modified on 17 May 2013 at 15:34 ••• 21,940 views

Release62:Nudged Elastic Band (NEB) and Zero Temperature String Methods

From NWChem

Jump to: navigation, search

Contents

Nudged Elastic Band (NEB) method

The NEB module is an implementation of the nudged elastic band (NEB) method of Jonsson et al., and it is one of two drivers in NWChem that can be used to perform minimum energy path optimizations. NEB can be used at all levels of theory, including SCF, HF, DFT, PSPW, BAND, MP2, RIMP2, CCSD, TCE.

Input to the NEB modules is contained with the NEB block

 NEB
  ...
 END

To run a NEB calculation the following the following task directives is used

TASK <theory> NEB
TASK <theory> NEB ignore

where <theory> is SCF, HF, DFT, PSPW, BAND, MP2, CCSD, TCE, etc.. The Task directive with the ignore option is recommended, otherwise NWChem will crash if the path is not optimized in the allowed maximum number of iterations.

Optional input for this module is specified within the compound directive,

 NEB
    NBEADS <integer nbeads default 5>
    KBEADS <float kbeads default 0.1>
    MAXITER <integer maxiter default 5>
    STEPSIZE <integer stepsize default 1.0>
    NHIST <integer nhist default 5>
    ALGORITHM <integer algorithm default 0>
    [loose | default | tight]
    GMAX <float gmax default 0.00045>
    GRMS <float grms default 0.00030>
    XMAX <float xmax default 0.00018>
    XMRS <float xmrs default 0.00012>
    [IMPOSE]
    [HASMIDDLE]
    [XYZ_PATH <string xyzfilename>]
    [RESET]
    [PRINT_SHIFT <integer print_shift default 0>]
 END

The following list describes the input for the NEB block

  • <nbeads> - number of beads (or images) used to represent the path
  • <kbeads> - value for the NEB spring constant
  • <maxiter> - maximum number of NEB path optimizations to be performed
  • <stepsize> - value for the stepsize used in the optimization. Typically less than 1.
  • <nhist> - number of histories to use for quasi-Newton optimization (algorithm =0)
  • LOOSE|DEFAULT|TIGHT - options specifying thresholds for convergence
  • <gmax> - value for the maximum gradient used to determine convergence
  • <grms> - value for the root mean square gradient used to determine convergence
  • <xmax> - value for the maximum cartesian step used to determine convergence
  • <xrmx> - value for the root mean square cartesian step used to determine convergence
  • <algorithm> - 0: quasi-Newton Fixed Point optimization, 1: dampled Verlet optimization, 2: refining conjugate gradient optimization
  • IMPOSE - if specified causes the initial geometries used to specify the path to be aligned with one another
  • HASMIDDLE - if specified causes the initial path to use the the "midgeom" geometry to be used as the midpoint, i.e. the initial path is defined as a linear morphing from "geometry" --> "midgeom" --> "endgeom"
  • XYZ_PATH - if specified the initial path is defined from the sequence of geometries contained in <xyzfilename>
  • RESET - if specified causes the NEB optimization and path to be started from scratch
  • <print_shift> - setting the PRINT_SHIFT directive causes the path energies and geometries to be outputed every <print_shift> steps. The current path energies are appended to the file jobname.neb_epath and the current geometries are appended to the file jobname.nebpath_"current iteration".xyz.

Setting up initial path

There are three different ways to define the initial path for NEB optimization.

  • Linear interpolation between two geometries

The geometries in the path are defined by

 \vec{R}^i_{xyz} = \vec{R}^1_{xyz} + \frac{i-1}{nbeads-1} (\vec{R}^{nbeads}_{xyz}-\vec{R}^{1}_{xyz}), \; for \;  i=1,nbeads

where the starting geometry ( \vec{R}^1_{xyz} ) is entered in the geometry block labeled "geometry", e.g.

geometry nocenter noautosym noautoz
O  0.00000000    -0.02293938     0.00000000
H  0.00000000     0.55046969     0.75406534
H  0.00000000     0.55046969    -0.75406534
end

and the last geometry in the path ( \vec{R}^{nbeads}_{xyz} ) in entered in the geometry block label "endgeom", e.g.

geometry endgeom nocenter noautosym noautoz
O  0.00000000     0.02293938     0.00000000
H  0.00000000    -0.55046969     0.75406534
H  0.00000000    -0.55046969    -0.75406534
end
  • Linear interpolation between three geometries

The geometries for this path are defined by

 \vec{R}^i_{xyz} = \vec{R}^1_{xyz} + \frac{i-1}{nbeads/2-1} (\vec{R}^{nbeads/2}_{xyz}-\vec{R}^{1}_{xyz}), \; for \;  i=1,nbeads/2

and

 \vec{R}^i_{xyz} = \vec{R}^{nbeads/2}_{xyz} + \frac{i-nbeads/2}{nbeads/2-1} (\vec{R}^{nbeads}_{xyz}-\vec{R}^{nbeads/2}_{xyz}), \; for \;  i=nbeads/2+1,nbeads

where the starting (), middle ( \vec{R}^{nbeads/2}_{xyz} ) and last ( \vec{R}^{nbeads}_{xyz} ) geometries are entered in the geometry blocks "geometry", "midgeom" and "endgeom" respectively, e.g.

geometry nocenter noautosym noautoz
O  0.00000000    -0.02293938     0.00000000
H  0.00000000     0.55046969     0.75406534
H  0.00000000     0.55046969    -0.75406534
end
geometry midgeom nocenter noautosym noautoz
O  0.00000000     0.00000000     0.00000000
H  0.00000000     0.00000000     1.00000000
H  0.00000000     0.00000000    -1.00000000
end
geometry endgeom nocenter noautosym noautoz
O  0.00000000     0.02293938     0.00000000
H  0.00000000    -0.55046969     0.75406534
H  0.00000000    -0.55046969    -0.75406534
end
  • Using xyz_path to explicitly input a path of geometries

The "xyz_path" option can also be used to define the initial path.

...
NEB
   ...
  XYZ_PATH path.xyz
END
...

where path.xyz contains a list of geometries in xyz format, e.g.

--------------- path.xyz ------------------
                   3
energy=  -17.107207699285738     
O           0.000000   -0.022939    0.000000
H           0.000000    0.550469    0.754065
H           0.000000    0.550469   -0.754065
                   3
energy=  -17.094903833074170     
O          -0.000003   -0.110080   -0.000000
H          -0.000000    0.273180    0.847029
H          -0.000000    0.273180   -0.847029
                   3
energy=  -17.063823686395292     
O          -0.000000   -0.000080   -0.000000
H           0.000000   -0.000002    0.941236
H           0.000000   -0.000002   -0.941236
                   3
energy=  -17.094944036147005     
O          -0.000000    0.110472   -0.000000
H          -0.000000   -0.273172    0.846957
H          -0.000000   -0.273172   -0.846957
                   3
energy=  -17.107208157343706     
O           0.000000    0.022939    0.000000
H           0.000000   -0.550469    0.754065
H           0.000000   -0.550469   -0.754065
--------------- path.xyz ------------------

Convergence criteria

The defaults may be used, or the directives LOOSE, DEFAULT, or TIGHT specified to use standard sets of values, or the individual criteria adjusted. All criteria are in atomic units. GMAX and GRMS control the maximum and root mean square gradient in the coordinates. XMAX and XRMS control the maximum and root mean square of the Cartesian step.

                LOOSE    DEFAULT    TIGHT
       GMAX   0.0045d0   0.00045   0.000015   
       GRMS   0.0030d0   0.00030   0.00001
       XMAX   0.0054d0   0.00180   0.00006
       XRMS   0.0036d0   0.00120   0.00004

NEB Tutorial 1: H2O Inversion

(input:Media:h2o-neb.nw, output:Media:h2o-neb.nwout, datafiles: Media:h2o-neb.neb_epath.dat Media:h2o-neb.neb_final_epath.dat )

(xyzfiles: Media: h2o-neb.nebpath_000001.xyz Media: h2o-neb.nebpath_000005.xyz Media: h2o-neb.nebpath_000010.xyz Media: h2o-neb.nebpath_000020.xyz Media: h2o-neb.nebpath_final.xyz )

File:H2o.gif
H2O Inversion Pathway
400px


Title "H2O inversion calculation"
echo
start h2o-neb
memory 1600 mb
permanent_dir ./perm2
scratch_dir      ./perm2
geometry nocenter noautosym noautoz
O  0.00000000    -0.02293938     0.00000000
H  0.00000000     0.55046969     0.75406534
H  0.00000000     0.55046969    -0.75406534
end
geometry endgeom nocenter noautosym noautoz
O  0.00000000     0.02293938     0.00000000
H  0.00000000    -0.55046969     0.75406534
H  0.00000000    -0.55046969    -0.75406534
end
#### Gaussian DFT ####
basis
* library 3-21G
end
dft
  xc b3lyp
  maxiter 5001
  cgmin
end
neb
  nbeads 10
  kbeads 1.0
  maxiter 10
  stepsize 0.10
  print_shift 1
end
task dft neb ignore
neb
  # increase the number of images
  nbeads 20
  kbeads 1.0
  stepsize 1.0
  maxiter 30
  loose
end
task dft neb ignore

After each optimization step the path energies are outputed as follows

neb: Path Energy #                    9
neb:                     1  -75.970000166349976     
neb:                     2  -75.973958450556779     
neb:                     3  -75.973964391052448     
neb:                     4  -75.973965560274110     
neb:                     5  -75.973961077512683     
neb:                     6  -75.973087554095144     
neb:                     7  -75.965847261117744     
neb:                     8  -75.950292780255126     
neb:                     9  -75.932932759963109     
neb:                    10  -75.921912278179292     
neb:                    11  -75.921834552460439     
neb:                    12  -75.932680002200939     
neb:                    13  -75.949868818688529     
neb:                    14  -75.965372754426866     
neb:                    15  -75.972788885848303     
neb:                    16  -75.973958649400714     
neb:                    17  -75.973965255113598     
neb:                    18  -75.973964962774133     
neb:                    19  -75.973959526041568     
neb:                    20  -75.970000163960066     


Another way to keep track of the optimization process is to run the following grep command on the output file.

[WE24397:NWChem/NEB/Example2] bylaska% grep @ h2o-neb.nwout
@neb  
@neb NEB Method
@neb algorithm      =         0
@neb maxiter        =        10
@neb nbeads         =        10
@neb nhist          =         5
@neb natoms         =         3
@neb stepsize       = 0.100E+01
@neb trust          = 0.100E+00
@neb kbeads         = 0.100E+00
@neb Gmax tolerance = 0.450E-03
@neb Grms tolerance = 0.300E-03
@neb Xmax tolerance = 0.180E-03
@neb Xrms tolerance = 0.120E-03
@neb  
@neb Step    Intrinsic E    Mid-Point E      Minimum E      Maximum E   Gmax     Grms     Xrms     Xmax   Walltime
@neb ---- -------------- -------------- -------------- -------------- -------- -------- -------- -------- --------
@neb    1     -75.951572     -75.921109     -75.970632     -75.921109  0.55875  0.01606  0.14221  1.54029    454.9
@neb    2     -75.953755     -75.923180     -75.972590     -75.923177  0.38930  0.01116  0.01588  0.45644    624.4
@neb    3     -75.956726     -75.924391     -75.972861     -75.924387  0.25587  0.00961  0.03673  0.83118    805.2
@neb    4     -75.957861     -75.924279     -75.973059     -75.924275  0.23572  0.00894  0.01793  0.24399    971.8
@neb    5     -75.959613     -75.925045     -75.973869     -75.925036  0.10257  0.00464  0.03197  0.20350   1152.8
@neb    6     -75.959964     -75.925503     -75.973957     -75.925486  0.04762  0.00196  0.00905  0.10433   1316.4
@neb    7     -75.960068     -75.925822     -75.973956     -75.925791  0.03897  0.00141  0.00308  0.04432   1519.9
@neb    8     -75.960091     -75.925914     -75.973959     -75.925877  0.03707  0.00127  0.00070  0.01691   2055.8
@neb    9     -75.960129     -75.926078     -75.973962     -75.926028  0.03353  0.00108  0.00127  0.03707   2297.2
@neb   10     -75.960142     -75.926142     -75.973963     -75.926085  0.03199  0.00101  0.00054  0.00420   2756.6
@neb   NEB calculation not converged
@neb  
@neb NEB Method
@neb algorithm      =         0
@neb maxiter        =        30
@neb nbeads         =        20
@neb nhist          =         5
@neb natoms         =         3
@neb stepsize       = 0.100E+01
@neb trust          = 0.100E+00
@neb kbeads         = 0.100E+01
@neb Gmax tolerance = 0.450E-02
@neb Grms tolerance = 0.300E-02
@neb Xmax tolerance = 0.540E-02
@neb Xrms tolerance = 0.360E-02
@neb  
@neb Step    Intrinsic E    Mid-Point E      Minimum E      Maximum E   Gmax     Grms     Xrms     Xmax   Walltime
@neb ---- -------------- -------------- -------------- -------------- -------- -------- -------- -------- --------
@neb    1     -75.960225     -75.921704     -75.973965     -75.921669  0.24799  0.00398  0.00272  0.08741   3966.5
@neb    2     -75.960339     -75.921782     -75.973965     -75.921745  0.24794  0.00328  0.00199  0.12148   5023.2
@neb    3     -75.960424     -75.921742     -75.973965     -75.921701  0.19390  0.00286  0.00164  0.08342   5741.4
@neb    4     -75.960494     -75.921849     -75.973965     -75.921804  0.19681  0.00266  0.00143  0.09030   6079.7
@neb    5     -75.960646     -75.921874     -75.973965     -75.921820  0.17459  0.00240  0.00241  0.22047   6751.5
@neb    6     -75.960674     -75.921856     -75.973965     -75.921797  0.14246  0.00165  0.00060  0.00256   7572.3
@neb    7     -75.960724     -75.921884     -75.973966     -75.921817  0.13004  0.00153  0.00082  0.05401   7893.3
@neb    8     -75.960747     -75.921892     -75.973966     -75.921822  0.12809  0.00149  0.00038  0.00237   8631.2
@neb    9     -75.960792     -75.921912     -75.973966     -75.921835  0.12267  0.00142  0.00075  0.05081   9222.0
@neb   10     -75.960813     -75.921923     -75.973966     -75.921841  0.11902  0.00138  0.00035  0.00212  10163.2
@neb   11     -75.960834     -75.921934     -75.973966     -75.921846  0.11569  0.00135  0.00035  0.00203  10478.3
@neb   12     -75.961060     -75.922060     -75.973966     -75.921889  0.07709  0.00104  0.00365  0.30944  10863.8
@neb   13     -75.961255     -75.922186     -75.973966     -75.921919  0.04600  0.00087  0.00309  0.19999  11357.0
@neb   14     -75.961405     -75.922286     -75.973966     -75.921927  0.03549  0.00079  0.00244  0.03857  11860.0
@neb  NEB calculation converged

Zero Temperature String Method

The STRING module is an implementation of the zero temperature string method of vanden Eijden et al., and it is one of two drivers in NWChem that can be used to perform minimum energy path optimizations. STRING can be used at all levels of theory, including SCF, HF, DFT, PSPW, BAND, MP2, RIMP2, CCSD, TCE.

Input to the STRING module is contained with the STRING block

 STRING
  ...
 END

To run a STRING calculation the following the following task directives is used

TASK <theory> STRING
TASK <theory> STRING ignore

where <theory> is SCF, HF, DFT, PSPW, BAND, MP2, CCSD, TCE, etc.. The Task directive with the ignore option is recommended, otherwise NWChem will crash if the path is not optimized in the allowed maximum number of iterations.

Optional input for this module is specified within the compound directive,


 STRING
    NBEADS <integer nbeads default 5>
    MAXITER <integer maxiter default 5>
    STEPSIZE <integer stepsize default 1.0>
    NHIST <integer nhist default 5>
    INTERPOL <integer algorithm default 1>
    FREEZE1 <logical freeze1 default .false.>
    FREEZEN <logical freezen default .false.>
    TOL <float tol default 0.00045>
    [IMPOSE]
    [HASMIDDLE]
    [XYZ_PATH <string xyzfilename>]
    [RESET]
    PRINT_SHIFT <integer print_shift default 0>
 END

The following list describes the input for the NEB block

  • <nbeads> - number of beads (or images) used to represent the path
  • <maxiter> - maximum number of NEB path optimizations to be performed
  • <stepsize> - value for the stepsize used in the optimization. Typically less than 1.
  • <nhist> - number of histories to use for quasi-Newton optimization (algorithm =0)
  • <tol> - value for the maximum gradient used to determine convergence
  • <freeze1> - .true.: first bead of simulation frozen, .false.:first bead of simulation not frozen.
  • <freezen> - .true.:last bead of simulation frozen, .false.:last bead of simulation not frozen
  • <interpol> - 1: linear, 2: spline, 3: Akima spline
  • IMPOSE - if specified causes the initial geometries used to specify the path to be aligned with one another
  • HASMIDDLE - if specified causes the initial path to use the the "midgeom" geometry to be used as the midpoint, i.e. the initial path is defined as a linear morphing from "geometry" --> "midgeom" --> "endgeom"
  • XYZ_PATH - if specified the initial path is defined from the sequence of geometries contained in <xyzfilename>
  • RESET - if specified causes the NEB optimization and path to be started from scratch
  • <print_shift> - setting the PRINT_SHIFT directive causes the path energies and geometries to be outputed every <print_shift> steps. The current path energies are appended to the file jobname.neb_epath and the current geometries are appended to the file jobname.nebpath_"current iteration".xyz.


Setting up the initial path

There are three different ways to define the initial path for NEB optimization.

  • Linear interpolation between two geometries

The geometries in the path are defined by

 \vec{R}^i_{xyz} = \vec{R}^1_{xyz} + \frac{i-1}{nbeads-1} (\vec{R}^{nbeads}_{xyz}-\vec{R}^{1}_{xyz}), \; for \;  i=1,nbeads

where the starting geometry ( \vec{R}^1_{xyz} ) is entered in the geometry block labeled "geometry", e.g.

geometry nocenter noautosym noautoz
O  0.00000000    -0.02293938     0.00000000
H  0.00000000     0.55046969     0.75406534
H  0.00000000     0.55046969    -0.75406534
end

and the last geometry in the path ( \vec{R}^{nbeads}_{xyz} ) in entered in the geometry block label "endgeom", e.g.

geometry endgeom nocenter noautosym noautoz
O  0.00000000     0.02293938     0.00000000
H  0.00000000    -0.55046969     0.75406534
H  0.00000000    -0.55046969    -0.75406534
end
  • Linear interpolation between three geometries

The geometries for this path are defined by

 \vec{R}^i_{xyz} = \vec{R}^1_{xyz} + \frac{i-1}{nbeads/2-1} (\vec{R}^{nbeads/2}_{xyz}-\vec{R}^{1}_{xyz}), \; for \;  i=1,nbeads/2

and

 \vec{R}^i_{xyz} = \vec{R}^{nbeads/2}_{xyz} + \frac{i-nbeads/2}{nbeads/2-1} (\vec{R}^{nbeads}_{xyz}-\vec{R}^{nbeads/2}_{xyz}), \; for \;  i=nbeads/2+1,nbeads

where the starting (), middle ( \vec{R}^{nbeads/2}_{xyz} ) and last ( \vec{R}^{nbeads}_{xyz} ) geometries are entered in the geometry blocks "geometry", "midgeom" and "endgeom" respectively, e.g.

geometry nocenter noautosym noautoz
O  0.00000000    -0.02293938     0.00000000
H  0.00000000     0.55046969     0.75406534
H  0.00000000     0.55046969    -0.75406534
end
geometry midgeom nocenter noautosym noautoz
O  0.00000000     0.00000000     0.00000000
H  0.00000000     0.00000000     1.00000000
H  0.00000000     0.00000000    -1.00000000
end
geometry endgeom nocenter noautosym noautoz
O  0.00000000     0.02293938     0.00000000
H  0.00000000    -0.55046969     0.75406534
H  0.00000000    -0.55046969    -0.75406534
end
  • Using xyz_path to explicitly input a path of geometries

The "xyz_path" option can also be used to define the initial path, e.g.

...
STRING
   ...
  XYZ_PATH path.xyz
END
...

String Tutorial 1:HCN --> HNC path optimization

(input:Media:HCN-string1.nw, output:Media:HCN-string1.nwout, datafiles: Media:HCN-string1.string_epath.dat Media:HCN-string1.string_final_epath.dat )

(xyzfiles: Media: HCN-string1.stringpath_000001.xyz Media: HCN-string1.stringpath_000005.xyz Media: HCN-string1.stringpath_000010.xyz Media: HCN-string1.stringpath_000020.xyz Media: HCN-string1.stringpath_000030.xyz Media: HCN-string1.stringpath_final.xyz )

HCN-->HNC Reaction Pathway
Hcn-hnc-dft.png

In this example, the path energy for the reaction HCN --> HNC is calculated.

#  
# The initial path has the Carbon moving through the Nitrogen.
# So for this simulation to work that atom avoidance code needs to work.
# Because the initial path is so stiff the wavefunction optimizer needs to requires
# lots of iterations during the early stages of the path optimization.
#
#
Title "HCN --> HNC Zero-Temperature String Simulation"
echo
start hcn-hnc-dft
permanent_dir ./perm
scratch_dir ./perm
geometry noautoz noautosym
C         0.00000000     0.00000000    -0.49484657
N         0.00000000     0.00000000     0.64616359
H         0.00000000     0.00000000    -1.56151539
end
geometry endgeom  noautoz noautosym
C         0.00000000     0.00000000     0.73225318 
N         0.00000000     0.00000000    -0.42552059
H         0.00000000     0.00000000    -1.42351006
end
#### Gaussian DFT ####
basis
* library 3-21G
end
dft
  xc b3lyp
  maxiter 501
end
string
  nhist 10
  nbeads 10
  maxiter 10
  stepsize 0.10
  print_shift 1
  # don't allow the end points of the path to move
  freeze1 .true.
  freezeN .true.
end
task dft string ignore
string
  # increase the number of images
  nbeads 20
  maxiter 20
  # allow the end points of the path to move
  freeze1 .false.
  freezeN .false.
end
task dft string ignore


After each optimization step the path energies are outputed as follows

 string: Path Energy #                    2
 string:                     1  -92.906682492969779
 string:                     2  -92.743446565848473
 string:                     3  -92.751945829987775
 string:                     4  -92.756507971834026
 string:                     5  -92.726984154346979
 string:                     6  -92.701651474021503
 string:                     7  -92.672613497521183
 string:                     8  -92.825096796032099
 string:                     9  -92.716422030970662
 string:                    10  -92.881713271394148

Another way to keep track of the optimization process is to run the following grep command on the output file.

[WE24397:NWChem/NEB/Example2] bylaska% grep @ HCN-dft.out
@zts
@zts String method.
@zts Temperature          =   0.00000
@zts Covergence Tolerance =   0.00010
@zts Step Size            =   0.10000
@zts Maximum Time Steps   =        10
@zts Number of replicas   =        10
@zts Number of histories  =        10
@zts String Interpolator  =         1
@zts First Replica        = frozen
@zts Last Replica         = frozen
@zts
@zts  Step     xrms     xmax        E start       E middle          E end          E max      E average
@zts     1 0.460700 2.602234    -92.9066825    -83.4767173    -92.8817133    -83.4767173    -91.6169775
@zts     2 0.862226 5.405612    -92.9066825    -92.3028437    -92.8817133    -92.3028437    -92.6631831
@zts     3 0.105285 0.530157    -92.9066825    -92.3289676    -92.8817133    -92.3289676    -92.6702949
@zts     4 0.134687 0.740991    -92.9066825    -92.3512584    -92.8817133    -92.3512584    -92.6821949
@zts     5 0.117113 0.916210    -92.9066825    -92.3767826    -92.8817133    -92.3767826    -92.6899234
@zts     6 0.124464 0.844439    -92.9066825    -92.4195957    -92.8817133    -92.4195957    -92.7045117
@zts     7 0.092105 0.731434    -92.9066825    -92.4510785    -92.8817133    -92.4510785    -92.7156403
@zts     8 0.049227 0.330651    -92.9066825    -92.4690983    -92.8817133    -92.4690983    -92.7288274
@zts     9 0.032819 0.177356    -92.9066825    -92.4827444    -92.8817133    -92.4827444    -92.7344806
@zts    10 0.076249 0.444246    -92.9066825    -92.4930430    -92.8817133    -92.4930430    -92.7381477
@zts The string calculation failed to converge
@zts Bead number    1  Potential Energy =     -92.906682487840
@zts Bead number    2  Potential Energy =     -92.850640135623
@zts Bead number    3  Potential Energy =     -92.819370566454
@zts Bead number    4  Potential Energy =     -92.680821335407
@zts Bead number    5  Potential Energy =     -92.505231918657
@zts Bead number    6  Potential Energy =     -92.493042984646
@zts Bead number    7  Potential Energy =     -92.637367419044
@zts Bead number    8  Potential Energy =     -92.775376312982
@zts Bead number    9  Potential Energy =     -92.831230727986
@zts Bead number   10  Potential Energy =     -92.881713271394
@zts
@zts String method.
@zts Temperature          =   0.00000
@zts Covergence Tolerance =   0.00010
@zts Step Size            =   0.10000
@zts Maximum Time Steps   =        20
@zts Number of replicas   =        20
@zts Number of histories  =        10
@zts String Interpolator  =         1
@zts First Replica        = moves
@zts Last Replica         = moves
@zts
@zts  Step     xrms     xmax        E start       E middle          E end          E max      E average
@zts     1 1.039809 5.039486    -92.9071472    -92.4998400    -92.8820628    -92.4998400    -92.7500136
@zts     2 0.192562 0.999019    -92.9073958    -92.5259828    -92.8821500    -92.5259828    -92.7624061
@zts     3 0.244943 1.236459    -92.9075306    -92.5735140    -92.8821223    -92.5735140    -92.7816692
@zts     4 0.207031 1.093667    -92.9075888    -92.6229190    -92.8821177    -92.6154678    -92.7979112
@zts     5 0.056648 0.293829    -92.9075975    -92.6672565    -92.8821033    -92.6507897    -92.8101666
@zts     6 0.078950 0.555245    -92.9076044    -92.7245122    -92.8822536    -92.7014407    -92.8241914
@zts     7 0.065564 0.521110    -92.9076101    -92.7539982    -92.8822915    -92.7376310    -92.8326007
@zts     8 0.050188 0.319477    -92.9076113    -92.7695725    -92.8824219    -92.7612604    -92.8378464
@zts     9 0.055301 0.322130    -92.9076168    -92.7754581    -92.8825732    -92.7740099    -92.8408900
@zts    10 0.038769 0.195102    -92.9076177    -92.7775695    -92.8826652    -92.7775695    -92.8425440
@zts    11 0.064900 0.273480    -92.9076215    -92.7800330    -92.8827175    -92.7800330    -92.8443574
@zts    12 0.062593 0.266337    -92.9076224    -92.7823972    -92.8826993    -92.7823972    -92.8458976
@zts    13 0.205437 0.948190    -92.9076243    -92.7842034    -92.8826408    -92.7842034    -92.8469810
@zts    14 0.015025 0.068924    -92.9076247    -92.7844362    -92.8826536    -92.7844362    -92.8472227
@zts    15 0.129208 0.602636    -92.9076254    -92.7849856    -92.8826676    -92.7849856    -92.8477169
@zts    16 0.013479 0.056561    -92.9076260    -92.7855201    -92.8826783    -92.7855201    -92.8481626
@zts    17 0.472858 2.220715    -92.9076271    -92.7878088    -92.8826913    -92.7878088    -92.8497919
@zts    18 0.162617 0.766201    -92.9076273    -92.7879912    -92.8826934    -92.7879912    -92.8499197
@zts    19 0.013204 0.060562    -92.9076276    -92.7885097    -92.8826994    -92.7885097    -92.8502675
@zts    20 0.718205 3.423813    -92.9076278    -92.7905066    -92.8827009    -92.7895258    -92.8514863
@zts The string calculation failed to converge
@zts Bead number    1  Potential Energy =     -92.907627751439
@zts Bead number    2  Potential Energy =     -92.905047596626
@zts Bead number    3  Potential Energy =     -92.897944354806
@zts Bead number    4  Potential Energy =     -92.887494117302
@zts Bead number    5  Potential Energy =     -92.874059841858
@zts Bead number    6  Potential Energy =     -92.857382758537
@zts Bead number    7  Potential Energy =     -92.837207959079
@zts Bead number    8  Potential Energy =     -92.815902497386
@zts Bead number    9  Potential Energy =     -92.798474907121
@zts Bead number   10  Potential Energy =     -92.789525765222
@zts Bead number   11  Potential Energy =     -92.790506632257
@zts Bead number   12  Potential Energy =     -92.799861168980
@zts Bead number   13  Potential Energy =     -92.814252430183
@zts Bead number   14  Potential Energy =     -92.830704548760
@zts Bead number   15  Potential Energy =     -92.847248091296
@zts Bead number   16  Potential Energy =     -92.861557132126
@zts Bead number   17  Potential Energy =     -92.871838446832
@zts Bead number   18  Potential Energy =     -92.878543965696
@zts Bead number   19  Potential Energy =     -92.881844751735
@zts Bead number   20  Potential Energy =     -92.882700859222

A plotting program (e.g. gnuplot, xmgrace) can be used to look at final path as well as the the convergence of the path i.e.,

[WE24397:NEB/Example2/perm] bylaska% gnuplot
	G N U P L O T
	Version 4.6 patchlevel 0    last modified 2012-03-04 
	Build System: Darwin x86_64
	Copyright (C) 1986-1993, 1998, 2004, 2007-2012
	Thomas Williams, Colin Kelley and many others
	gnuplot home:     http://www.gnuplot.info
	faq, bugs, etc:   type "help FAQ"
	immediate help:   type "help"  (plot window: hit 'h')
Terminal type set to 'aqua'
gnuplot> set xlabel "Reaction Coordinate"
gnuplot> set ylabel "Energy (kcal/mol)"
gnuplot> set yrange [0:100]
gnuplot> set grid
gnuplot> set style data linespoints
gnuplot> plot "hcn-hnc-dft.string_epath" using 1:($2+92.908)*27.2116*23.06,"hcn-hnc-dft.string_final_epath" using 1:($2+92.908)*27.2116*23.06
gnuplot> 

String Tutorial 2:

Title "2SiO4H4 --> H3O3Si-O-SiO3H3 + H2O"
echo
start sio4h4-dimer
memory 1800 mb
permanent_dir ./perm
scratch_dir   ./perm
geometry noautoz noautosym
Si        -3.90592       -0.11789        0.03791 
O         -2.32450       -0.24327       -0.05259
O         -4.45956       -1.13247        1.13159
O         -4.53584       -0.45118       -1.38472
O         -4.28179        1.37363        0.44838
Si         1.27960        0.06912        0.14555
O          2.85122        0.23514        0.32761
O          0.54278        0.38513        1.52092
O          0.94484       -1.42248       -0.29913
O          0.75605        1.07390       -0.97272
H         -1.66762       -0.74425       -0.29362
H         -4.05734        2.06481        0.90983
H         -4.30983       -1.85807        1.57116
H         -4.43621       -0.88060       -2.12508
H          3.59374       -0.16315        0.50572
H          0.36896        0.10990        2.31839
H          0.53993       -2.15495       -0.09488
H          0.43207        1.85525       -1.13531
end
geometry endgeom  noautoz noautosym
Si        -3.07373        0.18232       -0.24945
O         -1.50797        0.23823       -0.53062
O         -3.36758       -0.93058        0.85023
O         -3.83958       -0.20093       -1.59101
O         -3.57993        1.59735        0.27471
Si        -0.05186        0.25441        0.11277
O          0.94679       -0.58168       -0.80206
O         -0.10091       -0.40972        1.55838
O          1.41035       -3.75872        1.22931
O          0.47135        1.75206        0.24209
H          1.03624       -4.62405        0.92620
H         -3.81554        2.06192        0.96069
H         -3.97094       -1.38510        1.26383
H         -4.39754       -0.73964       -1.96563
H          1.45990       -0.57144       -1.49361
H         -0.44444       -0.37536        2.34765
H          2.15751       -4.00850        1.82933
H          0.77180        2.44229       -0.17616
end
nwpw
  simulation_cell
     SC 18.0
  end
  cutoff 30.0
  lmbfgs
end
string
  nhist 10
  nbeads 10
  maxiter 10
  stepsize 0.10
  print_shift 1
  # don't allow the end points of the path to move
  freeze1 .true.
  freezeN .true.
end
task pspw string ignore
string
  # increase the number of images
  nbeads 20
  maxiter 20
  # allow the end points of the path to move
  freeze1 .false.
  freezeN .false.
end
task pspw string ignore

String Tutorial 3: Combining NEB and String path optimizations