Recent Posts

1

Miscellaneous / How to use HAGA for optimization?

« Last post by Robin1991 on May 13, 2024, 05:28:26 PM »

Hello,

I have just found that there is HAGA optimization (with own executable) implemented in TM
I just don`t know how to use it properly and where I can find the information regarding this.
Maybe someone can say anything about this.

Thanks for the help,
Robin

2

Ridft, Rdgrad, Dscf, Grad / Re: Hyperfine Coupling Interaction Decomposition

« Last post by yannickf on May 13, 2024, 04:43:58 PM »

Dear Joan,


you can only use X2C with PCC and the derivatives of the unitary decoupling in the release version. So, you always get the full analytical derivative. DKH and BSS are not implemented as these are outdated theories and I did not want to implement it. You can simply use "$pnmr hfc-only" and get the results for the total HFC, the paramagnetic spin-orbit term, and the sum of the Fermi-contact and the spin-dipole term. Decoupling the Fermi-contact and the spin-dipole term is only possible in the non-relativistic limit (you can delete $rx2c before an mpshift run and use the non-relativistic EPR operators with the X2C-optimized density, i.e. without PCC). Below is a modified control:

Code: [Select]

$title
Cu(SMe)4 B3LYP ANO-DK3 Uncontracted HFC Term Decomposition Test 1
$symmetry c1
$coord    file=coord
$optimize
 internal   off
 redundant  off
 cartesian  on
$atoms
    basis =ANO-DK3
    jbas  =universal
$basis    file=basis
$uhfmo_alpha   file=alpha
$uhfmo_beta   file=beta
$uhf
$alpha shells
 a       1-66                                   ( 1 )
$beta shells
 a       1-65                                   ( 1 )
$scfiterlimit      300
$scfdamp   start=1.000  step=0.050  min=0.100
$scfdump
$scfdiis
$maxcor    500 MiB  per_core
$energy    file=energy
$grad    file=gradient
$pop
$rx2c
$finnuc
$snso
$snsopara 0
$dft
    functional   b3-lyp
    gridsize   5a
$scfconv   9
$scforbitalshift  closedshell=.05
$ricore      500
$rij
$jbas    file=auxbasis
$pnmr hfc-only
$rundimensions
   natoms=21
$last step     mpshift

PCC is always activated for NMR, EPR, and pNMR properties as results without PCC are typically wrong by orders of magnitude. With this settings you will get the following output for every nucleus.
1) Scalar-relativistic part, i.e. the Fermi-contact plus spin-dipole interaction (molecular frame)
2) The paramagnetic spin-orbit part (molecular frame)
3) Total HFC tensor in principal axis system
4) Eigenvectors for the principal axis system

Version 7.8 prints some more results and the Euler transformations.

Hope this helps.

Best wishes,

Yannick

3

Ridft, Rdgrad, Dscf, Grad / Re: Hyperfine Coupling Interaction Decomposition

« Last post by jcardol on May 13, 2024, 01:03:18 PM »

Dear Yannick,

thank you so much for your reply and pointing out all the related literature, it was really helpful! I was wondering, then, how to obtain the hyperfine coupling Fermi Contact, Spin Dipole Contribution and Spin-Orbit coupling terms individually from a mpshift calculation. I am currently using as input:

Code: [Select]

$title
$symmetry c1
$coord    file=coord
$optimize
 internal   off
 redundant  off
 cartesian  on
$atoms
    basis =ANO-DK3
    jbas  =universal
$basis    file=basis
$uhfmo_alpha   file=alpha
$uhfmo_beta   file=beta
$uhf
$alpha shells
 a       1-66                                   ( 1 )
$beta shells
 a       1-65                                   ( 1 )
$scfiterlimit      300
$scfdamp   start=1.000  step=0.050  min=0.100
$scfdump
$scfdiis
$maxcor    500 MiB  per_core
$energy    file=energy
$grad    file=gradient
$pop
$rdkh 4
$finnuc
$snso
$snsopara 0
$pcc
$dft
    functional   b3-lyp
    gridsize   5a
$scfconv   9
$scforbitalshift  closedshell=.05
$ricore      500
$rij
$jbas    file=auxbasis
$pnmr fc sd psoso temp=  298.00
$rundimensions
   natoms=21
$last step     mpshift


Before launching the mpshift calculation, I use the ridft module to calculate an energy. I have been testing with DKH, but it outputs too big of a HFC constants, I've tried with and without PCC and changing SNSO but it remains too high. Instead when usign X2C one-component calculation it gives reasonable values although I am not capable of finding the paramagnetic output of the Hyperfine Tensor decomposition, just the PSOSO in all calculations.

Thanks for your time and patience and best regards!

Joan

4

Ridft, Rdgrad, Dscf, Grad / two-electron integrals

« Last post by Dawood234 on May 10, 2024, 06:29:50 PM »

Hello,
we were wondering if there is a way to extract two-electron integrals using Kohn-Sham orbitals. For example, is it possible to run a HF or CIS calculation but inputting KS orbitals and outputting the two-electron integrals before the SCF cycles begin?
I could not find a way to do it in the documentation or from searching the forum.
Thank you.
Best,
Davood

5

Escf and Egrad / Re: egrad density difference (ed.plt)

« Last post by Helph on April 15, 2024, 09:39:25 AM »

Thank you very much for taking the time to answer 

It's now clear that when dealing with multiple transitions, only visualizing orbitals is not enough. Egrad proves to be more accurate as it considers all contributions involved in a specific excitation.

6

Ridft, Rdgrad, Dscf, Grad / Re: correlation energy in TMHF or Lh12ct-SsifPW9

« Last post by yannickf on April 12, 2024, 01:12:46 AM »

Dear Franz,

Turbomole V7.8.1 should be available for general users now. So, you can now get the energy contributions for local hybrid functionals.

Best wishes,

Yannick

7

Statpt / Re: statpt - geoired.f: lowest eigenvalue is too small

« Last post by jmkormanik on April 09, 2024, 10:29:54 AM »

Hi,

thank you for your response. Cartesian coordinates are usually, unfortunately, unreasonable for a system of this size, ~400 atoms is a lot and convergence easily takes 600+ steps in jobex even in internal redundant coordinates (these systems are sometimes quite far from equilibrium). Do you think that the optimization of Cartesian and internal redundant would then take roughly the same amount of time?

However, if there is no better solution, then this is what we'll have to do - either use Cartesian coordinates only or do several (~50) steps in Cartesian and then try to switch to ired and see if statpt catches on. If not, continue in Cartesian.

Thank you for the response! I was mostly wondering if there's anything else that we missed.

Have a nice day,

jmkormanik

8

Statpt / Re: statpt - geoired.f: lowest eigenvalue is too small

« Last post by uwe on April 08, 2024, 07:20:40 PM »

Hi,

easiest solution is to switch to cartesian coordinates.

Code: [Select]

$optimize
 internal   off
 redundant  off
 cartesian  on
If the internal redundant coordinates show very low eigenvalues (or if they are so low that statpt refuses to use them), they are most likely not more efficient than simply running the optimization in cartesian ones.

9

Statpt / Re: statpt - geoired.f: lowest eigenvalue is too small

« Last post by jmkormanik on April 08, 2024, 04:22:33 PM »

Hello, can anyone help, do you have any insights to offer? Thank you very much.

10

Parallel Runs / Re: Can turbomole run with multiple GPUs?

« Last post by uwe on April 05, 2024, 04:59:12 PM »

Hello,

whether or not multiple GPUs can be invoked and used correctly mainly depends on the version and the capabilities of the compilers that are used for building the binaries. The developer's version of Turbomole, which is build with GNU compiler, does work with multiple GPUs. The official release version, however, was built using NVidia's HPC compiler suite which showed some errors when using more than one GPU.
The latest version of Turbomole (released in 2024, V7.8.1) which used a very recent NVidia compiler release, might work better than older versions. Please send me a personal message and I will try to help. The results of our discussion (whether or not it is running on your system with multiple GPUs) can then be posted here in this thread.