User Tools

Site Tools


exercises:2018_uzh_cmest:path_optimization_neb

Path optimization using NEB

In one of the last exercises you calculated the energy for Ethane for two slightly different geometries and noticed that the geometry optimization was not able to change one structure into the other with lower energy. As presented in the lecture, it may happen quiet often that a minimization algorithm gets stuck in a local minimum, respectively it is not guaranteed to find the global minimum.

In this exercise, we will therefore perform Nudged Elastic Band (NEB) calculations using the same molecule as before and investigate the energy path between the two geometries.

Following are four geometry files you should put/create in a new exercise directory:

ethane_1_opt.xyz
       8
 i =       12, E =       -14.9518242480
  C         6.7709731556        5.9999999991        6.2147000005
  C         5.2290258859        6.0000000007        6.2147000001
  H         4.8193442852        6.0000000005        5.1955521086
  H         4.8183461262        6.8823861870        6.7231540475
  H         4.8183461248        5.1176138164        6.7231540484
  H         7.1806551657        5.9999999994        5.1955522543
  H         7.1816533264        6.8823860579        6.7231539750
  H         7.1816533250        5.1176139384        6.7231539741
ethane_s1.xyz
       8
 i =      76, E =      -14.9559722838
  C         6.7640435612        6.0000000003        5.9997401503
  C         5.2359564388        5.9999999997        6.0002598497
  H         4.8332682152        6.0000000001        7.0232923722
  H         4.8330445407        5.1142018851        5.4887808109
  H         4.8330445407        6.8857981148        5.4887808113
  H         7.1667317848        5.9999999999        4.9767076278
  H         7.1669554593        6.8857981149        6.5112191891
  H         7.1669554593        5.1142018852        6.5112191887
ethane_ts.xyz
       8
 i =      76, E =      -14.9518421887
  C         6.7713019119        5.9963482236        5.9999305287
  C         5.2296433443        6.0033677689        6.0065640271
  H         4.8226439717        6.0198987997        7.0258669051
  H         4.8144385670        5.1135492476        5.5115353199
  H         4.8205805610        6.8779314573        5.4838302216
  H         7.1765327350        5.1402836770        5.4459664211
  H         7.1832404413        6.9024073310        5.5339472993
  H         7.1852801526        5.9447026992        7.0166786010
ethane_s2.xyz
       8
 i =      76, E =      -14.9559544815
  C         6.7635882192        5.9976047386        6.0012623335
  C         5.2356460989        6.0047571469        6.0004006617
  H         4.8320044755        6.0067336493        7.0232527262
  H         4.8290963082        5.1202363024        5.4891970991
  H         4.8366693774        6.8919269855        5.4881514124
  H         7.1628913027        5.1334052558        6.5507178317
  H         7.1674882265        5.9517120047        4.9798028986
  H         7.1709320051        6.9026197936        6.4738574951

You could in principle start from the geometries you already optimized. In fact, the files ethane_1_opt.xyz and ethane_s1.xyz are geometry optimizations resulting from the previous ethane1.xyz and ethane2.xyz with slightly different settings (more details below) and some modifications.

The input file you need then looks as follows:

ethane_neb_aba.inp
&GLOBAL
  PROJECT ethane_neb_aba
  RUN_TYPE BAND 
  PRINT_LEVEL MEDIUM
&END GLOBAL

&FORCE_EVAL
  METHOD Quickstep              ! Electronic structure method (DFT,...)
  &DFT
    BASIS_SET_FILE_NAME  BASIS_MOLOPT
    POTENTIAL_FILE_NAME  POTENTIAL

    &POISSON                    ! Solver requested for non periodic calculations
      PERIODIC XYZ
    &END POISSON
    &SCF                        ! Parameters controlling the convergence of the scf. This section should not be changed. 
      SCF_GUESS ATOMIC
      EPS_SCF 1.0E-6
      MAX_SCF 300
    &END SCF
    &XC                        ! Parameters needed to compute the electronic exchange potential 
      &XC_FUNCTIONAL PBE
      &END XC_FUNCTIONAL
    &END XC
  &END DFT

  &SUBSYS
    &CELL
      ABC 12. 12. 12.
      PERIODIC XYZ
    &END CELL
    &TOPOLOGY                    ! Section used to center the atomic coordinates in the given box. Useful for big molecules
      &CENTER_COORDINATES
      &END
      COORD_FILE_FORMAT xyz
      COORD_FILE_NAME  ./ethane_1_opt.xyz
    &END
    &KIND H
      ELEMENT H
      BASIS_SET DZVP-MOLOPT-GTH
      POTENTIAL GTH-PBE-q1
    &END KIND
    &KIND C
      ELEMENT C
      BASIS_SET DZVP-MOLOPT-GTH
      POTENTIAL GTH-PBE-q4
    &END KIND
  &END SUBSYS
&END FORCE_EVAL

&MOTION
  &BAND
    BAND_TYPE CI-NEB
    NUMBER_OF_REPLICA 8
    K_SPRING 0.05
    &CONVERGENCE_CONTROL
      MAX_FORCE 0.0010
      RMS_FORCE 0.0050
    &END
    ROTATE_FRAMES TRUE
    ALIGN_FRAMES TRUE
    &CI_NEB
      NSTEPS_IT 2
    &END
    &OPTIMIZE_BAND
      OPT_TYPE DIIS
      OPTIMIZE_END_POINTS FALSE
      &DIIS
        MAX_STEPS 1000
      &END
    &END
    &PROGRAM_RUN_INFO
    &END
    &CONVERGENCE_INFO
    &END

    &REPLICA
      COORD_FILE_NAME ./ethane_s1.xyz
    &END
    &REPLICA
      COORD_FILE_NAME ./ethane_ts.xyz
    &END
    &REPLICA
      COORD_FILE_NAME ./ethane_s2.xyz
    &END
  &END BAND
&END MOTION

One notable difference to the previous input files is the specification of the periodic boundary conditions in the &POISSON section and in the &CELL (and the increased size of the box – now 12 Å instead of 10 Å). This is to use a different solver configuration which behaves better in combination with NEB and if the box is big enough it will not change the physics.

To run this simulation, we use a slightly different command which will return right away:

$ nohup mpirun -np 4 cp2k.popt -i ethane_neb_aba.inp -o ethane_neb_aba.out &

In the background, the process is still running, which you can verify by either watching the changes to the output file (exit this command with CTRL+c) using

$ tail -f ethane_neb_aba.out

or by looking at the list of your processes:

$ ps uxf

Note the use of mpirun -np 4 together with the cp2k.popt executable. This tells the machine to run it using the MPI parallelization system on 4 cores, which should speed up your calculations. The prefix nohup lets the command continue even if you log out and the ampersand & makes it run in the background. Remember that tcopt2 is shared among you and that there is only a limited number of cores. Before launching, you can use the command top to check if there are enough cores available for your calculation.

This may take a couple of hours. Continue with the exercises below once the calculation finishes.

Visualize the trajectory and plot the energy curve

When you take another peek at the input file you used to run the calculation, you will notice that we specified NUMBER_OF_REPLICA 8, which means that CP2K will generate in total 8 beads (3 we specified in the &REPLICA sections, 5 will be generated automatically by interpolation).

You should therefore find 8 files named ethane_neb_aba-pos-Replica_nr_1-1.xyz..ethane_neb_aba-pos-Replica_nr_8-1.xyz in your exercise directory, containing the optimization of each bead. To get the trajectory over the band, we extract the last frame (see the tip below) and write it into a separate file named ethane_neb_aba_8r.xyz:

$ for i in {1..8} ; do tail -n 10 "ethane_neb_aba-pos-Replica_nr_${i}-1.xyz" >> ethane_neb_aba_8r.xyz ; done

Look at the movement again using VMD.

The anatomy of a XYZ file:

A XYZ file consists of one or multiple blocks (frames) of the following:

       8  <-- the number of atoms
 i =       0, E =      -14.9559722838 <-- some comment, CP2K writes the number of the iteration and the resulting energy here
  C         6.7640435612        6.0000000003        5.9997401503  <-- the atomic symbol and position
  C         5.2359564388        5.9999999997        6.0002598497  <-- .. for the specified number of atoms
  H         4.8332682152        6.0000000001        7.0232923722
  H         4.8330445407        5.1142018851        5.4887808109
  H         4.8330445407        6.8857981148        5.4887808113
  H         7.1667317848        5.9999999999        4.9767076278
  H         7.1669554593        6.8857981149        6.5112191891
  H         7.1669554593        5.1142018852        6.5112191887

Extracting the last frame (the optimized geometry in case of a geometry optimization) is therefore simple once you know the number of atoms using the command tail, which can be used to get the last N lines of a file using the switch -n N.

In our case of 8 atoms this is:

$ tail -n 10 geo_opt_output.xyz

Since CP2K writes the energy in the comment line of each frame in the XYZ file (see tip above), we can extract the energy values for each bead directly from the newly generated ethane_neb_aba_8r.xyz:

awk '/E =/ {i=i+1; printf "%s %16.8f\n", i, $6}' ethane_neb_aba_8r.xyz

Create a plot for this energy curve.

Vibrational analysis

To verify whether the point at the highest energy is actually a transition state, we will be doing a vibrational analysis.

First identify the bead with the highest energy (see exercise above) and create a new XYZ file named ethane_neb_aba_TS.xyz with the respective coordinates (extracted from either the correct ethane_neb_aba-pos-Replica_nr_N-1.xyz file or the ethane_neb_aba_8r.xyz).

Use the following input file and the same command as above (with different input and output file names of course) to generate the analysis.

ethane_TS_va.inp
&GLOBAL
  PROJECT ethane_TS_va
  RUN_TYPE NORMAL_MODES
  PRINT_LEVEL MEDIUM
&END GLOBAL

&FORCE_EVAL
  METHOD Quickstep              ! Electronic structure method (DFT,...)
  &DFT
    BASIS_SET_FILE_NAME  BASIS_MOLOPT
    POTENTIAL_FILE_NAME  POTENTIAL

    &POISSON                    ! Solver requested for non periodic calculations
      PERIODIC XYZ
    &END POISSON
    &SCF                        ! Parameters controlling the convergence of the scf. This section should not be changed. 
      SCF_GUESS ATOMIC
      EPS_SCF 1.0E-7
      MAX_SCF 300
    &END SCF
    &XC                        ! Parametes needed to compute the electronic exchange potential 
      &XC_FUNCTIONAL PBE
      &END XC_FUNCTIONAL
    &END XC
  &END DFT

  &SUBSYS
    &CELL
      ABC 12. 12. 12.
      PERIODIC XYZ
    &END CELL
    &TOPOLOGY                    ! Section used to center the atomic coordinates in the given box. Useful for big molecules
      &CENTER_COORDINATES
      &END
      COORD_FILE_FORMAT xyz
      COORD_FILE_NAME  ./ethane_neb_aba_TS.xyz
    &END
    &KIND H
      ELEMENT H
      BASIS_SET DZVP-MOLOPT-GTH
      POTENTIAL GTH-PBE-q1
    &END KIND
    &KIND C
      ELEMENT C
      BASIS_SET DZVP-MOLOPT-GTH
      POTENTIAL GTH-PBE-q4
    &END KIND
  &END SUBSYS
&END FORCE_EVAL

&VIBRATIONAL_ANALYSIS
  NPROC_REP 1
  DX 0.01
  FULLY_PERIODIC
  &PRINT
    &MOLDEN_VIB
    &END
    &CARTESIAN_EIGS
    &END
    &PROGRAM_RUN_INFO
      &EACH
        REPLICA_EVAL 1
      &END
    &END
  &END
&END

Once this run completes, you can grep for the modes frequency with:

$ grep "VIB|Frequency" ethane_TS_va.out

The presence of a negative (imaginary) mode means that it is actually a transition state (and not stable). Are there any of that kind for this structure ? If so, what are the atomic displacements associated with it ?

Note that cp2k also produces a ethane_TS_va-VIBRATIONS-1.mol file. It is intended to be read by the Molden program, which makes the visualization of the modes possible.

exercises/2018_uzh_cmest/path_optimization_neb.txt · Last modified: 2020/08/21 10:15 by 127.0.0.1