[CP2K-user] [CP2K:20037] Matsui-Akaogi Force Field for bulk TiO2

[Léon Luntadila Lufungula]

Dear Matthias,

Thanks for the concise answers to my questions!

Kind regards,
Léon

On Monday 18 March 2024 at 09:58:37 UTC+1 Krack Matthias wrote:

> I had never an issue with that or considered explicitly the actual FFT 
> lengths provided by the employed FFT library. I select values for GMAX 
> appropriate for the actual cell size (e.g. 1 or 2 points per Angstrom). The 
> code should select automatically available FFT lengths equal or larger than 
> the requested values.
>
>  
>
> Best
>
>  
>
> Matthias
>
>  
>
> *From: *cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Léon 
> Luntadila Lufungula <Leon.luntad... at uantwerpen.be>
> *Date: *Friday, 15 March 2024 at 11:32
> *To: *cp2k <cp... at googlegroups.com>
> *Subject: *Re: [CP2K:20034] Matsui-Akaogi Force Field for bulk TiO2
>
> P.S. For GMAX it is stated in the manual that "The number of points needs 
> to be FFTable (which depends on the library used) and odd for EWALD.", 
> but I don't know which numbers are FFTable for FFTW3, could you point me in 
> the right direction?
>
> All the best,
> Léon
>
> On Friday 15 March 2024 at 11:14:02 UTC+1 Léon Luntadila Lufungula wrote:
>
> Dear Matthias,
>
>  
>
> Thanks for pointing me to the right NONBONDED implementation, it seems to 
> work well with BUCKMORSE 
> <https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/FORCEFIELD/NONBONDED/BUCKMORSE.html>. 
> I also found my error with the other WILLIAMS parameters, I messed up the 
> exponent for the units of C (kcalmol*angstrom^-6 instead of 
> kcalmol*angstrom^6), so now they are working as they should! I was 
> wondering though, how important it is to make sure your cell dimensions are 
> larger than RCUT? I tried redoing the calculation with a 1x1x1 cell instead 
> of 8x8x3 and the results seem to be more or less the same...
>
>  
>
> Kind regards,
>
> Léon
>
>  
>
> On Thursday 14 March 2024 at 20:52:55 UTC+1 Krack Matthias wrote:
>
> Hi Léon
>
>  
>
> I think there is no need to use GENPOT, because that nonbonded force field 
> contribution is implemented in CP2K as BUCKMORSE 
> <https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/FORCEFIELD/NONBONDED/BUCKMORSE.html>. 
> There are often typos in papers with force field parameters and it is 
> advisable to cross-check the parameter tables for the same force field in 
> different papers for consistency. 
>
> Small differences to other codes can have many reasons: splining quality, 
> single- vs double-precision, SPME mesh size, type and threshold values of 
> the employed optimizer etc.
>
>  
>
> Best
>
>  
>
> Matthias
>
>  
>
> *From: *cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Léon 
> Luntadila Lufungula <Leon.luntad... at uantwerpen.be>
> *Date: *Thursday, 14 March 2024 at 11:48
> *To: *cp2k <cp... at googlegroups.com>
> *Subject: *Re: [CP2K:20027] Matsui-Akaogi Force Field for bulk TiO2
>
> Hi Matthias,
>
>  
>
> Thanks for pointing this out for me, I've adjusted the input accordingly 
> and now the calculation seems to work fine. I've increased GMAX to 2 points 
> per angstrom and also used MULTIPLE_UNIT_CELL 8 8 3 (each cell vector 
> should thus be around 30 Angstrom in length), because I read that the cell 
> edge should be at least twice the size of RCUT (i.e., > 24 Angstrom). I 
> tried several versions of the Matsui-Akaogi Force Field to try and get 
> decent values for the cell parameters after a cell optimization as was 
> shown in their original paper 
> <https://www.tandfonline.com/doi/abs/10.1080/08927029108022432>:
>
>  
>
> However, no matter which definition I try, I always get cell parameters 
> which are quite a bit larger than those reported and I can't seem to get 
> the original FF definition to work with GENPOT. The parameter sets I tried 
> are attached and labeled as follows:
>
> - Matsui-Akaogi (MA) 
> <https://www.tandfonline.com/doi/abs/10.1080/08927029108022432>: original 
> definition with GENPOT [CPASSERT failed]
>
>
> - Naicker (N) <https://pubs.acs.org/doi/10.1021/jp050963q>: seemingly 
> unaltered WILLIAMS version of the original MA definition, although I'm not 
> 100% sure how to convert the original function into a WILLIAMS type 
> definition. But f(Bi+Bj) does coincide with rho_ij and Cij coincides with 
> (Ci+Cj) [a = 3.8865, c = 10.0187]
>
> - Predota (P) <https://pubs.acs.org/doi/10.1021/jp037197c>: slightly 
> altered version to better simulate TiO2 surfaces [a = 3.8865, c = 10.0187]
>
> - Alimommadi (A) <https://pubs.acs.org/doi/10.1021/jp207272e>: parameters 
> should be the same as Predota but are given in eV instead of kcalmol, but 
> give slightly different result [a = 3.886, c = 10.0177]
>
>  
>
> Do you know why this could be? Is there still something wrong with my 
> settings perhaps or should I just accept that the result might not be 
> ideal? Also, I don't see what I'm doing wrong with the GENPOT definition of 
> the original MA FF, so any help would be greatly appreciated!
>
>  
>
> Kind regards,
>
> Léon
>
> On Wednesday 13 March 2024 at 18:26:08 UTC+1 Krack Matthias wrote:
>
> Hi Léon
>
>  
>
> your force field definition is incorrect. The unit for the C parameters 
> should be [kcalmol*angstrom^-6] instead of [kcalmol*angstrom^-1] and the B 
> parameter for Ti-O should be something like 5.15 rather than 0.00515. The 
> simulation cell is also too small for the given RCUT value of 12, but that 
> shouldn’t matter.
>
>  
>
> HTH
>
>  
>
> Matthias
>
>  
>
> *From: *cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Léon 
> Luntadila Lufungula <Leon.luntad... at uantwerpen.be>
> *Date: *Wednesday, 13 March 2024 at 12:20
> *To: *cp2k <cp... at googlegroups.com>
> *Subject: *[CP2K:20022] Matsui-Akaogi Force Field for bulk TiO2
>
> Dear all,
>
>  
>
> I want to test a force field for TiO2 so that I can use it for 
> pre-equilibrating my system for AIMD calculations. However, I have no 
> experience in using force fields and I'm having trouble implementing the FF 
> calculations. I took the FF parameters from a paper from Naicker et al 
> <https://pubs.acs.org/doi/10.1021/jp050963q> and tried to calculate the 
> energy of the anatase crystal structure, before going on to a cell 
> optimization of the structure to compare to DFT. 
>
>  
>
> However, when I try to run the calculation I get an error indicating 
> "GEOMETRY wrong or EMAX_SPLINE too small!", however, the geometry is 
> definitely correct. So I tried increasing EMAX_SPLINE to 1E2 and then got 
> "SPLINE_INFO| Number of points:  2346797 obtained accuracy 0.953210E-07. MM 
> SPLINE: no convergence on required accuracy (adjust EPS_SPLINE and rerun)". 
> I then tried to increase EPS_SPLINE to 1E-6 and the energy calculation ran 
> without a problem. I tried to run a CELL_OPT with these settings but the 
> optimization seems to be going nowhere... (see attached input and output 
> files) Can anyone tell me what I'm doing wrong? Is it perhaps the 
> definition of the forcefield?
>
>  
>
> Kind regards,
>
> Léon
>
> -- 
> You received this message because you are subscribed to the Google Groups 
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to cp2k+uns... at googlegroups.com.
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/e185a18f-fcbc-477a-ba8e-95c0191b4ae4n%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/e185a18f-fcbc-477a-ba8e-95c0191b4ae4n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
> -- 
> You received this message because you are subscribed to the Google Groups 
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to cp2k+uns... at googlegroups.com.
>
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/9e3e09dd-33ad-448c-bbff-bc163b4b46a0n%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/9e3e09dd-33ad-448c-bbff-bc163b4b46a0n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
> -- 
> You received this message because you are subscribed to the Google Groups 
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to cp2k+uns... at googlegroups.com.
>
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/9c7c3e20-f171-4713-8e0b-a54fcdf922cfn%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/9c7c3e20-f171-4713-8e0b-a54fcdf922cfn%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/a21b3938-e103-4c31-89fc-b4d1c54d4d5dn%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240318/5307a810/attachment-0001.htm>