Response vector not found on file RSPVEC - SOCs calculations

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.)

escudero
Posts: 6
Joined: 07 Jan 2015, 11:20
First name(s): Daniel
Last name(s): Escudero
Affiliation: CEISAM, Université de Nantes
Country: France

Response vector not found on file RSPVEC - SOCs calculations

Post by escudero » 07 Jan 2015, 13:07

Dear Dalton team,

I am computing SOCs between higher-lying singlet and triplet excited states (see attached inputs and outputs).
For some molecular systems I get the following error:

Response label X1MNF-SO with frequency -0.06040605 and symmetry 1
not found on file RSPVEC

--- SEVERE ERROR, PROGRAM WILL BE ABORTED ---

Hopefully you can help me.
With best wishes,

Daniel
Attachments
test.out
(83.34 KiB) Downloaded 350 times
test.mol
(1.52 KiB) Downloaded 307 times
test.dal
(228 Bytes) Downloaded 326 times

olav
Posts: 145
Joined: 28 Aug 2013, 06:20
First name(s): Olav
Last name(s): Vahtras

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by olav » 07 Jan 2015, 13:38

Dear Escudero,

Your sample input is an 8 hour calculation. Have you tried to reproduce the problem on something smaller? You say this error shows up for "some molecules", what makes them different from others?

Regards,
Olav

escudero
Posts: 6
Joined: 07 Jan 2015, 11:20
First name(s): Daniel
Last name(s): Escudero
Affiliation: CEISAM, Université de Nantes
Country: France

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by escudero » 07 Jan 2015, 14:00

Dear Olav,

Thank you for your quick answer. For the following molecule (which is even larger) it was working perfectly fine:

BASIS
6-31G*
xxx
xxx
Atomtypes=2 Angstrom
Charge=6.0 Atoms=26
C -0.009964 0.477275 0.107367
C -0.031723 -0.030963 1.421969
C 1.194835 -0.422494 1.995707
C 2.389792 -0.305273 1.288416
C 2.390748 0.196448 -0.015505
C 1.184007 0.584512 -0.603021
C -1.302746 -0.148904 2.175414
C -2.077891 1.026941 2.452061
C -3.318797 0.911845 3.184165
C -3.765442 -0.377441 3.629543
C -2.990307 -1.553286 3.352881
C -1.749387 -1.438191 2.620798
C -1.632598 2.319901 2.096931
C -2.375023 3.475737 2.390795
C -3.566960 3.365096 3.093861
C -4.015952 2.098513 3.503209
C -1.052243 -2.624870 2.301760
C -1.501256 -3.891449 2.711097
C -2.693204 -4.002082 3.414147
C -3.435621 -2.846239 3.708001
C -5.036456 -0.495382 4.382991
C -6.262979 -0.103565 3.809372
C -7.457932 -0.220787 4.516670
C -7.458919 -0.722799 5.820480
C -6.252212 -1.111150 6.407876
C -5.058247 -1.003907 5.697483
Charge=1.0 Atoms=18
H -0.678083 2.427903 1.596249
H -6.270959 0.277852 2.792177
H -8.391949 0.076089 4.046375
H -8.390443 -0.810193 6.373141
H -6.239659 -1.494276 7.425101
H -4.122451 -1.298361 6.164169
H 1.202838 -0.803682 3.012988
H 3.323835 -0.601922 1.758801
H 3.322275 0.283839 -0.568162
H 1.171432 0.967407 -1.620333
H -0.945785 0.771496 -0.359414
H -1.998910 4.446730 2.081282
H -4.147370 4.247282 3.348684
H -4.918284 2.033280 4.098985
H -0.149907 -2.559648 1.705992
H -0.920852 -4.773641 2.456280
H -3.069332 -4.973072 3.723650
H -4.390144 -2.954228 4.208673

If needed, I can send you the inputs and outputs for this one too.
Thanks,

Daniel

olav
Posts: 145
Joined: 28 Aug 2013, 06:20
First name(s): Olav
Last name(s): Vahtras

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by olav » 07 Jan 2015, 14:06

If you say that this example worked provide complete inputs and outputs for that case.

escudero
Posts: 6
Joined: 07 Jan 2015, 11:20
First name(s): Daniel
Last name(s): Escudero
Affiliation: CEISAM, Université de Nantes
Country: France

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by escudero » 07 Jan 2015, 14:18

Sure, please find enclosed the files.
p.s: Note that the only differences between the .dal files is the functional used and the number of roots. I am now doing some tests to check this.
Attachments
1.out
(87.07 KiB) Downloaded 330 times
1.mol
(1.45 KiB) Downloaded 294 times
1.dal
(232 Bytes) Downloaded 297 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: Response vector not found on file RSPVEC - SOCs calculat

Post by arnfinn » 07 Jan 2015, 14:30

I have started a calculation with your input, so see if I experience the same issues. Just to be sure, are you sure you have not deleted or overwritten your scratch directory by another calculation?

escudero
Posts: 6
Joined: 07 Jan 2015, 11:20
First name(s): Daniel
Last name(s): Escudero
Affiliation: CEISAM, Université de Nantes
Country: France

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by escudero » 07 Jan 2015, 14:48

Thank you Arnfinn. Yes, I am almost sure that the scratch directory wasn't deleted nor overwritten by another calculation.
Please keep me updated about this issue.

olav
Posts: 145
Joined: 28 Aug 2013, 06:20
First name(s): Olav
Last name(s): Vahtras

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by olav » 07 Jan 2015, 15:12

Most likely a bug related to the number of roots then. We will investigate this further.

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: Response vector not found on file RSPVEC - SOCs calculat

Post by arnfinn » 07 Jan 2015, 15:28

escudero wrote:Please keep me updated about this issue.
I got the same error...

escudero
Posts: 6
Joined: 07 Jan 2015, 11:20
First name(s): Daniel
Last name(s): Escudero
Affiliation: CEISAM, Université de Nantes
Country: France

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by escudero » 07 Jan 2015, 17:36

Dear Arnfinn and Olav,

Thank you very much for your answers. Just to inform you that when using STO-3G everything works fine... i'm puzzled with this.

Best,
Daniel

olav
Posts: 145
Joined: 28 Aug 2013, 06:20
First name(s): Olav
Last name(s): Vahtras

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by olav » 08 Jan 2015, 16:00

Hi Daniel

Please apply the included patch: in the top level directory

Code: Select all

$ gunzip qr.patch.gz
$ patch -p0 < qr.patch
This is a two-line fix in DALTON/rsp/hypinp.F which should solve your problem. If not, let us know.

Regards,
Olav
Attachments
qr.patch.gz
(252 Bytes) Downloaded 328 times

escudero
Posts: 6
Joined: 07 Jan 2015, 11:20
First name(s): Daniel
Last name(s): Escudero
Affiliation: CEISAM, Université de Nantes
Country: France

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by escudero » 17 Feb 2015, 11:04

[quote="olav"]Hi Daniel

Please apply the included patch: in the top level directory

Code: Select all

$ gunzip qr.patch.gz
$ patch -p0 < qr.patch
This is a two-line fix in DALTON/rsp/hypinp.F which should solve your problem. If not, let us know.


Thank you Olav,
It worked out.

Best,
Daniel

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

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by peppelicari » 03 Mar 2016, 23:05

Dear all,
I had similar problem. Here I attached three different calculations of test SOC job for water molecule.
In the first two jobs I apply C2 symmetry to water (even if the full symmetry is C2v). The first job terminates correctly, but in this test job I'm asking only for the "Z1MNF-SO" component in the Response module. In the second calculation I ask for all the three components. In this case the run terminates with the error message "Response label X1MNF-SO not found on file RSPVEC" (and basically I would get the same error message if the program has to calculate the Y1MNF-SO component). So I get error when X1MNF-SO and Y1MNF-SO are required. In the latter calculation, I added the "NoSymmetry" keyword to the *.mol file and asked again for all the three components; here the calculation runs just fine.
It seems something related to the symmetry. Am I writing something weird in the input? Could this patch be useful for my case?

I thank you for the help and wish you a nice day!
Regards,
Giuseppe
Attachments
test_SOC_water_3.out
"NoSymmetry" calculation. Run is successful.
(71.16 KiB) Downloaded 276 times
test_SOC_water_2.out
All component demanded. Run is aborted.
(83.3 KiB) Downloaded 275 times
test_SOC_water.out
Only Z1MNF-SO demanded. Run is successful.
(79.2 KiB) Downloaded 271 times

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: Response vector not found on file RSPVEC - SOCs calculat

Post by taylor » 04 Mar 2016, 09:50

I am not sure what you see as a problem here. If you run with symmetry (even C2 rather than C2v for water) then different components of the spin-orbit (SO) operator appear in different irreps. If you run just the z-component of the SO operator, the response equations are solved within the appropriate irrep. Since there is no response vector that has the symmetry of the x- or y-components of SO generated, there is nothing written to file. You can "restart" a z-component calculation, possibly looking for more states, because there's at least one converged vector to start the calculation. But you cannot "restart" the other components, because there was no calculation done in the first place for those irreps.

I suspect this is why it "works" when "Nosymmetry" is specified: then the point group is C1 and all SO components are in the same irrep. There is therefore a converged vector to "restart" from in this case.

Best regards
Pete

olav
Posts: 145
Joined: 28 Aug 2013, 06:20
First name(s): Olav
Last name(s): Vahtras

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by olav » 04 Mar 2016, 11:14

There may be some problem in the code, but you also have a problem in your input file. In testcase 2
Atomtypes=2 Angstrom Generators=1 XY
Charge=8.0 Atoms=1
O 0.0 0.0 0.0
Charge=1.0 Atoms=1
H 0.757 0.586 0.0
you have a linear water molecule
Bond angles (degrees):
----------------------

atom 1 atom 2 atom 3 angle
------ ------ ------ -----
bond angle: H _1 O H _2 180.000
Change the symmetry generator to X (reflection only), or just enter all atoms and skip the Generators keyword, the program will detect the symmetry.
Is it a good practice to check the total symmetry whenever you compare calculations with and without symmetry.
test_SOC_water_2.out:@ Final DFT energy: -76.340650630031
test_SOC_water_3.out:@ Final DFT energy: -76.391778348772
test_SOC_water.out:@ Final DFT energy: -76.340650630031
Regards,
Olav

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

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by peppelicari » 04 Mar 2016, 13:15

Dear Pete and Olav,
thanks a lot for your answers.

First of all, sorry for the molecule of water, I just picked up coordinates from internet without checking the symmetry. Now I replace that structure with a right one, as you can see from the following two outputs the DFT energies are the same for calculation with and without symmetry. Anyway, the error in the response vector is still there (now I set the symmetry to be detected automatically, so C2v point group with 4 irreducible representations).

About the comment of Pete, well at the beginning I had the same impression, if there is no response vector maybe the x- and y- contribution must be absent for the demanded point group. But then I looked at the calculation using the no symmetry and there, one can clearly see contributions to the SOC from y- component (in test_SOC_water_5.out):

@ Transition moment <B | A - <A> | C> in a.u. for
@ A operator label, symmetry, spin: Y1MNF-SO 1 1
@ B excited state no., symmetry, spin: 2 1 0
@ C excited state no., symmetry, spin: 1 1 1

@ B and C excitation energies, moment: 0.35815554 0.25366750 -0.00024522

Here there is a non negligible SOC constant of 54 cm-1 between S2 and T1 states (if I understand correctly the output, correct me please if I'm wrong). Then, how could be that the calculation with symmetry cannot calculate this value? Shouldn't the two calculations give the same numerical results (as for the case of the DFT energies pointed out by Olav)?

I always thank you for your precious help!
Regards,
Giuseppe
Attachments
test_SOC_water_4.out
(80.47 KiB) Downloaded 215 times
test_SOC_water_5.out
(51.78 KiB) Downloaded 214 times

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: Response vector not found on file RSPVEC - SOCs calculat

Post by taylor » 04 Mar 2016, 17:12

Sorry, I thought I had made this clear in my earlier posting. If you run with "Nosymmetry" all components are in the same irrep (since there's only one irrep) and thus assuming you have a restart file you can restart the calculation and ask for more properties and more roots and this is all fine because in this calculation it is indeed a restart. But if you run with symmetry and have only calculated one property, well, there are no vectors available that correspond to irreps other than that property. Since it is common that the number of functions in each irrep is different across the irreps, even the dimension of the response vectors will be different from irrep to irrep, which is why it is essentially impossible to "restart" such a calculation.

If you invoke "Nosymmetry", you throw away all of the structure that the group theory provides and your response vectors, for a closed-shell SCF/DFT case, will be of length number-of-occupied times number-of virtuals. This is true even if, because of the intrinsic symmetry, many of these elements will be zero. If you invoked symmetry, the vectors will be shorter, but then you have to ensure that if you are "restarting" a calculation, there are vectors from all irreps of interest available. This is trivial with "Nosymmetry", although it doesn't always mean that you get what you might want! But with symmetry, you need to have vectors available for all irreps of interest in order to restart.

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: Response vector not found on file RSPVEC - SOCs calculat

Post by peppelicari » 04 Mar 2016, 19:17

So let's see if I understood. In the case of water molecule there are 4 irrep (A1, A2, B1, B2), A1 transforms as z, B1 as x and B2 as y. Then, would one possible solution be to start 3 different calculations specifying only one irrep (even several states) and asking only for the SOC component that transforms as that irrep? For example specifying the following?:
.PROPRT
Z1MNF-SO
.ROOTS
2 0 0 0

If not what do you suggest as a solution for this? I mean how can I carry out the calculation obtaining the SOC between Sn and Tn being sure that I can get the total value of the SOC?

Thanks and I wish you a nice weekend!
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: Response vector not found on file RSPVEC - SOCs calculat

Post by kennethruud » 05 Mar 2016, 09:42

I think the easiest would be to ask for the lowest excited states in all relevant irreps as well as all components of the SO operators, i.e.

.PROPRT
X1MNF-SO
.PROPRT
Y1MNF-SO
.PROPRT
Z1MNF-SO
.ROOTS
2 2 2 0

(or
.ROOTS
2 2
if you are running in C2 symmetry).


Best regards,

Kenneth

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

Re: Response vector not found on file RSPVEC - SOCs calculat

Post by peppelicari » 05 Mar 2016, 09:56

Dear Kenneth,
I tried as you proposed but it didn't worked, this is the input for C2:
**DALTON INPUT
.RUN RESPONSE
**INTEGRALS
.MNF-SO
**WAVE FUNCTIONS
.DFT
CAMB3LYP
**RESPONSE
.TRPFLG
.MAXPHP
250
.MAXRM
1000
*QUADRATIC
.DOUBLE RESIDUE
.THCPP
2.0D-3
.MAXITL
100
.THCLR
2.0D-3
.ISPABC
1 0 1
.PROPRT
X1MNF-SO
.PROPRT
Y1MNF-SO
.PROPRT
Z1MNF-SO
.ROOTS
2 2
.PRINT
3
**END OF INPUT
Isn't this input as you propose? Well, this terminates with the error message that "Response label X1MNF-SO with frequency -0.02471493 and symmetry 2 not found on file RSPVEC". How should be my input then for C2?

Thanks!
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: Response vector not found on file RSPVEC - SOCs calculat

Post by taylor » 05 Mar 2016, 09:57

Kenneth saved me some work: he has posted some of what I was planning to write. An advantage of running this way is that you don't have to worry about issues, which happen quite frequently, such as T1 corresponding (roughly) to the triplet coupling of the occupancy that dominates S2, say. It is not always the case that there is a one-to-one correspondence between Sn and Tn for n values of interest: they may "cross". This can be explored first using linear response calculations.

I am not sure what you mean by "the total value of the SOC". If the symmetry is high enough that each component of the SO operator transforms according to a different irrep, then there can be only one matrix element coupling any Sm to any Tn. If the symmetry is lower then there may be multiple possible matrix elements, but Kenneth's suggestion will produce them all, provided the root specification covers the cases of interest.

Finally, the spatial part of the spin-orbit operator transforms as an axial vector and in C2v would be A2, B1, and B2, so (depending on orientation) symmetries 2, 3, and 4.

Best regards
Pete

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: Response vector not found on file RSPVEC - SOCs calculat

Post by taylor » 05 Mar 2016, 10:02

You are sure your attempt to follow Kenneth's recipe is not a restart and/or is not picking up an old scratch directory? Please upload outputs, by the way: pasting stuff inline produces problems with formatting and simply giving a one-line error message gives us very little to go on.

Best regards
Pete

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: Response vector not found on file RSPVEC - SOCs calculat

Post by taylor » 05 Mar 2016, 10:03

You are sure your attempt to follow Kenneth's recipe is not a restart and/or is not picking up an old scratch directory? Please upload outputs, by the way: pasting stuff inline produces problems with formatting and simply giving a one-line error message gives us very little to go on.

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: Response vector not found on file RSPVEC - SOCs calculat

Post by peppelicari » 05 Mar 2016, 10:25

Dear Pete,
these two calculations (water in C2v and in C2) terminates with error message. I think I'm using the input as suggested but I might be wrong. Can you have a look please and tell me if I set something wrong? If I'm able to setup the calculation for water in C2 point group I should be able also to treat my molecular system.

About the "the total value of the SOC" issue, for example there is an irrep where one has contribution from x-soc and y-soc component, I guess they can be summed up to get the "total value of the SOC", right? Must the sum be vectorial (something like sqrt(x-soc^2 + y-soc^2)) or the two values are really constant to be simply summed up?

Thanks!
Regards
Giuseppe
Attachments
test_SOC_water_6.out
(70.28 KiB) Downloaded 224 times
test_SOC_water_7.out
(60.36 KiB) Downloaded 204 times

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: Response vector not found on file RSPVEC - SOCs calculat

Post by taylor » 05 Mar 2016, 11:58

You have found a bug! At least, I think you have. I inadvertently (quickly cutting and pasting your inputs into an old job file) ran without noticing that it was a very old job file and that it actually invoked Dalton2011(!). This runs to completion and prints satisfactory (in the sense that they are very small as one would expect for this case) spin-orbit matrix elements between the excited states. But when I use exactly the same input files with Dalton2014 (see below) I get exactly the error message that you posted. This is all for the C2v case and I am using exactly your inputs, except that since my builds are sequential I need to add .DIRECT.

Unfortunately I do not have anything later than 2014 built on this machine, and it is not convenient at the moment to do a build, so this is all I have at hand. However, it seems clear that something has got messed up between 2011 and 2014. I will shortly post a bug report for this and hope that one of the response authors (I am not one!) can help.

Incidentally I also tried SCF instead of DFT and again get the same error as for the DFT calculation. This holds whether the SCF calculation is done with .DIRECT or without it.

Best regards
Pete

Post Reply

Who is online

Users browsing this forum: Google [Bot], xiongyan21 and 8 guests