| 1 |
mmeineke |
10 |
subroutine ssd_forces(natoms, i, j, atom1, atom2, dx, dy, dz, rij, pot, r2, & |
| 2 |
|
|
flx, fly, flz, tlx, tly, tlz, a) |
| 3 |
|
|
|
| 4 |
|
|
implicit none |
| 5 |
|
|
|
| 6 |
|
|
include 'f_ssd.inc' |
| 7 |
|
|
|
| 8 |
|
|
! This routine does only the sticky portion of the SSD potential |
| 9 |
|
|
! [Chandra and Ichiye, J. Chem. Phys. 111, 2701 (1999)]. |
| 10 |
|
|
! The Lennard-Jones and dipolar interaction must be handled separately. |
| 11 |
|
|
|
| 12 |
|
|
! We assume that the rotation matrices have already been calculated |
| 13 |
|
|
! and placed in the A(9, max_mol) array. |
| 14 |
|
|
! |
| 15 |
|
|
! i and j are pointers to the two SSD molecules, while atom1 and |
| 16 |
|
|
! atom2 are the pointers to the location of the atoms in the force |
| 17 |
|
|
! and position arrays. These are both necessary since the rotation |
| 18 |
|
|
! matrix is a property of the molecule, while the force is acting on |
| 19 |
|
|
! the atom. The indexing of atoms and molecules is not necessarily |
| 20 |
|
|
! the same in simulations of mixtures. |
| 21 |
|
|
|
| 22 |
|
|
|
| 23 |
|
|
integer :: i, j, atom1, atom2, natoms |
| 24 |
|
|
double precision dx, dy, dz, rij, pot, r2 |
| 25 |
|
|
double precision, dimension(natoms) ::flx, fly, flz, tlx, tly, tlz |
| 26 |
|
|
double precision, dimension(9,natoms):: a |
| 27 |
|
|
|
| 28 |
|
|
|
| 29 |
|
|
double precision xi, yi, zi, xj, yj, zj, xi2, yi2, zi2, xj2, yj2, zj2 |
| 30 |
|
|
double precision r3, r5, r6, s, sp, dsdr, dspdr |
| 31 |
|
|
double precision wi, wj, w, wip, wjp, wp |
| 32 |
|
|
double precision dwidx, dwidy, dwidz, dwjdx, dwjdy, dwjdz |
| 33 |
|
|
double precision dwipdx, dwipdy, dwipdz, dwjpdx, dwjpdy, dwjpdz |
| 34 |
|
|
double precision dwidux, dwiduy, dwiduz, dwjdux, dwjduy, dwjduz |
| 35 |
|
|
double precision dwipdux, dwipduy, dwipduz, dwjpdux, dwjpduy, dwjpduz |
| 36 |
|
|
double precision zif, zis, zjf, zjs, uglyi, uglyj |
| 37 |
|
|
double precision drdx, drdy, drdz |
| 38 |
|
|
double precision txi, tyi, tzi, txj, tyj, tzj |
| 39 |
|
|
double precision fxii, fyii, fzii, fxjj, fyjj, fzjj |
| 40 |
|
|
double precision fxij, fyij, fzij, fxji, fyji, fzji |
| 41 |
|
|
|
| 42 |
|
|
|
| 43 |
|
|
! Use molecular positions, since the SSD model has only one atom, and the |
| 44 |
|
|
! rotation matrix is for the molecule itself: |
| 45 |
|
|
|
| 46 |
|
|
r3 = r2*rij |
| 47 |
|
|
r5 = r3*r2 |
| 48 |
|
|
|
| 49 |
|
|
drdx = dx / rij |
| 50 |
|
|
drdy = dy / rij |
| 51 |
|
|
drdz = dz / rij |
| 52 |
|
|
|
| 53 |
|
|
! rotate the inter-particle separation into the two different |
| 54 |
|
|
! body-fixed coordinate systems: |
| 55 |
|
|
|
| 56 |
|
|
xi = a(1,i)*dx + a(2,i)*dy + a(3,i)*dz |
| 57 |
|
|
yi = a(4,i)*dx + a(5,i)*dy + a(6,i)*dz |
| 58 |
|
|
zi = a(7,i)*dx + a(8,i)*dy + a(9,i)*dz |
| 59 |
|
|
|
| 60 |
|
|
! negative sign because this is the vector from j to i: |
| 61 |
|
|
|
| 62 |
|
|
xj = -(a(1,j)*dx + a(2,j)*dy + a(3,j)*dz) |
| 63 |
|
|
yj = -(a(4,j)*dx + a(5,j)*dy + a(6,j)*dz) |
| 64 |
|
|
zj = -(a(7,j)*dx + a(8,j)*dy + a(9,j)*dz) |
| 65 |
|
|
|
| 66 |
|
|
xi2 = xi*xi |
| 67 |
|
|
yi2 = yi*yi |
| 68 |
|
|
zi2 = zi*zi |
| 69 |
|
|
|
| 70 |
|
|
xj2 = xj*xj |
| 71 |
|
|
yj2 = yj*yj |
| 72 |
|
|
zj2 = zj*zj |
| 73 |
|
|
|
| 74 |
|
|
call calc_s(rij, s, sp, dsdr, dspdr) |
| 75 |
|
|
|
| 76 |
|
|
wi = 2.0d0*(xi2-yi2)*zi / r3 |
| 77 |
|
|
wj = 2.0d0*(xj2-yj2)*zj / r3 |
| 78 |
|
|
w = wi+wj |
| 79 |
|
|
|
| 80 |
|
|
zif = zi/rij - 0.6d0 |
| 81 |
|
|
zis = zi/rij + 0.8d0 |
| 82 |
|
|
|
| 83 |
|
|
zjf = zj/rij - 0.6d0 |
| 84 |
|
|
zjs = zj/rij + 0.8d0 |
| 85 |
|
|
|
| 86 |
|
|
wip = zif*zif*zis*zis - SSD_w0 |
| 87 |
|
|
wjp = zjf*zjf*zjs*zjs - SSD_w0 |
| 88 |
|
|
wp = wip + wjp |
| 89 |
|
|
|
| 90 |
|
|
pot = pot + 0.5d0*SSD_v0*(s*w + sp*wp) |
| 91 |
|
|
|
| 92 |
|
|
dwidx = 4.0d0*xi*zi/r3 - 6.0d0*xi*zi*(xi2-yi2)/r5 |
| 93 |
|
|
dwidy = - 4.0d0*yi*zi/r3 - 6.0d0*yi*zi*(xi2-yi2)/r5 |
| 94 |
|
|
dwidz = 2.0d0*(xi2-yi2)/r3 - 6.0d0*zi2*(xi2-yi2)/r5 |
| 95 |
|
|
|
| 96 |
|
|
dwjdx = 4.0d0*xj*zj/r3 - 6.0d0*xj*zj*(xj2-yj2)/r5 |
| 97 |
|
|
dwjdy = - 4.0d0*yj*zj/r3 - 6.0d0*yj*zj*(xj2-yj2)/r5 |
| 98 |
|
|
dwjdz = 2.0d0*(xj2-yj2)/r3 - 6.0d0*zj2*(xj2-yj2)/r5 |
| 99 |
|
|
|
| 100 |
|
|
uglyi = zif*zif*zis + zif*zis*zis |
| 101 |
|
|
uglyj = zjf*zjf*zjs + zjf*zjs*zjs |
| 102 |
|
|
|
| 103 |
|
|
dwipdx = -2.0d0*xi*zi*uglyi/r3 |
| 104 |
|
|
dwipdy = -2.0d0*yi*zi*uglyi/r3 |
| 105 |
|
|
dwipdz = 2.0d0*(1.0d0/rij - zi2/r3)*uglyi |
| 106 |
|
|
|
| 107 |
|
|
dwjpdx = -2.0d0*xj*zj*uglyj/r3 |
| 108 |
|
|
dwjpdy = -2.0d0*yj*zj*uglyj/r3 |
| 109 |
|
|
dwjpdz = 2.0d0*(1.0d0/rij - zj2/r3)*uglyj |
| 110 |
|
|
|
| 111 |
|
|
dwidux = 4.0d0*(yi*zi2 + 0.5d0*yi*(xi2-yi2))/r3 |
| 112 |
|
|
dwiduy = 4.0d0*(xi*zi2 - 0.5d0*xi*(xi2-yi2))/r3 |
| 113 |
|
|
dwiduz = - 8.0d0*xi*yi*zi/r3 |
| 114 |
|
|
|
| 115 |
|
|
dwjdux = 4.0d0*(yj*zj2 + 0.5d0*yj*(xj2-yj2))/r3 |
| 116 |
|
|
dwjduy = 4.0d0*(xj*zj2 - 0.5d0*xj*(xj2-yj2))/r3 |
| 117 |
|
|
dwjduz = - 8.0d0*xj*yj*zj/r3 |
| 118 |
|
|
|
| 119 |
|
|
dwipdux = 2.0d0*yi*uglyi/rij |
| 120 |
|
|
dwipduy = -2.0d0*xi*uglyi/rij |
| 121 |
|
|
dwipduz = 0.0d0 |
| 122 |
|
|
|
| 123 |
|
|
dwjpdux = 2.0d0*yj*uglyj/rij |
| 124 |
|
|
dwjpduy = -2.0d0*xj*uglyj/rij |
| 125 |
|
|
dwjpduz = 0.0d0 |
| 126 |
|
|
|
| 127 |
|
|
! do the torques first since they are easy: |
| 128 |
|
|
! remember that these are still in the body fixed axes |
| 129 |
|
|
|
| 130 |
|
|
txi = 0.5d0*SSD_v0*(s*dwidux + sp*dwipdux) |
| 131 |
|
|
tyi = 0.5d0*SSD_v0*(s*dwiduy + sp*dwipduy) |
| 132 |
|
|
tzi = 0.5d0*SSD_v0*(s*dwiduz + sp*dwipduz) |
| 133 |
|
|
|
| 134 |
|
|
txj = 0.5d0*SSD_v0*(s*dwjdux + sp*dwjpdux) |
| 135 |
|
|
tyj = 0.5d0*SSD_v0*(s*dwjduy + sp*dwjpduy) |
| 136 |
|
|
tzj = 0.5d0*SSD_v0*(s*dwjduz + sp*dwjpduz) |
| 137 |
|
|
|
| 138 |
|
|
! go back to lab frame using transpose of rotation matrix: |
| 139 |
|
|
|
| 140 |
|
|
tlx(atom1) = tlx(atom1) + a(1,i)*txi + a(4,i)*tyi + a(7,i)*tzi |
| 141 |
|
|
tly(atom1) = tly(atom1) + a(2,i)*txi + a(5,i)*tyi + a(8,i)*tzi |
| 142 |
|
|
tlz(atom1) = tlz(atom1) + a(3,i)*txi + a(6,i)*tyi + a(9,i)*tzi |
| 143 |
|
|
|
| 144 |
|
|
tlx(atom2) = tlx(atom2) + a(1,j)*txj + a(4,j)*tyj + a(7,j)*tzj |
| 145 |
|
|
tly(atom2) = tly(atom2) + a(2,j)*txj + a(5,j)*tyj + a(8,j)*tzj |
| 146 |
|
|
tlz(atom2) = tlz(atom2) + a(3,j)*txj + a(6,j)*tyj + a(9,j)*tzj |
| 147 |
|
|
|
| 148 |
|
|
! Now, on to the forces: |
| 149 |
|
|
|
| 150 |
|
|
! first rotate the i terms back into the lab frame: |
| 151 |
|
|
|
| 152 |
|
|
fxii = a(1,i)*(s*dwidx+sp*dwipdx) + & |
| 153 |
|
|
a(4,i)*(s*dwidy+sp*dwipdy) + & |
| 154 |
|
|
a(7,i)*(s*dwidz+sp*dwipdz) |
| 155 |
|
|
fyii = a(2,i)*(s*dwidx+sp*dwipdx) + & |
| 156 |
|
|
a(5,i)*(s*dwidy+sp*dwipdy) + & |
| 157 |
|
|
a(8,i)*(s*dwidz+sp*dwipdz) |
| 158 |
|
|
fzii = a(3,i)*(s*dwidx+sp*dwipdx) + & |
| 159 |
|
|
a(6,i)*(s*dwidy+sp*dwipdy) + & |
| 160 |
|
|
a(9,i)*(s*dwidz+sp*dwipdz) |
| 161 |
|
|
|
| 162 |
|
|
fxij = -fxii |
| 163 |
|
|
fyij = -fyii |
| 164 |
|
|
fzij = -fzii |
| 165 |
|
|
|
| 166 |
|
|
fxjj = a(1,j)*(s*dwjdx+sp*dwjpdx) + & |
| 167 |
|
|
a(4,j)*(s*dwjdy+sp*dwjpdy) + & |
| 168 |
|
|
a(7,j)*(s*dwjdz+sp*dwjpdz) |
| 169 |
|
|
fyjj = a(2,j)*(s*dwjdx+sp*dwjpdx) + & |
| 170 |
|
|
a(5,j)*(s*dwjdy+sp*dwjpdy) + & |
| 171 |
|
|
a(8,j)*(s*dwjdz+sp*dwjpdz) |
| 172 |
|
|
fzjj = a(3,j)*(s*dwjdx+sp*dwjpdx)+ & |
| 173 |
|
|
a(6,j)*(s*dwjdy+sp*dwjpdy) + & |
| 174 |
|
|
a(9,j)*(s*dwjdz+sp*dwjpdz) |
| 175 |
|
|
|
| 176 |
|
|
fxji = -fxjj |
| 177 |
|
|
fyji = -fyjj |
| 178 |
|
|
fzji = -fzjj |
| 179 |
|
|
|
| 180 |
|
|
! now assemble these with the radial-only terms: |
| 181 |
|
|
|
| 182 |
|
|
flx(atom1) = flx(atom1) + 0.5d0*SSD_v0*(dsdr*drdx*w + dspdr*drdx*wp + & |
| 183 |
|
|
fxii + fxji) |
| 184 |
|
|
fly(atom1) = fly(atom1) + 0.5d0*SSD_v0*(dsdr*drdy*w + dspdr*drdy*wp + & |
| 185 |
|
|
fyii + fyji) |
| 186 |
|
|
flz(atom1) = flz(atom1) + 0.5d0*SSD_v0*(dsdr*drdz*w + dspdr*drdz*wp + & |
| 187 |
|
|
fzii + fzji) |
| 188 |
|
|
|
| 189 |
|
|
flx(atom2) = flx(atom2) + 0.5d0*SSD_v0*(-dsdr*drdx*w - dspdr*drdx*wp + & |
| 190 |
|
|
fxjj + fxij) |
| 191 |
|
|
fly(atom2) = fly(atom2) + 0.5d0*SSD_v0*(-dsdr*drdy*w - dspdr*drdy*wp + & |
| 192 |
|
|
fyjj + fyij) |
| 193 |
|
|
flz(atom2) = flz(atom2) + 0.5d0*SSD_v0*(-dsdr*drdz*w - dspdr*drdz*wp + & |
| 194 |
|
|
fzjj + fzij) |
| 195 |
|
|
|
| 196 |
|
|
return |
| 197 |
|
|
end subroutine ssd_forces |
| 198 |
|
|
|
| 199 |
|
|
subroutine calc_s(r, s, sp, dsdr, dspdr) |
| 200 |
|
|
|
| 201 |
|
|
! calculates the switching functions and their derivatives for a given |
| 202 |
|
|
! value of r |
| 203 |
|
|
|
| 204 |
|
|
double precision r, s, sp, dsdr, dspdr |
| 205 |
|
|
double precision rl, ru, rup |
| 206 |
|
|
! distances are in angstroms |
| 207 |
|
|
parameter(rl = 2.75d0, ru = 3.35d0, rup = 4.0d0) |
| 208 |
|
|
|
| 209 |
|
|
if (r.lt.rl) then |
| 210 |
|
|
s = 1.0d0 |
| 211 |
|
|
sp = 1.0d0 |
| 212 |
|
|
dsdr = 0.0d0 |
| 213 |
|
|
dspdr = 0.0d0 |
| 214 |
|
|
elseif (r.gt.rup) then |
| 215 |
|
|
s = 0.0d0 |
| 216 |
|
|
sp = 0.0d0 |
| 217 |
|
|
dsdr = 0.0d0 |
| 218 |
|
|
dspdr = 0.0d0 |
| 219 |
|
|
else |
| 220 |
|
|
sp = ((rup + 2.0d0*r - 3.0d0*rl) * (rup-r)**2)/((rup - rl)**3) |
| 221 |
|
|
dspdr = 6.0d0*(r-rup)*(r-rl)/((rup - rl)**3) |
| 222 |
|
|
|
| 223 |
|
|
if (r.gt.ru) then |
| 224 |
|
|
s = 0.0d0 |
| 225 |
|
|
dsdr = 0.0d0 |
| 226 |
|
|
else |
| 227 |
|
|
s = ((ru + 2.0d0*r - 3.0d0*rl) * (ru-r)**2)/((ru - rl)**3) |
| 228 |
|
|
dsdr = 6.0d0*(r-ru)*(r-rl)/((ru - rl)**3) |
| 229 |
|
|
endif |
| 230 |
|
|
endif |
| 231 |
|
|
|
| 232 |
|
|
return |
| 233 |
|
|
end subroutine calc_s |
| 234 |
|
|
|
| 235 |
|
|
|
| 236 |
|
|
|
| 237 |
|
|
|
