Running COSMO on NWChem, gathering energies for reaction coordinates
Clicked A Few Times
1:06:09 PM PST - Sat, Mar 9th 2013
I'm trying to optimize transition state structures in NWChem for ethyl bromide with iodide and propyl bromide with iodide, but am encountering some problems. Using the default convergence criteria, I was able to find a valid transition state. I want to be able to use this transition state in MOPAC to find the complexes semi-emirically, then calculate the energies for the complexes back in NWChem. The transition states I've found for ethyl bromide propyl bromide have not been successful in being used to calculate the corresponding complexes, so I've tried reoptimizing them with TIGHT convergence criteria. Propyl bromide complex calculations still have not been successful, and ethyl bromide has not been able to converge yet. I've been increasing the max iterations for ethyl, but other than this, I'm not quite sure if what else to do, or if there is anything I can change about the transition optimizations for this purpose. Any suggestions?

Also, after calculating things in vacuum, I've been optimizing the structures using COSMO to show the solvation effect of the reaction happening in acetone. It has worked with 3-21G, but when I try to run the same TS optimization with 6-311G, the transition state is found, but the vibrational frequency calculation continues to fail for MP2, though I'm not sure how to improve this for frequency calculations. Would anyone have suggestions for this also?

Forum Vet
1:51:23 PM PDT - Mon, Mar 11th 2013
A little lost in your first paragraph:

1. You got transition states from NWChem

2. You put them in MOPAC to find other transition states? What do you mean by complexes?

3. You are using TIGHT to reoptimize the transition states in 1.?

4. What kind of calculation are you trying to do, SCF, DFT, MP2? For DFT it might be worth increasing the accuracy of the grid. For SCF and MP2 you may want to increase the integral accuracy.

For COSMO: What is the error you see for the MP2 frequencies?

Clicked A Few Times
7:56:22 AM PDT - Thu, Mar 14th 2013
In COSMO I'm getting errors in the frequency calculation of the transition state for MP2 6-311G* and for ethyl bromide with iodide for HF 3-21 G. The error says
"hess_anal: energy failure 555"
"Last System Error Message from Task 0:: Inappropriate ioctl for device
0:0:hess_anal: energy failure:: 555"

I'm not sure what either of these error messages mean in regards to how to edit the input file so that they stop occuring. I have tried altering the driver by increasing the maxiter and telling it to use tight convergence criteria. I did this because the output file said the error is most commonly associated with not reaching convergence criteria. However, I am not sure how to prevent the energy calculations from failing again.

Forum Vet
8:58:15 AM PDT - Thu, Mar 14th 2013
Can you post the input for the two cases. You need to increase the number of iterations in your SCF block, rather then increase the number of iterations in driver which only drives the gradient,

Bert

Clicked A Few Times
9:25:03 AM PDT - Thu, Mar 14th 2013
Yeah. here's the input for the ethyl bromide with iodide for HF 3-21G:
basis noprint
* library 3-21G
end

scf
noprint "final vectors analysis"

singlet
end

driver
gmax 0.000015

grms 0.00001

maxiter 10000
end

task scf energy

property
 dipole

 mulliken
end

COSMO
dielec 20.7

rsolv 3.08
end

task scf property

task scf freq

and here's the input for methyl bromide with iodide for MP2 6-311G*
basis noprint
* library 6-311G*
end

scf
noprint "final vectors analysis"

singlet
end

driver
gmax 0.000015

grms 0.00001

maxiter 2000
end

task mp2 energy

property
 dipole

 mulliken
end

COSMO
dielec 20.7

rsolv 3.08
end

task scf property

task mp2 freq

I'm not sure I understand what you mean by increasing the number of iterations in the SCF block.

Thank you for helping me

Forum Vet
10:51:14 AM PDT - Fri, Mar 15th 2013
The driver input block is for geometry optimization only. I.e., increasing the number of iterations there tells the code to run more geometry optimization steps. You're not doing a geometry optimization so this won't do you any good.

By default the scf energy calculation has a maximum of 20 (See: http://nwchemgit.github.io/index.php/Release61:Hartree-Fock_Theory_for_Molecules#MAXITER_--_... ). It sounds like the SCF energy calculation for each numerical hessian step is not converging.

Your input decks do not have a geometry so I cannot try to run your cases to check.

Bert

Clicked A Few Times
4:17:17 PM PDT - Fri, Mar 15th 2013
Oh thanks! I'll check that link out and see about adding iterations to the SCF energy calculations. I just realized I forgot to include the geometry in the input file. Here's the whole input file for one of my Ethylbromide with iodide using HF, 3-21G runs

title "Et TS HF COSMO 3-21G"

echo

charge -1

geometry
zmatrix
C
C 1 B1
H 2 B2 1 A1
H 2 B3 1 A2 3 D1
H 2 B4 1 A3 3 D2
Br 1 B5 2 A4 3 D3
H 1 B6 2 A5 3 D4
H 1 B7 2 A6 3 D5
I 1 B8 2 A7 3 D6
variables
B1 1.5113002
B2 1.0798152
A1 107.71964
B3 1.0799266
A2 110.52351
D1 119.62510
B4 1.0799274
A3 110.52139
D2 -119.62395
B5 2.7091183
A4 99.289855
D3 -0.0360276
B6 1.0610559
A5 120.75073
D4 84.757276
B7 1.0610509
A6 120.75305
D5 -84.828261
B8 2.7265751
A7 99.971164
D6 179.96836
end
end

basis noprint
* library 3-21G
end

scf
noprint "final vectors analysis"

singlet
end

driver
gmax 0.000015

grms 0.00001

maxiter 2000
end

task scf energy

property
 dipole

 mulliken
end

COSMO
dielec 20.7

rsolv 3.08
end

task scf property

task scf freq

Forum Vet
2:00:34 PM PDT - Mon, Mar 18th 2013
Test case converges with NWChem 6.1.1
I just ran your test case input unmodified with NWChem 6.1.1 and it completes successfully. So, I'm confused as to why you are having the problems you have. Are these clean runs (i.e. there are not left over files in scratch, etc.)?

Clicked A Few Times
1:46:55 PM PDT - Thu, Mar 21st 2013
I have gotten those jobs to run now, and have been having successful runs without problem since these jobs. I have run jobs in COSMO with ethyl bromide and iodide, and have been running jobs with propyl bromide and iodide. Recently I have run into an interesting problem with both my COSMO and gas phase vibrational frequency jobs. Unfortunately, while the jobs are failing, the output files look like the calculation just stopped. There is no error message displayed, so I'm not sure how to adjust the input files for the jobs to run either successfully, or far enough to reach an error message. Thoughts?

(I'm currently running a transition state optimization job to determine if this problems extends through all of my calculations or just through the frequency calculations)

Forum Vet
9:04:13 AM PDT - Fri, Mar 22nd 2013
No idea. The ethyl bromide + iodide above here in this posting runs to completion without a problem.

1. What is the error messages you are getting (either in the output file or error file).
2. What are the last set of lines in the output file?


Forum >> NWChem's corner >> Running NWChem