Understanding

Molecular Simulation
From Algorithms to Applications

Daan Frenkel
FOM Institute for Atomic and Molecular Physics,
Amsterdam, The Netherlands

Department of Chemical Engineering,

Faculty of Sciences
University of Amsterdam
Amsterdam, The Netherlands

Berend Smit
Department of Chemical Engineering
Faculty of Sciences
University of Amsterdam
Amsterdam, The Netherlands

ACADEMIC PRESS
A Division of Harcourt, Inc.
San Diego San Francisco New York
Boston London Sydney Tokyo
Contents

Preface to the Second Edition xiii

Preface xv

List of Symbols xix

1 Introduction 1

Part I Basics 7
2 Statistical Mechanics 9
2.1 Entropy and Temperature 9
2.2 Classical Statistical Mechanics 13
2.2.1 Ergodicity 15
2.3 Questions and Exercises 17

3 Monte Carlo Simulations V 23

3.1 The Monte Carlo Method . 23
3.1.1 Importance Sampling . 24
3.1.2 The Metropolis Method 27
3.2 A Basic Monte Carlo Algorithm 31
3.2.1 The Algorithm 31
3.2.2 Technical Details 32
3.2.3 Detailed Balance versus Balance 42
3.3 Trial Moves 43
3.3.1 Translational Moves 43
3.3.2 Orientational Moves 48
3.4 Applications 51
3.5 Questions and Exercises 58
vi Contents

4 Molecular Dynamics Simulations 63

4.1 Molecular Dynamics: The Idea 63
4.2 Molecular Dynamics: A Program 64
4.2.1 Initialization 65
4.2.2 The Force Calculation 67
4.2.3 Integrating the Equations of Motion 69
4.3 Equations of Motion 71
4.3.1 Other Algorithms 74
4.3.2 Higher-Order Schemes 77
4.3.3 Liouville Formulation of Time-Reversible Algorithms . 77
4.3.4 Lyapunov Instability 81
4.3.5 One More Way to Look at the Verlet Algorithm 82
4.4 Computer Experiments 84
4.4.1 Diffusion 87
4.4.2 Order-n Algorithm to Measure Correlations 90
4.5 Some Applications : 97
4.6 Questions and Exercises 105

Part II Ensembles 109

5 Monte Carlo Simulations in Various Ensembles 111
5.1 General Approach 112
5.2 Canonical Ensemble 112
5.2.1 Monte Carlo Simulations 113
5.2.2 Justification of the Algorithm 114
5.3 Microcanonical Monte Carlo 114
5.4 Isobaric-Isothermal Ensemble 115
5.4.1 Statistical Mechanical Basis 116
5.4.2 Monte Carlo Simulations ; 119
5.4.3 Applications 122
5.5 Isotension-Isothermal Ensemble 125
5.6 Grand-Canonical Ensemble 126
5.6.1 Statistical Mechanical Basis 127
5.6.2 Monte Carlo Simulations 130
5.6.3 Justification of the Algorithm 130
5.6.4 Applications 133
5.7 Questions and Exercises 135
6 Molecular Dynamics in Various Ensembles 139
6.1 Molecular Dynamics at Constant Temperature 140
6.1.1 The Andersen Thermostat 141
6.1.2 Nose-Hoover Thermostat 147
Contents vii

6.1.3 Nose-Hoover Chains 155

6.2 Molecular Dynamics at Constant Pressure 158
6.3 Questions and Exercises 160

Part III Free Energies and Phase Equilibria 165

7 Free Energy Calculations 167
7.1 Thermodynamic Integration 168
7.2 Chemical Potentials 172
7.2.1 The Particle Insertion Method 173
7.2.2 Other Ensembles 176
7.2.3 Overlapping Distribution Method 179
7.3 Other Free Energy Methods 183
7.3.1 Multiple Histograms 183
7.3.2 Acceptance Ratio Method 189
7.4 Umbrella Sampling . . . 192
7.4.1 Nonequilibrium Free Energy Methods 196
7.5 Questions and Exercises 199
8 The Gibbs Ensemble 201
8.1 The Gibbs Ensemble Technique 203
8.2 The Partition Function 204
8.3 Monte Carlo Simulations 205
8.3.1 Particle Displacement 205
8.3.2 Volume Change 206
8.3.3 Particle Exchange 208
8.3.4 Implementation 208
8.3.5 Analyzing the Results 214
8.4 Applications \. 220
8.5 Questions and Exercises . . \ 223
9 Other Methods to Study Coexistence 225
9.1 Semigrand Ensemble 225
9.2 Tracing Coexistence Curves 233

10 Free Energies of Solids 241

10.1 Thermodynamic Integration 242
10.2 Free Energies of Solids 243
10.2.1 Atomic Solids with Continuous Potentials 244
10.3 Free Energies of Molecular Solids 245
10.3.1 Atomic Solids with Discontinuous Potentials 248
10.3.2 General Implementation Issues 249
10.4 Vacancies and Interstitials 263
viii Contents

10.4.1 Free Energies 263

10.4.2 Numerical Calculations 266
11 Free Energy of Chain Molecules 269
11.1 Chemical Potential as Reversible Work 269
11.2 Rosenbluth Sampling 271
11.2.1 Macromolecules with Discrete Conformations 271
11.2.2 Extension to Continuously Deformable Molecules . . . 276
11.2.3 Overlapping Distribution Rosenbluth Method 282
11.2.4 Recursive Sampling 283
11.2.5 Pruned-Enriched Rosenbluth Method 285

Part IV Advanced Techniques 289

12 Long-Range Interactions 291
12.1 EwaldSums 292
12.1.1 Point Charges 292
12.1.2 Dipolar Particles 300
12.1.3 Dielectric Constant 301
12.1.4 Boundary Conditions 303
12.1.5 Accuracy and Computational Complexity 304
12.2 Fast Multipole Method 306
12.3 Particle Mesh Approaches 310
12.4 Ewald Summation in a Slab Geometry 316

13 Biased Monte Carlo Schemes 321

13.1 Biased Sampling Techniques 322
13.1.1 Beyond Metropolis 323
13.1.2 Orientational Bias 323
13.2 Chain Molecules 331
13.2.1 Configurational-Bias Monte Carlo 331
13.2.2 Lattice Models 332
13.2.3 Off-lattice Case 336
13.3 Generation of Trial Orientations 341
13.3.1 Strong Intramolecular Interactions 342
13.3.2 Generation of Branched Molecules 350
13.4 Fixed Endpoints 353
13.4.1 Lattice Models : . . . 353
13.4.2 Fully Flexible Chain 355
13.4.3 Strong Intramolecular Interactions 357
13.4.4 Rebridging Monte Carlo 357
13.5 Beyond Polymers 360
13.6 Other Ensembles 365
Contents ix

13.6.1 Grand-Canonical Ensemble 365

13.6.2 Gibbs Ensemble Simulations 370
13.7 Recoil Growth 374
13.7.1 Algorithm 376
13.7.2 Justification of the Method 379
13.8 Questions and Exercises 383
14 Accelerating Monte Carlo Sampling 389
14.1 Parallel Tempering 389
14.2 Hybrid Monte Carlo 397
14.3 Cluster Moves 399
14.3.1 Clusters 399
14.3.2 Early Rejection Scheme 405

15 Tackling Time-Scale Problems 409

15.1 Constraints 410
15.1.1 Constrained and Unconstrained Averages 415
15.2 On-the-Fly Optimization: Car-Parrinello Approach 421
15.3 Multiple Time Steps 424
16 Rare Events 431
16.1 Theoretical Background 432
16.2 Bennett-Chandler Approach 436
16.2.1 Computational Aspects 438
16.3 Diffusive Barrier Crossing 443
16.4 Transition Path Ensemble 450
16.4.1 Path Ensemble 451
16.4.2 Monte Carlo Simulations 454
16.5 Searching for the Saddle Point 462

17 Dissipative Particle Dynamics ; 465

17.1 Description of the Technique 466
17.1.1 Justification of the Method 467
17.1.2 Implementation of the Method 469
17.1.3 DPD and Energy Conservation 473
17.2 Other Coarse-Grained Techniques 476

Part V Appendices 479

A Lagrangian and Hamiltonian 481
A.I Lagrangian 483
A.2 Hamiltonian 486
A.3 Hamilton Dynamics and Statistical Mechanics 488
 Contents

A.3.1 Canonical Transformation 489

A.3.2 Symplectic Condition 490
A.3.3 Statistical Mechanics 492
B Non-Hamiltonian Dynamics 495
B.I Theoretical Background 495
B.2 Non-Hamiltonian Simulation of the N,V,T Ensemble 497
B.2.1 The Nose-Hoover Algorithm 498
B.2.2 Nose-Hoover Chains 502
B.3 The N,P,T Ensemble 505
C Linear Response Theory 509
C.I Static Response 509
C.2 Dynamic Response 511
C.3 Dissipation 513
C.3.1 Electrical Conductivity 516
C.3.2 Viscosity 518
C.4 Elastic Constants 519
D Statistical Errors 525
D.I Static Properties: System Size 525
D.2 Correlation Functions 527
D.3 Block Averages 529

E Integration Schemes 533

E.I Higher-Order Schemes 533
E.2 Nose-Hoover Algorithms 535
E.2.1 Canonical Ensemble 536
E.2.2 The Isothermal-Isobaric Ensemble 540

F Saving CPU Time 545

F.I VerletList 545
F.2 Cell Lists 550
F.3 Combining the Verlet and Cell Lists 550
F.4 Efficiency 552

G Reference States 559

G.I Grand-Canonical Ensemble Simulation 559

H Statistical Mechanics of the Gibbs "Ensemble" 563

H.I Free Energy of the Gibbs Ensemble 563
H.1.1 Basic Definitions 563
H.1.2 Free Energy Density 565
H.2 Chemical Potential in the Gibbs Ensemble 570
Contents xi

I Overlapping Distribution for Polymers 573

J Some General Purpose Algorithms 577

K Small Research Projects 581

K.I Adsorption in Porous Media 581
K.2 Transport Properties in Liquids 582
K.3 Diffusion in a Porous Media 583
K.4 Multiple-Time-Step Integrators 584
K.5 Thermodynamic Integration 585
L Hints for Programming 587

Bibliography 589

Author Index 619

Index 628