Recent Posts

Pages: [1] 2 3 ... 10
1
News and Announcements / TURBOMOLE 7.5.1 released
« Last post by uwe on March 02, 2021, 04:14:46 PM »
TURBOMOLE V7.5.1 has been released (January 2021)

New features
Fixes and enhancements
  • Check added for outlying charge correction when using COSMO FINE cavity (Isorad) to ensure stability of the algorithm.

  • Vibrational frequencies for unrestricted DFT calculations using MGGA functionals with the libxc library re-enabled.

  • Enhance relativistic calculations with general-contracted bases


2
Define / Re: heavy atoms without ecp
« Last post by Scoops on March 02, 2021, 12:31:33 PM »
Hi,

I'm having an issue with running all electron calculations too (on some small actinide molecules). I have removed the ecp from the control file, but having a problem with using both eht and hcore, with the following message coming up after reading the basis sets, when using hcore:

 symmetry group of the molecule :   c2v

 the group has the following generators :
   c2(z)
   mirror plane sigma(xz)

    4 symmetry operations found

 there are 4 real representations :   a1   a2   b1   b2

 maximum number of shells which are related by symmetry :  2

  missing closed shell occupation number declaration 

****
I haven't tried manually specifying the occupation numbers, would you recommend giving that a go? If so any advice on how to go about it? 

Thanks in advance

Sophie
3
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by uwe on February 24, 2021, 08:17:33 PM »
Hi,

just as a check: Distort the molecule along the vibrational mode with imaginary frequency and generate a series with different amplitudes (use TmoleX to do that or the vibration tool on the command line, run 'vibration -help' first to see the options and then use the -vibro option). Then, for each of the generated structures, run a single-point calculation and plot the energy (or excited state energy) vs the distortion. It will be very flat, but perhaps you can see whether it is a minimum or a maximum.

Regards,
Uwe
4
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by antti_karttunen on February 24, 2021, 05:08:55 PM »
Hi,

I'm not aware of any possibility to use 5 points instead of 3 in the numerical differentiation. The ruecker binary that generates the displacements only takes in $delta, $type (central/polyedr), $dimension (small, norm, big - this is a mystery to me!), and $freezeoption (nofreeze).

I agree that this sounds like a numerical artifact, the imaginary frequency is so small. From NumForce side, I only can think of two possibilities:
1) Using smaller or larger displacement (I guess you already tried these)
2) Using polyedr (if you so far used only central).

From ricc2 side, I don't know if increasing nexc could little bit affect the numerics. I guess you would see much larger imaginary frequency if this would be an issue of root flipping for some displacement.

I feel your pain, we are right now struggling with similar annoying small imaginary mode in phonon dispersion calculations for a solid-state material  ;).

Best,
Antti
5
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by marand on February 24, 2021, 02:21:11 PM »
Dear Antti, I believe I did just what you described. To be sure, I have just repeated the calculations, dnd the problematic imaginary frequency still remains.

I am quite positive that this is some kind of numerical artifact, but I do not know how can I get rid of it.

Perhaps it is possible to make Turbomole calculate gradients from 5 points instead of 3?

Marcin
6
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by antti_karttunen on February 23, 2021, 11:25:56 PM »
Hi,

About NumForce and symmetry: Earlier today I had a quick look and at least for a simple test system (C2v symmetry) this ran:
1. Using point group C2v, optimize the structure of the S1 state with jobex -level cc2 (+other possible parameters)
2. Copy all files from the optimization directory to new directory.
3. Remove any C2v-related ricc2 keyword from control (like $excitation_energies_CCS_____1^a1__ file=exstates)
4. Change $ricc2 data group to calculate gradients for S1 state in C1 point group:

Code: [Select]
$ricc2
geoopt model=cc2 state=(s1)
$excitations
irrep=a nexc=1
# or possibly few more states in nexc, whatever works best for your system

5. Run NumForce with
Code: [Select]
NumForce -central -c -level cc2It will generate displacements based on C2v. The gradients will be calculated for displaced structures (always with C1 symmetry and irrep a). Hessian will be obtained for C2v-symmetric.

So, in principle the process technically works, but I don't know if this is actually reasonable. H2O runs fine, but I did get a large imaginary frequency. So, this is bit messy and I have doubts if this would actually help. Unfortunately, I don't have any other possible solutions in mind. Hopefully someone could also contribute!

Best,
Antti
7
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by marand on February 23, 2021, 02:12:42 PM »
I have never been able to compute excited state frequencies with symmetry. There has never been a complete hessian after all the displacement gradients were completed.
Ground state is ok, though.

What I normally do is I reduce symmetry to c1 (with define), set coordinates to cartesian, and finally change excitated state symmetry from e.g. b1 to a. So far I have never enountered problems of this kind, except of course close energetic proximity of some other state (which in c1 symmetry may interact with my target state), in which case all sorts of trouble may happen.

Can one do calculations of the excited state frequencies with symmetry, counting on the NumForce module to be smart enough to reduce the symmetry for the non totally symmetric displacements, changing the excited state representation accordingly? This would be marvellous, but I do not think it is ok with turbomole.

Yours!
Marcin
 
8
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by antti_karttunen on February 23, 2021, 09:55:58 AM »
Hi

I'm just bit confused about this one thing:

Quote
Geometry optimization was conducted in C2v symmetry, and only then the symmetry was artificially reduced to c1 so that the NumForce would run properly.

So did you manually reduce symmetry C2v -> C1 before NumForce? Because this choice will significantly affect the number of displacements and possibly also have some numerical consequences (symmetry of the Hessian). What happens if you run NumForce directly for the C2v-symmetric structure?

Best,
Antti
9
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by marand on February 23, 2021, 09:31:46 AM »
Thanks for all the advice! However, I think I have tried those possibilities.

Geometry optimization was conducted in C2v symmetry, and only then the symmetry was artificially reduced to c1 so that the NumForce would run properly.
Coord and gradient was obtained with symmetry. The thight convergence for orbitals was required in exactly the way as shown, and the citeria were even tighter (10^-10 for energy and density).

Geometry was optimized with the gcart 5 option.

The strangest thing is that the imaginary frequency appears even though the PES (along the imaginary frequency related normal mode) shows minimum, and geometry distorted along this coordinate returns to the planar one after reoptimization of the geometry. I must say, I am at a loss at to what to do now.

Best regards!
Marcin

10
Aoforce and Numforce / Re: Stubborn imaginary frequency
« Last post by antti_karttunen on February 22, 2021, 11:35:10 PM »
Thanks for the clarification. Seems like a tough case. One question and one more comment:

1) You have used C2v point group symmetry during the excited state optimization. Is there some particular reason why you remove symmetry before running NumForce? Is this because NumForce anyway breaks symmetry and then $excitations keyword used foir C2v will not be valid? I'm just guessing here since I don't remember if I ever ran excited state frequency calculations with ricc2, but what if after the geometry optimization you just put in control file:
Code: [Select]
$ricc2
geoopt model=cc2 state=(s1)
$excitations
irrep=a nexc=1
# or possibly few more states in nexc, whatever works best for your system
And then run NumForce (for the C2v symmetric structure). I'm just thinking if symmetry could help to reduce some numerical noise here.

2) You mentioned increasing convergence criteria for orbitals, but what keywords did you exactly use? Did you try something like
Code: [Select]
$scfconv 8
$denconv 1d-8
or even tighter during the optimization? And then jobex with -gcart 5?
Pages: [1] 2 3 ... 10