Switch between internal and cartesian coordinates

[Sibo Lin]

Dear NwChem community,

I want to find a transition state by first doing a minimization where an internal coordinate is set to be constant and fixed. Then I want to release that constraint and do a saddle optimization. Both steps use internal coordinates by default, but can I make NWChem switch to cartesian coordinates for the saddle optimization? I ask because I believe Cartesian optimization to be slower but more robust for certain transition states. Below is my current code, which does saddle optimization in internal coordinates. Should I switch to stepper instead of driver  for the cartesian saddlepoint optimization?

geometry noautosym

  load guess.xyz

  zcoord 

     bond 1 31 r constant

    end

end

task dft optimize

geometry adjust noautoz

  zcoord 

    bond 1 31 r

  end

end

task dft saddle

  

[Edoardo Aprà]

Here is a way to do the transition state search in cartesian coordinates after a geometry optimization step in internal coordinates.

After the initial geometry optimization, I do it again with the driver options maxiter=0 and xyz so that an xyz filed named prefix-0.xyz is generated.

Then the prefix-0.xyz file is loaded in combination with the noautoz keyword for the task saddle step.

start sad_switch

geometry
 O     0.00000000    0.00000000    0.11713500
 H     0.00000000   -0.77079100   -0.46854100
 H     0.00000000    0.77079100   -0.46854100
  zcoord
    bond 1 2 r constant
  end
end
basis; * library sto-3g ;end
dft; xc hfexch ;end
task dft optimize
driver
 xyz
 maxiter 0
end
task dft optimize ignore

geometry noautoz
  load sad_switch-000.xyz
end
driver
 clear
 maxiter 99
end
task dft saddle
task dft freq

[Sibo Lin]

Thank you Edo!  This works well for what I want to try.