Download - Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC
![Page 1: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/1.jpg)
Geometry Optimization, Molecular Dynamics
and Vibrational Spectra
Pablo Ordejón
ICMAB-CSIC
![Page 2: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/2.jpg)
Born-Oppenheimer dynamics
Nuclei are much slower than electrons
en mm electronic
nuclear
decoupling
Classical Nuclear Dynamics
![Page 3: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/3.jpg)
Optimization and MD basic cycle.
![Page 4: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/4.jpg)
Extracting information from the Potential Energy Surface (PES)
Optimizations and Phonons:Optimizations and Phonons:- We move We move onon the PES the PES- Local vs global minimaLocal vs global minima- PES is harmonic close to minPES is harmonic close to min..
MDMD- We move We move overover the PES the PES (KE)(KE)- Good Sampling is Good Sampling is required!!required!!
![Page 5: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/5.jpg)
OPTIMIZATION METHODS
![Page 6: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/6.jpg)
Theory of geometry optimization
Gradients Hessian
=1 for quadratic region
![Page 7: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/7.jpg)
Methods of Optimisation
• Energy only:– simplex
• Energy and first derivatives (forces):– steepest descents (poor convergence)– conjugate gradients (retains information)– approximate Hessian update
• Energy, first and second derivatives– Newton-Raphson– Broyden (BFGS) updating of Hessian (reduces inversions)– Rational Function Optimisation (for transition states/ and soft
modes)
SIESTA presently uses conjugate gradients and BFGS
![Page 8: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/8.jpg)
Steepest Descents
• Start at xo
• Calculate gradient g(x) = f(x)
• Minimize f(x) along the line defined by the gradient
• Start again until tolerance is reached
![Page 9: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/9.jpg)
Steepest Descents
Performance depends on• Eigenvalues of Hessian (λmax / λmin)• Starting point
![Page 10: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/10.jpg)
Conjugate Gradients
• Same idea, but retaining information about previous steps
• Search directions ‘conjugate’ (orthogonal) to previous
• Convergence assured for N steps
See “Numerical Recipes” by Press et al (Cambridge) for details
![Page 11: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/11.jpg)
Using the Hessian
• Energy, first and second derivatives:- Newton-Raphson: An approximation of H at a position (Xk) is calculated.Then finding the inverse of that Hessian (H-1), and solving the equation P = -H-
1*F(Xk) gives a good search direction P. Later, a line search procedure has to determine just how much to go in that direction (producing the scalar alpha). The new position is given by: Xk+1 = Xk + alpha*P
• Broyden (BFGS): updating of an approximate HessianBasic idea, update the Hessian along the minimization to fit:
... using only the forces!!
![Page 12: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/12.jpg)
Optimization Variables in SIESTA(1)
• Set runtype to conjugate gradients or Broyden: MD.TypeOfRun CG, Broyden
• Set maximum number of iterative steps:MD.NumCGsteps 100
• Optionally set force tolerance:MD.MaxForceTol 0.04 eV/Ang
• Optionally set maximum displacement:MD.MaxCGDispl 0.2 Bohr
![Page 13: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/13.jpg)
Optimization Variables in SIESTA(2)
• By default: optimisations are done for a fixed cell
• To allow unit cell to vary:MD.VariableCell true
• Optionally set stress tolerance:MD.MaxStressTol 1.0 Gpa
• Optionally set cell preconditioning:MD.PreconditionVariableCell 5.0 Ang
• Set an applied pressure:MD.TargetPressure 5.0 GPa
![Page 14: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/14.jpg)
Advice on Optimizations in SIESTA(I)
• Make sure that your MeshCutoff is large enough:
- Mesh leads to space rippling- If oscillations are large convergence is slow- May get trapped in wrong local minimum
![Page 15: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/15.jpg)
Advice on Optimizations in SIESTA(II)
Ill conditioned systems (soft modes) can slow down optimizations, very sensitive to mesh cuttoff.
- Use constraints when relevant.
Fixed to Si BulkFixed to Si Bulk
![Page 16: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/16.jpg)
Advice on Optimizations in SIESTA(III)
Decouple Degrees of freedom (relax separately different parts of the system).
Look at the evolution of relevant physics quantities (band structure, Ef).
No constraints
Fix the Zeolite,Fix the Zeolite,Its relaxation is noIts relaxation is noLonger relevant.Longer relevant.FFtubetube<0.04 eV/A<0.04 eV/AFFzeolzeol>0.1 eV/A>0.1 eV/A
![Page 17: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/17.jpg)
More Advice on Optimisation…..
• Optimise internal degrees of freedom first• Optimise unit cell after internals• Exception is simple materials (e.g. MgO)• Large initial pressure can cause slow
convergence• Small amounts of symmetry breaking can occur• Check that geometry is sufficiently converged
(as opposed to force - differs according to Hessian)
• SCF must be more converged than optimisation• Molecular systems are hardest to optimise
![Page 18: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/18.jpg)
Using Constraints
• The following can currently be constrained:- atom positions
- cell strains
- Z-matrix (internal coordinates)• User can create their own subroutine (constr)• To fix atoms:
• To fix stresses:
Stress notation: 1=xx, 2=yy, 3=zz, 4=yz, 5=xz, 6=xy
![Page 19: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/19.jpg)
What you hope for!
![Page 20: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/20.jpg)
MOLECULAR DYNAMICS (MD)
![Page 21: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/21.jpg)
•Follows the time evolution of a systemFollows the time evolution of a system•Solve Newton’s equations of motion:Solve Newton’s equations of motion:
•Treats electrons quantum mechanicallyTreats electrons quantum mechanically•Treats nuclei Treats nuclei classicallyclassically
•Hydrogen may raise issuesHydrogen may raise issues:: tunnelling tunnelling, zero , zero point E...point E...
•Allows study of dynamic processesAllows study of dynamic processes•Annealing of complex materialsAnnealing of complex materials•IInfluence of temperaturenfluence of temperature and pressure and pressure
•Time averages Time averages vvs Statistical averagess Statistical averages
Molecular Dynamics
2
2 )()()(
dttxd
mtamxd
dEtF
![Page 22: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/22.jpg)
Ergodicity
• In MD we want to replace a full sampling on the appropriate statistical ensemble by a SINGLE very long trajectory.
• This is OK only if system is ergodic.
• Ergodic Hypothesis: a phase point for any isolated system passes in succession through every point compatible with the energy of the system before finally returning to its original position in phase space. This journey takes a Poincare cycle.
• In other words, Ergodic hypothesis: each state consistent with our knowledge is equally “likely”.
– Implies the average value does not depend on initial conditions.– <A>time= <A>ensemble , so <Atime> = (1/NMD) = ∑t=1,N At is good
estimator.
• Are systems in nature really ergodic? Not always! – Non-ergodic examples are glasses, folding proteins (in practice)
and harmonic crystals (in principle).
![Page 23: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/23.jpg)
Different aspects of ergodicity
• The system relaxes on a “reasonable” time scale towards a unique equilibrium state (microcanonical state)
• Trajectories wander irregularly through the energy surface eventually sampling all of accesible phase space.
• Trajectories initially close together separate rapidily (Sensitivity to initial conditions).
Ergodic behavior makes possible the use ofstatistical methods on MD of small system. Small round-off errors and other mathematicalapproximations should not matter.
![Page 24: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/24.jpg)
From J.M. Haile: MD From J.M. Haile: MD SimulationsSimulations
Particle in a smooth/rough circle
![Page 25: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/25.jpg)
Molecular Dynamics(I)
In Molecular Dynamics simulations, one computes the evolution of the positions and velocities with time, solving Newton’s equations.
• Algorithm to integrate Newton’s equations: “Verlet”
• Initial conditions in space and time.
'')''(')()()()(
)()()(
'1
000
2
2
00
dttFdttttvtxtx
dttxd
mtamxd
dEtF
t
t m
t
t
![Page 26: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/26.jpg)
Molecular Dynamics(II)
• Initialize positions and momenta at t=0 (initial conditions in space and time)
• Solve F = ma to determine r(t), v(t). (integrator)
– We need to make time discrete, instead of continuous!!!
•Calculate the properties of interest along the trajectory
•Estimate errors
•Use the results of the simulation to answer physical questions!!.
t0 t1 t2 tNtn tn+
1
tn-1
h=t
![Page 27: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/27.jpg)
Molecular Dynamics III
• Timestep must be small enough to accurately sample highest frequency motion
• Typical timestep is 1 fs (1 x 10-15 s)
• Typical simulation length: Depends on the system of study!!(the more complex the PES the longer the simulation time)
• Is this timescale relevant to your process?
• Simulation has two parts– equilibration – when properties do not depend on time– production (record data)
• Results:– diffusion coefficients– Structural information (RDF’s,) – Free energies / phase transformations (very hard!)
• Is your result statistically significant?
![Page 28: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/28.jpg)
Integrator: Verlet algorithm
The most commonly used algorithm:
r(t+h) = r(t) + v(t) h + 1/2 a(t) h2 + b(t) h3 + O(h4) (Taylor series expansion)r(t-h) = r(t) - v(t) h + 1/2 a(t) h2 - b(t) h3 + O(h4)
r(t+h) = 2 r(t) - r(t-h) + a(t) h2 + O(h4) Sum
v(t) = (r(t+h) - r(t-h))/(2h) + O(h2) Difference (estimated velocity)
• Trajectories are obtained from the first equation. Velocities are not necessary.
• Errors in trajectory: O(h4)• Preserves time reversal symmetry.• Excellent energy conservation. • Modifications and alternative schemes exist (leapfrog, velocity
Verlet), always within the second order approximation• Higher order algorithms: Gear
![Page 29: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/29.jpg)
Different ensembles: conserved magnitudes
• NVE (Verlet): NVE (Verlet): Microcanonical.Microcanonical.
• Integrates Newtons equations of Integrates Newtons equations of motion, for N particles, in a fixed motion, for N particles, in a fixed volume V.volume V.
• Natural time evolution of the Natural time evolution of the system: E is a constant of motionsystem: E is a constant of motion
• NVT (Nose): NVT (Nose): CanonicalCanonical
• System in thermal contact with a System in thermal contact with a heat bath.heat bath.
• Extended Lagrangian:Extended Lagrangian:• N particles + Thermostat, mass N particles + Thermostat, mass
Q.Q.
• NPE (Parrinello-NPE (Parrinello-Rahman) (isobaric)Rahman) (isobaric)
• Extended LagrangianExtended Lagrangian• Cell vectors are dynamical Cell vectors are dynamical
variables with an associated variables with an associated mass.mass.
• NPT (Nose-NPT (Nose-Parrinello-Rahman)Parrinello-Rahman)
• 2 Extended Lagrangians2 Extended Lagrangians• NVT+NPE.NVT+NPE.
Same sampling In the thermodynamic limit
![Page 30: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/30.jpg)
Nose-Hoover thermostat
• MD in canonical distribution (TVN)• Introduce a friction force (t)
T Reservoir
SYSTEM
p(t))(t)F(q,dt
dpt
Dynamics of friction coefficient to get canonical Dynamics of friction coefficient to get canonical ensemble.ensemble.
Feedback makes Feedback makes K.E.=3/2kTK.E.=3/2kT
Q= fictitious “heat bath mass”. Large Q is weak Q= fictitious “heat bath mass”. Large Q is weak couplingcoupling
Qddt
1
2 mivi
2 3N
2kBT
![Page 31: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/31.jpg)
Hints
• Nose Mass: Match a vibrational frequency of the system, better high energy frequency
![Page 32: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/32.jpg)
Molecular Dynamics in SIESTA(1)
• MD.TypeOfRun VerletNVE ensemble dynamics
• MD.TypeOfRun NoseNVT dynamics with Nose thermostat
• MD.TypeOfRun ParrinelloRahmanNPE dynamics with P-R barostat
• MD.TypeOfRun NoseParrinelloRahmanNPT dynamics with thermostat/barostat
• MD.TypeOfRun AnnealAnneals to specified p and T
Variable Cell
![Page 33: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/33.jpg)
Molecular Dynamics in SIESTA(2)
• Setting the length of the run:MD.InitialTimeStep 1MD.FinalTimeStep 2000
• Setting the timestep:MD.LengthTimeStep 1.0 fs
• Setting the temperature:MD.InitialTemperature 298 KMD.TargetTemperature 298 K
• Setting the pressure:MD.TargetPressure 3.0 Gpa
• Thermostat / barostat parameters:MD.NoseMass / MD.ParrinelloRahmanMass
Maxwell-Boltzmann
![Page 34: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/34.jpg)
Annealing in SIESTA
• MD can be used to optimize structures:MD.Quench true - zeros velocity when opposite to force
• MD annealing:MD.AnnealOption PressureMD.AnnealOption TemperatureMD.AnnealOption TemperatureAndPressure
• Timescale for achieving targetMD.TauRelax 100.0 fs
![Page 35: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/35.jpg)
Output Files
• SystemLabel.MDE conserved quantity
• SystemLabel.MD (unformatted; post-process with iomd.F)• SystemLabel.ANI (coordinates in xyz format)• If Force Constants run: SystemLabel.FC
![Page 36: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/36.jpg)
Visualisation and Analysis
Molekelhttp://www.cscs.ch/molekel
XCRYSDENhttp://www.xcrysden.org/
GDIShttp://gdis.seul.org/
![Page 37: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/37.jpg)
VIBRATIONAL SPECTRA
![Page 38: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/38.jpg)
Vibrational spectra: Phonons
Calculating Dynamical Matrix: Mass weighted Hessian Matrix
• Frozen Phonon approximation: • Numerical evaluation of the second derivatives. (finite differences).
• Density Functional Perturbation Theory (Linear Response): • Perturbation theory used to obtain analytically the Energy second derivatives within a
self consistent procedure.
Molecular dynamics: Green-Kubo linear response.• Link between time correlation functions and the response of the system to weak
perturbations.
Harmonic Approx.Harmonic Approx.
Beyond Harmonic ApproxBeyond Harmonic Approx.
![Page 39: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/39.jpg)
Phonons in Siesta (I)
• Frozen Phonon approximation:
MD.TypeOfRun FC
MD.FCDispl 0.04 Bohr (default)
Total number of SCF cycles: 3 X 2 X N = 6N
(x,y,z) (+,-) Nat
Output file: SystemLabel.FC
Building and diagonalization of Dynamical matrix: Vibra Suite (Vibrator) (in /Util)
![Page 40: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/40.jpg)
Phonons in Siesta (II)
1. Relax the system: Max F<0.02 eV/Ang
2. Increase MeshCutof, and run FC.
3. If possible, test the effect of MaxFCDispl.
![Page 41: Geometry Optimization, Molecular Dynamics and Vibrational Spectra Pablo Ordejón ICMAB-CSIC](https://reader035.vdocuments.us/reader035/viewer/2022062221/56649f1d5503460f94c34794/html5/thumbnails/41.jpg)
Phonons and MD
• MD simulations (NVE)
• Fourier transform of Velocity-Velocity autocorrelation function.
• Anharmonic effects: (T)
• Expensive, but information available for MD simulations.