Input Block 'String'
Clicked A Few Times
4:16:22 PM PST - Tue, Jan 15th 2013
Hello all,

I noticed in the QA/test directory there's h2o_zts and subdirectories with similar names (e.g. ch4_zts). By looking at those input files, I guess it's monitoring 'transition' and I have a couple of questions:

1. Can this be used to search transition state structure?

2. Where can I find more documentation about this String block? I didn't find much from the 'Documentation' link from this site.

Thank you!

Best,
Yue

Forum Vet
6:11:23 PM PST - Tue, Jan 15th 2013
String
1. Yes, it can be used to search for transition states, just like NEB.

2. This is a capability under development, and we are developing the documentation for this functionality. I'll have the person responsible for this develop provide more input.

Bert

Gets Around
9:28:12 AM PST - Wed, Jan 16th 2013
Which version of NWChem are you using? The input to the NEB and string codes has changed quite a bit in the last few months, so the input will be highly dependent on what version you're using.

Gets Around
9:30:24 AM PST - Wed, Jan 16th 2013
The basic input for NEB looks like

echo
start ch4_neb_dat

permanent_dir ./permh
scratch_dir ./permh


basis
 * library 3-21G
end



scf
end


neb
  maxiter 200

  nhist 15

  xyz_path ./jjj.xyz

  print_shift 1
end

task scf neb ignore

Gets Around
9:36:44 AM PST - Wed, Jan 16th 2013
you can also enter extra geometries, i.e. middle and end geometries, to define the path. The labels of these geometries are "midgeom", "endgeom", e.g.

This run never really converges well because the two H's that are doing all the real work pop off to form H2 and do a little dance and then come back to the CH2 group they left behind. Optimizing the details of the little dance is about as likely has finding blinker fluid at the auto parts store

Note that this sends H1 right through the Carbon - atom avoidance code must work



echo

start ch4_string

permanent_dir ./permh
scratch_dir ./permh

geometry nocenter
 symmetry c1
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 1.089000
H 1.026719 0.000000 -0.363000
H -0.513360 -0.889165 -0.363000
H -0.513360 0.889165 -0.363000
end

geometry endgeom nocenter
 symmetry c1
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 -1.089000
H 1.026719 0.000000 0.363000
H -0.513360 -0.889165 0.363000
H -0.513360 0.889165 0.363000
end

basis
 * library STO-3G
end

scf
  maxiter 5001
end


string
  #tol  0.004d0

  nhist 10

  stepsize 1.00d0

  nbeads  10

  maxiter 100

  interpol 1

  freeze1 .true.

  freezeN .true.
end
task scf string ignore
task scf string ignore

string
   nbeads 20
end
task scf string ignore

Gets Around
9:38:39 AM PST - Wed, Jan 16th 2013
You can also go back and forth between neb and string calculations, e.g.

This run never really converges well because the two H's that are doing all the real work pop off to form H2 and do a little dance and then come back to the CH2 group they left behind. Optimizing the details of the little dance is about as likely has finding blinker fluid at the auto parts store

Note that this sends H1 right through the Carbon - atom avoidance code must work


echo
start ch4_neb

permanent_dir ./permh
scratch_dir ./permh

geometry nocenter
 symmetry c1
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 1.089000
H 1.026719 0.000000 -0.363000
H -0.513360 -0.889165 -0.363000
H -0.513360 0.889165 -0.363000
end

geometry endgeom nocenter
 symmetry c1
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 -1.089000
H 1.026719 0.000000 0.363000
H -0.513360 -0.889165 0.363000
H -0.513360 0.889165 0.363000
end

basis
 * library STO-3G
end

scf
  thresh 1.0d-10

  maxiter 5001
end

neb
  impose

  nhist 10

  nbeads 10

  maxiter 100

  stepsize 0.10d0

  kbeads 0.1
end
task scf neb ignore

neb
  kbeads 1.0

  nbeads 20
end
task scf neb ignore

string
 nbeads 20

 nhist 10

 maxiter 100

 stepsize 1.0

 freeze1 .true.

 freezeN .true.

 interpol 1
end

task scf string ignore

Clicked A Few Times
1:48:46 PM PST - Wed, Jan 16th 2013
I'm running NWChem 6.1.


Quote:Bylaska Jan 16th 4:28 pm
Which version of NWChem are you using? The input to the NEB and string codes has changed quite a bit in the last few months, so the input will be highly dependent on what version you're using.

Gets Around
10:00:10 AM PST - Fri, Jan 18th 2013
The input I posted is in the development tree and will part of the next release. You can download a development snapshot and use it, or you can use the input in the 6.1 QA. The older version of the string code doesn't converge very quickly, but it should work.

Clicked A Few Times
5:11:56 PM PST - Wed, Jan 23rd 2013
can you explain a bit what those keywords in the string section mean? I guess 'nbeads' is the number of optimization points one wants, and maxiter is easy to understand. Beyond that, what is nhist? How to choose proper stepsize? What is interpol doing? What do freeze1 and freezeN do?

Thank you!


Quote:Bylaska Jan 16th 4:36 pm
you can also enter extra geometries, i.e. middle and end geometries, to define the path. The labels of these geometries are "midgeom", "endgeom", e.g.

This run never really converges well because the two H's that are doing all the real work pop off to form H2 and do a little dance and then come back to the CH2 group they left behind. Optimizing the details of the little dance is about as likely has finding blinker fluid at the auto parts store

Note that this sends H1 right through the Carbon - atom avoidance code must work



echo

start ch4_string

permanent_dir ./permh
scratch_dir ./permh

geometry nocenter
 symmetry c1
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 1.089000
H 1.026719 0.000000 -0.363000
H -0.513360 -0.889165 -0.363000
H -0.513360 0.889165 -0.363000
end

geometry endgeom nocenter
 symmetry c1
C 0.000000 0.000000 0.000000
H 0.000000 0.000000 -1.089000
H 1.026719 0.000000 0.363000
H -0.513360 -0.889165 0.363000
H -0.513360 0.889165 0.363000
end

basis
 * library STO-3G
end

scf
  maxiter 5001
end


string
  #tol  0.004d0

  nhist 10

  stepsize 1.00d0

  nbeads  10

  maxiter 100

  interpol 1

  freeze1 .true.

  freezeN .true.
end
task scf string ignore
task scf string ignore

string
   nbeads 20
end
task scf string ignore

Clicked A Few Times
5:46:56 PM PST - Wed, Jan 23rd 2013
Are NEB function (for pure QM instead of QM/MM) and input also changed a lot from 6.1 to 6.2? Can you also explain a bit what 'impose', 'nhist', 'stepsize' and 'kbeads' do? Thank you!

Quote:Bylaska Jan 18th 5:00 pm
The input I posted is in the development tree and will part of the next release. You can download a development snapshot and use it, or you can use the input in the 6.1 QA. The older version of the string code doesn't converge very quickly, but it should work.

Gets Around
10:07:04 PM PDT - Tue, Mar 12th 2013
Hi, Bylaska Jan.
How I can add extra geometries, i.e. middle and end geometries, to NEB task?
Can I add an initial MO vectors to geometries, in NEB or STRING tasks?
My version of NWCHEM is 6.1.1. Thank you!


Forum >> NWChem's corner >> Running NWChem