Cavity generation problem with symmetry in PCM

Find answers or ask questions regarding Dalton calculations.
Please upload an output file showing the problem, if applicable.
(It is not necessary to upload input files, they can be found in the output file.)

Post Reply
peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Cavity generation problem with symmetry in PCM

Post by peppelicari » 18 Feb 2016, 16:48

Dear all,
here I'm again!
I'm running quadratic response calculation on a molecule with C(2) symmetry. Since I want to use PCM solvation model and insert by myself the radii for the spheres I made the attached input file (TPA_THF_prova2.out). In the molecule specification I introduced the generator XY and inserted only the necessary atoms, the rest being generated by the C(2) rotation. Then, in the *.dal file I included the radii for carbons, sulphur and nitrogen (not hydrogens) in the *PCMCAV section.

The calculation failed without a clear message error (as can be seen from the file err_TPA_THF_prova2.out), it simply crashed. Even in the output there is no clear indication of what exactly went wrong. I'm wondering if this is maybe a problem with the sulphur, which is the fixed point in the point group, so (let's say) half in each portion of the two symmetric parts.
Could this give a problem when generating the cavity? Some symmetry breaking? I hope some of you have some hint about it or some workaround to solve the issue.

Thanks a lot for the help and I wish you a great day!
Best regards,
Giuseppe
Attachments
err_TPA_THF_prova2.out
(2.56 KiB) Downloaded 246 times
TPA_THF_prova2.out
(72.07 KiB) Downloaded 249 times

peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Re: Cavity generation problem with symmetry in PCM

Post by peppelicari » 18 Feb 2016, 16:55

I want also to let you know that if I do the calculation without symmetry it works correctly. So it is something really related to the symmetry settings, maybe I make some mistake in the input and I am not able to identify. It is a pity if I cannot use the symmetry.
Thanks again!
Giuseppe

taylor
Posts: 595
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: Cavity generation problem with symmetry in PCM

Post by taylor » 18 Feb 2016, 17:29

This time I fear the problems are of your making rather than the program's! Your err_... file does in fact contain a valid error message: the program does not "just crash". About twenty lines down is a Fortran runtime error because an unexpected end-of-file has been encountered on DALTON.INP, which is the internal name for the file that contains the contents of your dal file. And there will be an unexpected end-of-file, because your dal file ends

Code: Select all

**END
and this is not a valid keyword. The manual is quite explicit that keywords, whether **, *, or .-prefixed are read as seven characters (including the prefix characters). You therefore need at least

Code: Select all

**END O
and I would suggest something like

Code: Select all

**END OF DATA
is more usefully readable.

In addition, I do not think the keyword

Code: Select all

**WAVEFUNCTION
is valid either: at least I hope it is not. "Wave function" is two words, not one.

Best regards
Pete

peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Re: Cavity generation problem with symmetry in PCM

Post by peppelicari » 18 Feb 2016, 17:44

Dear Pete,
you are totally right! I feel ashamed for this, apparently I must have deleted part of "**END OF INPUT". Since the job was starting and stopping before the cavity generation I thought it was a problem with that part. In the calculation without symmetry the input was correct, so I did not have any problem.
Sorry again, that was my mistake. But your reply is always ready and prompt. Infinite thanks!

I wish you a nice evening!
Giuseppe

p.s.: I celebrated my birthday only with beers, not champaign unfortunately! :)

peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Re: Cavity generation problem with symmetry in PCM

Post by peppelicari » 23 Feb 2016, 11:53

Hello,
actually I'm still having problems with this calculation. The programs says that the cavity is inconsistent. Is there anything wrong in the input? I'm just using the radii of carbon, sulphur and nitrogen.

Thanks for the help as usual!
Regards,
Giuseppe
Attachments
err_TPA_THF.out
(2.89 KiB) Downloaded 224 times
TPA_THF.out
(74.34 KiB) Downloaded 238 times

arnfinn
Posts: 231
Joined: 28 Aug 2013, 08:02
First name(s): Arnfinn
Middle name(s): Hykkerud
Last name(s): Steindal
Affiliation: UiT
Country: Norway
Location: UiT The Arctic University of Norway

Re: Cavity generation problem with symmetry in PCM

Post by arnfinn » 23 Feb 2016, 12:55

When you specify the symmetry in the mol-file like you do here, Dalton will make a rotation about the z-axis and "create" new atoms. You end up with 40 carbons, for instance. Is that what you want?

If you know the symmetry beforehand, you have to remove the "obsolete" atoms, which will be created afterwards by Dalton. BTW, are you sure you have a molecule with symmetry (you might have; I am no expert in symmetry)?

taylor
Posts: 595
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: Cavity generation problem with symmetry in PCM

Post by taylor » 23 Feb 2016, 13:50

I think Arnfinn has pointed you to the problem here. When you specify symmetry you specify (as, as far as I can see, you have) only non-redundant atoms. For example, your molecule has one S, which is just about at the coordinate origin, and presumably, e.g., two N atoms, only one of which you have specified, so the other will be generated by the two-fold rotation. Like Arnfinn, there is at least one area I'm not expert in, and in my case one of those is solvent/PCM. I have no idea how the cavity is generated or what to specify, but it may be that the "spheres" are not generated automatically for the symmetry-dependent centres. I don't know, although I thought this should be documented in the manual.

Two other observations, one that might conceivably be related to your problem, although I doubt it, and one that is certainly not but is just a suggestion.

First, you will see that Dalton determines that "The molecule is planar". This is a consequence of the diagonalization of the moment-of-inertia tensor and relationships between the resulting eigenvalues. So your molecule is planar, and it has a two-fold axis. Rephrasing this, there is a plane of symmetry as well as a two-fold axis. In that case your symmetry must be higher than C2. If the plane is perpendicular to the rotation axis the point group is C2h and if the rotation axis lies in the plane the group is C2v. If this is unexpected perhaps you have an error with the geometry, which might (although as I said I doubt it) cause problems at the PCM step. If the molecule indeed has a higher symmetry, in general one should use the highest symmetry available.

My suggestion, looking at your mol file, is that it would pay off to label your atoms specifically. You have 20 symmetry-independent carbons, for instance, and it seems to me that if you calculate any properties that relate to particular centres it is going to be a hell of a job to figure out which is which. A few moments to label them as C01, C02,...C20 will make subsequent analysis much easier. As you see the program generates dependent atoms with their own labels, but this does not distinguish them from one another if the symmetry-distinct atoms are all labelled the same. This situation is of course the same even if there is no symmetry if all the carbons are labelled C.

Best regards
Pete

peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Re: Cavity generation problem with symmetry in PCM

Post by peppelicari » 23 Feb 2016, 15:01

Dear Arnfinn and Pete,
in the *.mol file I left the atoms that are needed to create the entire molecule after a C(2) rotation through the sulphur atom, the S is the fixed point in the point group. The whole molecule contains 73 atoms (40 C, 2 N, 1 S, 30 H), but in the *.mol file I specified 37 atoms (36 + 1 S, which is the fixed point). The atoms are the correct ones, as you could verify from the .xyz file attached and I'm sure that there is symmetry (for example the auto-detection of symmetry, using the whole molecule, detects the C(2) point group without problem). I thought it could have been because of the S atom, which being the fixed centre might have given troubles after the rotation in the generation of the cavity around it. But then I removed the sphere from S and got the same problem ("Cavity is inconsistent with symmetry") in the Hermit module.

Unfortunately I did not find any info in the manual. Only that the .ICESPH=1 option cannot work together with molecular symmetry. But I'm using .ICESPH=2.

Any suggestion? How could I specify the spheres if I decide to input the whole molecule with auto detection of symmetry?

Thanks!
Giuseppe
Attachments
XYZ_TPA_THF.out
xyz file of the "half-molecule"
(2.94 KiB) Downloaded 232 times

arnfinn
Posts: 231
Joined: 28 Aug 2013, 08:02
First name(s): Arnfinn
Middle name(s): Hykkerud
Last name(s): Steindal
Affiliation: UiT
Country: Norway
Location: UiT The Arctic University of Norway

Re: Cavity generation problem with symmetry in PCM

Post by arnfinn » 23 Feb 2016, 15:24

It seems like I was wrong (not very uncommon:)).

You can take a look at the test case "pcm_neq_exc_sym" to see how to specify the spheres with symmetry (the molecule input format is a little bit old fashion, but I guess you will understand it anyway). In short, you have to specify spheres also for the "redundant atoms". Thus, you have to specify 43 spheres in your dalton input file (if I have done the math correctly).

peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Re: Cavity generation problem with symmetry in PCM

Post by peppelicari » 23 Feb 2016, 16:14

Dear Arnfinn,
thanks for the hint, that solved my problem.
Anyway, in the test job the molecule is simply formaldehyde (C2v point group), so one needs to add only one "redundant" sphere. In my case it is a bit more tricky to understand in which order they need to be placed. However, I made the first try just inserting the redundant symmetric atoms in the same order as they were in the *.mol file and it works! For example for my case, I had the original 22 atoms (indexes from 1 to 22, 20C + 1S + 1N), then I added other 21 atoms (indexes from 38 to 58, 20C + 1N), skipping all the hydrogens and without repeating, of course, the sulphur. This seems to work.
Maybe it could be useful for other people that will encounter the same issue.

I thank a lot to all of you again for your valuable help and more important I wish you a nice working week!
Best regards,
Giuseppe

taylor
Posts: 595
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: Cavity generation problem with symmetry in PCM

Post by taylor » 24 Feb 2016, 09:19

I think it is very unfortunate that the PCM code apparently is incapable of generating spheres around symmetry-related centres and will take this up separately on the Dalton developers' list. In the meantime there is no problem (other than the labour involved) to come up with the necessary input: the program internally generates the redundant centres by considering the distinct centres one at a time. For example, in ethene with a single distinct C in the input followed by a distinct H, the program will first generate the second carbon, then for the H it will generate the three other H atoms. Since you need to have the same cavity around each symmetry-related atom (otherwise the overall symmetry will be broken) you simply need to replicate each atom, one after another.

Note that if you want to have different radii around different centres of the same type (I don't know why one would, but I have no clue about these things anyway) you probably need more information about each centre. See my last posting about giving atoms unique identifiers!

And speaking of my last posting, you did not respond on the question of how your "C2" molecule can also be planar without having a higher symmetry?

Best regards
Pete

peppelicari
Posts: 23
Joined: 05 Feb 2016, 16:26
First name(s): Giuseppe
Last name(s): Licari
Affiliation: University of Geneva
Country: Switzerland

Re: Cavity generation problem with symmetry in PCM

Post by peppelicari » 24 Feb 2016, 09:55

Dear Pete,
thanks for explaining how one should do for other symmetries, like for ethene, and thanks for considering to improve this feature in Dalton code.

My molecule is not planar, if you open the xyz file attached above you might look at the molecule (well, half of it!). It doesn't have a vertical or horizontal plane of symmetry, because phenyls and thiophene are not in the same plane (this in the ground state). But there is still a C(2) axis which make the molecule symmetric (for example H2O2 is another example of molecule with C2 symmetry). I did not state that my molecule is planar.

I wish you a nice day!
Best regards,
Giuseppe

kennethruud
Posts: 269
Joined: 27 Aug 2013, 16:42
First name(s): Kenneth
Last name(s): Ruud
Affiliation: UiT The Arctic University of Norway
Country: Norway

Re: Cavity generation problem with symmetry in PCM

Post by kennethruud » 24 Feb 2016, 10:42

The handling of symmetry in Dalton is described in Journal of Computational Chemistry 25, 375 (2004). It is somewhat difficult to do this very generally, and the choices made in Dalton has been for instance based on a insisting on e.g. water calculated in C2v, C1 or Cs symmetry should always give the same energy, which is not normally straightforwardly achieved with the cavity generation routines. I quote a couple of paragraphs from the article:

"With respect to symmetry, the DALTON code11 can use the D2h abelian point group and all its subgroups (D2, C2ν, C2h, C2, Cs). The GEPOL algorithm has been adapted to this feature by always starting with a single sphere tessellation in D2h symmetry: in other words, according to the algorithm described in paper I, only one-eighth of a sphere is tessellated; this portion is then replicated to match the specified point group symmetry. Finally, the algorithm of paper I is followed to get to the UFS part of the cavity. This change in procedure compared to paper I allows us to ensure that the cavity possesses the full symmetry of D2h and its subgroups. If the molecule has a higher point-group symmetry than specified in the program input, the original symmetry is formally retained in the design of the cavity. For instance, performing a calculation on water imposing Cs symmetry will produce the same numerical result as if it had been performed in C2ν symmetry. This would not have been the case if the cavities for Cs and C2ν symmetries were built in a different way.

Another feature of the present implementation is the invariance of the cavity tessellation with respect to the choice of the molecular reference frame. This is achieved by rotating each tessellated sphere of the cavity into a predetermined orientation, which is fixed by the molecular symmetry and the moment of inertia tensor of the molecule, without any reference to the external frame. This allows us to obtain the exact same result for two calculations performed with a molecule oriented in two different ways. Without this step, the rotation of the molecule would not correspond to a rotation in the sphere tessellation, and thus even though the same set of spheres would be obtained, the tessellation would differ, and consequently, a slightly different description of the solvent response would be obtained. Although these differences would be small, invariance to molecular rotation and applied point group symmetry are highly desirable criteria for any use of molecular point group symmetry in ab initio calculations, requirements that are met by the scheme presented here."


Best regards,

Kenneth

Post Reply

Who is online

Users browsing this forum: No registered users and 12 guests