NAMD
Macros | Functions
ComputeNonbondedTI.C File Reference
#include "ComputeNonbondedInl.h"
#include "ComputeNonbondedAlch.h"
#include "ComputeNonbondedBase.h"

Go to the source code of this file.

Macros

#define TIFLAG
 
#define CALCENERGY
 
#define NBTYPE   NBPAIR
 
#define FULLELECT
 
#define MERGEELECT
 
#define SLOWONLY
 
#define NBTYPE   NBSELF
 
#define FULLELECT
 
#define MERGEELECT
 
#define SLOWONLY
 
#define NBTYPE   NBPAIR
 
#define FULLELECT
 
#define MERGEELECT
 
#define SLOWONLY
 
#define NBTYPE   NBSELF
 
#define FULLELECT
 
#define MERGEELECT
 
#define SLOWONLY
 

Functions

void ti_vdw_force_energy_dUdl (BigReal A, BigReal B, BigReal r2, BigReal myVdwShift, BigReal switchdist2, BigReal cutoff2, BigReal switchfactor, Bool vdwForceSwitching, BigReal myVdwLambda, BigReal alchVdwShiftCoeff, Bool alchWCAOn, BigReal myRepLambda, BigReal *alch_vdw_energy, BigReal *alch_vdw_force, BigReal *alch_vdw_dUdl)
 

Macro Definition Documentation

#define CALCENERGY

Definition at line 144 of file ComputeNonbondedTI.C.

#define FULLELECT

Definition at line 189 of file ComputeNonbondedTI.C.

#define FULLELECT

Definition at line 189 of file ComputeNonbondedTI.C.

#define FULLELECT

Definition at line 189 of file ComputeNonbondedTI.C.

#define FULLELECT

Definition at line 189 of file ComputeNonbondedTI.C.

#define MERGEELECT

Definition at line 191 of file ComputeNonbondedTI.C.

#define MERGEELECT

Definition at line 191 of file ComputeNonbondedTI.C.

#define MERGEELECT

Definition at line 191 of file ComputeNonbondedTI.C.

#define MERGEELECT

Definition at line 191 of file ComputeNonbondedTI.C.

#define NBTYPE   NBPAIR

Definition at line 187 of file ComputeNonbondedTI.C.

#define NBTYPE   NBSELF

Definition at line 187 of file ComputeNonbondedTI.C.

#define NBTYPE   NBPAIR

Definition at line 187 of file ComputeNonbondedTI.C.

#define NBTYPE   NBSELF

Definition at line 187 of file ComputeNonbondedTI.C.

#define SLOWONLY

Definition at line 194 of file ComputeNonbondedTI.C.

#define SLOWONLY

Definition at line 194 of file ComputeNonbondedTI.C.

#define SLOWONLY

Definition at line 194 of file ComputeNonbondedTI.C.

#define SLOWONLY

Definition at line 194 of file ComputeNonbondedTI.C.

#define TIFLAG

Definition at line 143 of file ComputeNonbondedTI.C.

Function Documentation

void ti_vdw_force_energy_dUdl ( BigReal  A,
BigReal  B,
BigReal  r2,
BigReal  myVdwShift,
BigReal  switchdist2,
BigReal  cutoff2,
BigReal  switchfactor,
Bool  vdwForceSwitching,
BigReal  myVdwLambda,
BigReal  alchVdwShiftCoeff,
Bool  alchWCAOn,
BigReal  myRepLambda,
BigReal alch_vdw_energy,
BigReal alch_vdw_force,
BigReal alch_vdw_dUdl 
)
inline

Copyright (c) 1995, 1996, 1997, 1998, 1999, 2000 by The Board of Trustees of the University of Illinois. All rights reserved.

Definition at line 17 of file ComputeNonbondedTI.C.

References A, vdw_forceandenergy(), vdw_fswitch_forceandenergy(), vdw_fswitch_shift(), and vdw_switch().

22  {
23  BigReal U, F, dU, switchmul, switchmul2;
24  /*
25  * The variable naming here is unavoidably awful. A number immediately after
26  * a variable implies a distance to that power. The suffix "_2" indicates an
27  * evaluation at alchLambda2. The prefix "inv" means the inverse (previously
28  * this also used underscores - it was awful).
29  */
30  if (alchWCAOn) {
31  /* THIS BRANCH IS PARTIALLY INCOMPLETE AND BLOCKED INSIDE SimParameters
32  *
33  * The problem is that WCA creates TWO energy terms and thus two different
34  * TI derivatives, but the reduction code is currently only set up for
35  * one derivative. There is no such problem with FEP because the energy
36  * is not decomposed. In principle, the contribution to the free energy
37  * from each derivative is zero while the other is changing (e.g. while
38  * repulsion is changing dispersion is not), but the derivatives are still
39  * non-zero and it is NAMD convention to report them even as such.
40  * However, even if we were willing to break this convention, both
41  * derivatives still compute to the free energy at the boundary where
42  * repulsion is completely on and dispersion is exactly zero - two
43  * derivatives are thus required for a complete working implementation.
44  *
45  */
46 
47  // WCA-on, auxilliary, lambda-dependent cutoff based on Rmin
48  //
49  // Avoid divide by zero - correctly zeroes interaction below.
50  const BigReal Rmin2 = (B <= 0.0 ? 0.0 : powf(2.0*A/B, 1.f/3));
51  if (myRepLambda < 1.0) {
52  // modified repulsive soft-core
53  // All of the scaling is baked into the shift and cutoff values. There
54  // are no linear coupling terms out front.
55  const BigReal WCAshift = Rmin2*(1 - myRepLambda)*(1 - myRepLambda);
56  if (r2 <= Rmin2 - WCAshift) {
57  const BigReal epsilon = B*B/(4.0*A);
58  vdw_forceandenergy(A, B, r2 + WCAshift, &U, &F);
59  *alch_vdw_energy = U + epsilon;
60  *alch_vdw_force = F;
61  *alch_vdw_dUdl = Rmin2*(1 - myRepLambda)*F;
62  } else {
63  *alch_vdw_energy = 0.0;
64  *alch_vdw_force = 0.0;
65  *alch_vdw_dUdl = 0.0;
66  }
67  } else { // myRepLambda == 1.0
68  if (vdwForceSwitching) {
69  // force and potential switching
70  if (r2 <= Rmin2) {
71  // normal LJ, potential w/two shifts
72  const BigReal epsilon = B*B/(4.0*A);
73  vdw_fswitch_shift(A, B, switchdist2, cutoff2, &dU);
74  vdw_forceandenergy(A, B, r2, &U, &F);
75  *alch_vdw_energy = U + (1 - myVdwLambda)*epsilon + myVdwLambda*dU;
76  *alch_vdw_force = F;
77  *alch_vdw_dUdl = dU - epsilon;
78  } else if (r2 <= switchdist2) {
79  // normal LJ potential w/shift
80  vdw_fswitch_shift(A, B, switchdist2, cutoff2, &dU);
81  vdw_forceandenergy(A, B, r2, &U, &F);
82  *alch_vdw_energy = myVdwLambda*(U + dU);
83  *alch_vdw_force = myVdwLambda*F;
84  *alch_vdw_dUdl = U + dU;
85  } else { // r2 > switchdist
86  // normal LJ potential with linear coupling
87  vdw_fswitch_forceandenergy(A, B, r2, switchdist2, cutoff2, &U, &F);
88  *alch_vdw_energy = myVdwLambda*U;
89  *alch_vdw_force = myVdwLambda*F;
90  *alch_vdw_dUdl = U;
91  }
92  } else {
93  // potential switching - also correct for no switching
94  if (r2 <= Rmin2) {
95  // normal LJ potential w/shift
96  const BigReal epsilon = B*B/(4.0*A);
97  vdw_forceandenergy(A, B, r2, &U, &F);
98  *alch_vdw_energy = U + (1 - myVdwLambda)*epsilon;
99  *alch_vdw_force = F;
100  *alch_vdw_dUdl = -epsilon;
101  } else { // r2 > Rmin2
102  // normal LJ potential with linear coupling
103  vdw_switch(r2, switchdist2, cutoff2, switchfactor, &switchmul, \
104  &switchmul2);
105  vdw_forceandenergy(A, B, r2, &U, &F);
106  *alch_vdw_energy = myVdwLambda*switchmul*U;
107  *alch_vdw_force = myVdwLambda*(switchmul*F + switchmul2*U);
108  *alch_vdw_dUdl = switchmul*U;
109  }
110  }
111  }
112  } else { // WCA-off
113  if (vdwForceSwitching) {
114  // force and potential switching
115  if (r2 <= switchdist2) {
116  // normal LJ potential w/shift
117  vdw_forceandenergy(A, B, r2 + myVdwShift, &U, &F);
118  vdw_fswitch_shift(A, B, switchdist2 + myVdwShift, cutoff2, &dU);
119  *alch_vdw_energy = myVdwLambda*(U + dU);
120  *alch_vdw_force = myVdwLambda*F;
121  *alch_vdw_dUdl = U + 0.5*myVdwLambda*alchVdwShiftCoeff*F + dU;
122  } else { // r2 > switchdist2
123  vdw_fswitch_forceandenergy(A, B, r2 + myVdwShift, \
124  switchdist2 + myVdwShift, cutoff2, &U, &F);
125  *alch_vdw_energy = myVdwLambda*U;
126  *alch_vdw_force = myVdwLambda*F;
127  *alch_vdw_dUdl = U + 0.5*myVdwLambda*alchVdwShiftCoeff*F;
128  }
129  } else {
130  // potential switching - also correct for no switching
131  vdw_switch(r2, switchdist2, cutoff2, switchfactor, &switchmul, \
132  &switchmul2);
133  vdw_forceandenergy(A, B, r2 + myVdwShift, &U, &F);
134  *alch_vdw_energy = myVdwLambda*switchmul*U;
135  *alch_vdw_force = myVdwLambda*(switchmul*F + switchmul2*U);
136  *alch_vdw_dUdl = switchmul*(U + 0.5*myVdwLambda*alchVdwShiftCoeff*F);
137  }
138  }
139 }
void vdw_forceandenergy(const BigReal A, const BigReal B, const BigReal r2, BigReal *U, BigReal *F)
const BigReal A
void vdw_switch(const BigReal r2, const BigReal switchdist2, const BigReal cutoff2, const BigReal switchfactor, BigReal *switchmul, BigReal *switchmul2)
void vdw_fswitch_forceandenergy(const BigReal A, const BigReal B, const BigReal r2, const BigReal switchdist2, const BigReal cutoff2, BigReal *U, BigReal *F)
const BigReal B
void vdw_fswitch_shift(const BigReal A, const BigReal B, const BigReal switchdist2, const BigReal cutoff2, BigReal *dU)
__global__ void const int const TileList *__restrict__ TileExcl *__restrict__ const int *__restrict__ const int const float2 *__restrict__ const int *__restrict__ const float3 const float3 const float3 const float4 *__restrict__ const float cutoff2
double BigReal
Definition: common.h:114