle_quartic_omp.h"
#include "atom.h"
#include "comm.h"
#include "force.h"
#include "neighbor.h"
#include "domain.h"

#include "math_const.h"

#include <math.h>

#include "suffix.h"
using namespace LAMMPS_NS;
using namespace MathConst;

#define SMALL 0.001

/* ---------------------------------------------------------------------- */

AngleQuarticOMP::AngleQuarticOMP(class LAMMPS *lmp)
  : AngleQuartic(lmp), ThrOMP(lmp,THR_ANGLE)
{
  suffix_flag |= Suffix::OMP;
}

/* ---------------------------------------------------------------------- */

void AngleQuarticOMP::compute(int eflag, int vflag)
{

  if (eflag || vflag) {
    ev_setup(eflag,vflag);
  } else evflag = 0;

  const int nall = atom->nlocal + atom->nghost;
  const int nthreads = comm->nthreads;
  const int inum = neighbor->nanglelist;

#if defined(_OPENMP)
#pragma omp parallel default(none) shared(eflag,vflag)
#endif
  {
    int ifrom, ito, tid;

    loop_setup_thr(ifrom, ito, tid, inum, nthreads);
    ThrData *thr = fix->get_thr(tid);
    ev_setup_thr(eflag, vflag, nall, eatom, vatom, thr);

    if (evflag) {
      if (eflag) {
        if (force->newton_bond) eval<1,1,1>(ifrom, ito, thr);
        else eval<1,1,0>(ifrom, ito, thr);
      } else {
        if (force->newton_bond) eval<1,0,1>(ifrom, ito, thr);
        else eval<1,0,0>(ifrom, ito, thr);
      }
    } else {
      if (force->newton_bond) eval<0,0,1>(ifrom, ito, thr);
      else eval<0,0,0>(ifrom, ito, thr);
    }

    reduce_thr(this, eflag, vflag, thr);
  } // end of omp parallel region
}

template <int EVFLAG, int EFLAG, int NEWTON_BOND>
void AngleQuarticOMP::eval(int nfrom, int nto, ThrData * const thr)
{
  int i1,i2,i3,n,type;
  double delx1,dely1,delz1,delx2,dely2,delz2;
  double eangle,f1[3],f3[3];
  double dtheta,dtheta2,dtheta3,dtheta4,tk;
  double rsq1,rsq2,r1,r2,c,s,a,a11,a12,a22;

  const double * const * const x = atom->x;
  double * const * const f = thr->get_f();
  const int * const * const anglelist = neighbor->anglelist;
  const int nlocal = atom->nlocal;

  for (n = nfrom; n < nto; n++) {
    i1 = anglelist[n][0];
    i2 = anglelist[n][1];
    i3 = anglelist[n][2];
    type = anglelist[n][3];

    // 1st bond

    delx1 = x[i1][0] - x[i2][0];
    dely1 = x[i1][1] - x[i2][1];
    delz1 = x[i1][2] - x[i2][2];

    rsq1 = delx1*delx1 + dely1*dely1 + delz1*delz1;
    r1 = sqrt(rsq1);

    // 2nd bond

    delx2 = x[i3][0] - x[i2][0];
    dely2 = x[i3][1] - x[i2][1];
    delz2 = x[i3][2] - x[i2][2];

    rsq2 = delx2*delx2 + dely2*dely2 + delz2*delz2;
    r2 = sqrt(rsq2);

    // angle (cos and sin)

    c = delx1*delx2 + dely1*dely2 + delz1*delz2;
    c /= r1*r2;

    if (c > 1.0) c = 1.0;
    if (c < -1.0) c = -1.0;

    s = sqrt(1.0 - c*c);
    if (s < SMALL) s = SMALL;
    s = 1.0/s;

    // force & energy

    dtheta = acos(c) - theta0[type];
    dtheta2 = dtheta * dtheta;
    dtheta3 = dtheta2 * dtheta;
    tk =  2.0 * k2[type] * dtheta + 3.0 * k3[type] * dtheta2 + 4.0 * k4[type] * dtheta3;

    if (EFLAG) {
      dtheta4 = dtheta3 * dtheta;
      eangle = k2[type] * dtheta2 + k3[type] * dtheta3 + k4[type] * dtheta4;
    }

    a = -2.0 * tk * s;
    a11 = a*c / rsq1;
    a12 = -a / (r1*r2);
    a22 = a*c / rsq2;

    f1[0] = a11*delx1 + a12*delx2;
    f1[1] = a11*dely1 + a12*dely2;
    f1[2] = a11*delz1 + a12*delz2;
    f3[0] = a22*delx2 + a12*delx1;
    f3[1] = a22*dely2 + a12*dely1;
    f3[2] = a22*delz2 + a12*delz1;

    // apply force to each of 3 atoms

    if (NEWTON_BOND || i1 < nlocal) {
      f[i1][0] += f1[0];
      f[i1][1] += f1[1];
      f[i1][2] += f1[2];
    }

    if (NEWTON_BOND || i2 < nlocal) {
      f[i2][0] -= f1[0] + f3[0];
      f[i2][1] -= f1[1] + f3[1];
      f[i2][2] -= f1[2] + f3[2];
    }

    if (NEWTON_BOND || i3 < nlocal) {
      f[i3][0] += f3[0];
      f[i3][1] += f3[1];
      f[i3][2] += f3[2];
    }

    if (EVFLAG) ev_tally_thr(this,i1,i2,i3,nlocal,NEWTON_BOND,eangle,f1,f3,
                             delx1,dely1,delz1,delx2,dely2,delz2,thr);
  }
}