Molcas Forum
Support and discussions for Molcas and OpenMolcas users and developers
You are not logged in.
- Topics: Active | Unanswered
Announcement
Please note: The forum's URL has changed. The new URL is: https://molcasforum.univie.ac.at. Please update your bookmarks!
You can choose an avatar and change the default style by going to "Profile" → "Personality" or "Display".#1 2020-05-20 07:51:14
- cymantren
- Member
- Registered: 2017-09-18
- Posts: 37
Some internal inconsistency of the code was detected
Hello OpenMolcas-users,
I do an MECP calculation on TiB2-bent and came across a problem. Trying to solve it by simply using a new geometry, produces the same message again:
------------------------------------
***** Orthogonality violated *****
# Some internal inconsistency of the code was detected #.Does one have an idea, how to solve the problem ?
-------------------------------------
----------++ --------- Input file ---------
&GATEWAY
coord = TiB2-gew_ISC.Opt.xyz
basis=6-31G**
Constraints
a = EDiff
value
a =0.000
End of Constraints
>>> DO WHILE
&SEWARD
&SCF
&ALASKA
>>> COPY $WorkDir/$Project.RunFile $WorkDir/RUNFILE2
&SCF
UHF
ZSpin=2
&ALASKA
&SLAPAF
>>> END DO---------
THE XYZ-FILE:
----------------------
3
-897.33437522083807
Ti 0.00000000 0.00000000 -0.19690750
B 0.00000000 0.66931782 1.39690750
B 0.00000000 -0.66931782 1.39690750----------------------
Thank you in advance.
Best wishes
Alfred
Offline
#2 2020-05-20 09:35:59
- Ignacio
- Administrator
- From: Uppsala
- Registered: 2015-11-03
- Posts: 1,085
Re: Some internal inconsistency of the code was detected
It worked for me.
**********************************************************************************************************************
* Energy Statistics for Geometry Optimization *
**********************************************************************************************************************
Energy Grad Grad Step Estimated Geom Hessian
Iter Energy Change Norm Max Element Max Element Final Energy Update Update Index
1 -897.36033832 0.00000000 0.129765 -0.129765 dEdx001 -0.161394 nrc002 -897.36812462 RS-RFO None 0
2 -897.30527895 0.05505937 0.055224 0.055224 dEdx001 -0.153957 nrc001 -897.30626920 RS-RFO BFGS 0
3 -897.24792311 0.05735584 0.051849 0.051849 dEdx001 0.042672 nrc002 -897.24933624 RS-RFO BFGS 0
4 -897.24524881 0.00267430 0.030145 -0.030145 dEdx001 -0.017927 nrc002 -897.24556041 RS-RFO BFGS 0
5 -897.24444827 0.00080053 0.001360 -0.001360 dEdx001 -0.000936 nrc002 -897.24444890 RS-RFO BFGS 0
6 -897.24430000 0.00014828 0.000193 -0.000193 dEdx001 -0.000110 nrc002 -897.24430001 RS-RFO BFGS 0
+----------------------------------+----------------------------------+
+ Cartesian Displacements + Gradient in internals +
+ Value Threshold Converged? + Value Threshold Converged? +
+-----+----------------------------------+----------------------------------+
+ RMS + 5.3806E-05 1.2000E-03 Yes + 1.3629E-04 3.0000E-04 Yes +
+-----+----------------------------------+----------------------------------+
+ Max + 6.4639E-05 1.8000E-03 Yes + 1.9274E-04 4.5000E-04 Yes +
+-----+----------------------------------+----------------------------------+
Geometry is converged in 6 iterations to a Minimum Energy Crossing Point StructureYou can try saving the orbitals and restarting from them for each spin multiplicity.
Offline
#3 2020-05-24 11:19:35
- cymantren
- Member
- Registered: 2017-09-18
- Posts: 37
Re: Some internal inconsistency of the code was detected
Hello Ignacio,
thank you for your answer, which however sounds odd, because it's the same code and input.
It also appeard at a analogous calculation on ZnB2, which were already close to converge.
----------------------
A : Energy difference = -0.01385108 hartree, -36.36600954 kJ/mol
E(i) = -1826.36852102 hartree
E(j) = -1826.35466993 hartree
.......
+ Cartesian Displacements + Gradient in internals +
+ Value Threshold Converged? + Value Threshold Converged? +
+-----+----------------------------------+----------------------------------+
+ RMS + 1.0774E-01 1.2000E-03 No + 3.6184E-01 3.0000E-04 No +
+-----+----------------------------------+----------------------------------+
+ Max + 1.1181E-01 1.8000E-03 No + 5.1173E-01 4.5000E-04 No +
+-----+----------------------------------+----------------------------------+
Convergence not reached yet!
..........
#######################.
.# Convergence problem #.
.#######################.
Timing: Wall=720.13 User=765.96 System=51.51----------------------
I usually read the precalculated geometry in and repeat the calculation. Sometimes it works, sometimes not.
In the case of ZnB2 it does'nt work, as with TiB2.
So I will try as you suggested, but I want to remark, that if the same input-file works on your computer, but does'nt on my computer, so there seems to be an memory/hardware problem ?
ZnB2-bent
---------------------------------
3
-1826.3902316706151
Zn 0.00000000 0.00000000 0.01110524
B 0.00000000 0.65872559 1.38889476
B 0.00000000 -0.65872559 1.38889476----------------------------------
Can you please repeat the calculation on your computer system and tell me, if it works ? I suppose, it does.
What if there is no ISC, no MECP at all ? The program can't show "no MECP found", but one has to stop the calculations and think about the input again, if after the 2nd or 3nd time, there is no "Happy landing".
What else modifications are possible, to improve ones results in a MECP-calculation ? What's about the active space ? Would'nt it be difficult to set up, because its on the border of two geometries ?
Of course I could switch to a casscf/rasscf calculation, but before doing that, I wanted to calculate the whole system on the same level. Of course, later I wanted to improve the results by using mcscf/casscf/rasscf-approach and a better basis-set.
.
Thank you for your suggestions and help.
Best Regards
Alfred
Offline
#4 2020-05-25 09:19:22
- Ignacio
- Administrator
- From: Uppsala
- Registered: 2015-11-03
- Posts: 1,085
Re: Some internal inconsistency of the code was detected
So I will try as you suggested, but I want to remark, that if the same input-file works on your computer, but does'nt on my computer, so there seems to be an memory/hardware problem ?
It could be, but are you sure it's the same code? Are you using the latest OpenMolcas master? It could also be a compilation/library issue. Try with the internal LAPACK routines (no MKL), it will be slow, but not a problem for this case.
Can you please repeat the calculation on your computer system and tell me, if it works ?
What if there is no ISC, no MECP at all ?
Well, here you go:
48 -1826.40676009 -0.03226228 0.354781 0.354781 dEdx001 0.203585* nrc002 -1826.43336865 RS-RFO BFGS 0
49 -1826.44409349 -0.03733341 0.248842 -0.248842 dEdx001 -0.181220* nrc002 -1826.45902400 RS-RFO BFGS 0
50 -1826.40423807 0.03985542 0.148054 -0.148054 dEdx001 -0.211423* nrc002 -1826.41680083 RS-RFO BFGS 0
+----------------------------------+----------------------------------+
+ Cartesian Displacements + Gradient in internals +
+ Value Threshold Converged? + Value Threshold Converged? +
+-----+----------------------------------+----------------------------------+
+ RMS + 8.7459E-02 1.2000E-03 No + 1.0469E-01 3.0000E-04 No +
+-----+----------------------------------+----------------------------------+
+ Max + 1.0167E-01 1.8000E-03 No + 1.4805E-01 4.5000E-04 No +
+-----+----------------------------------+----------------------------------+
Convergence not reached yet!One can see that the geometry is just moving semi-randomly. There may be some instability with SCF calculations converging to different roots.
Offline
#5 2021-08-08 07:55:33
- cymantren
- Member
- Registered: 2017-09-18
- Posts: 37
Re: Some internal inconsistency of the code was detected
Dear Ignacio,
thank you for your reply, but I was busy with other things and had no time to work with OPENMOLCAS.
YES, some problems they may occur, are surely due to my old laptop, which has not al lot of memory.
Sometimes I came across a technical issue, which maybe somewhere connected to the memory of my laptop.
For Instance, I calculated the pyrazine-molecule and it worked, but using the optimized structure,I got a failure message.
These kind of message appear from time to time, but I have no idea why.
Please, can you tell me what may be the source of the error ? Is it also due to the memory ?
Thank you, in advance.
Kindly regards
--------------------------------------------------
&GATEWAY
COORD = MOLCAS.Opt.xyz
GROUP = FULL
TITLE = Molcas-Opt
BASIS = 6-31G**
>>> COPY -FORCE SingletOrbitals.RasOrb INPORB
>>> DO WHILE
&SEWARD
&RASSCF
SPIN = 1
NACTEL = 6,0,0
INACT = 18
RAS2 = 4
CIROOT = 1,1,1
LUMORB
&SLAPAF
>>> END DO
&MCKINLEY-- ----------------------------------
--- Start Module: gateway at Sun Aug 8 08:01:10 2021
......
Found SYMMETRY generators: x
........
&RASSCF
only a single process is used
available to each process: 2.0 GB of memory, 2 threads
pid: 28819
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()
###############################################################################
###############################################################################
### ###
### ###
### ERROR: Read error during input preprocessing. ###
### ###
### ###
###############################################################################
###############################################################################
###############################################################################
###############################################################################
### ###
### ###
### ERROR: Failure reading after INACTIVE keyword. ###
### ###
### ###
###############################################################################
###############################################################################
Error exit 9920 from PROC_INP.
###############################################################################
###############################################################################
### ###
### ###
### ERROR: Input processing failed. ###
### ###
### ###
###############################################################################
###############################################################################
RASSCF Error: Proc_Inp failed unexpectedly.
Check the output file for any previous messages
that can help explain the failure.
Here is a printing of the input file that
was processed:
&RASSCF
SPIN
1
NACTEL
6,0,0
INACT
18
RAS2
4
CIROOT
1,1,1
LUMORB
End of input
###############################################################################
###############################################################################
### ###
### ###
### WARNING: There were warnings during the execution ###
### Please, check the output with care! ###
### ###
### ###
###############################################################################
###############################################################################
###############################################################################
###############################################################################
### ###
### ###
### Location: AixCheck ###
### File: ORDINT ###
### ###
### ###
### Active unit. Should have been closed! ###
### ###
### ###
###############################################################################
###############################################################################
--- Stop Module: rasscf at Sun Aug 8 08:01:22 2021 /rc=_RC_INTERNAL_ERROR_ ---
*** files: xmldump
saved to directory /home/kernel07/bin/OM_pyrazine
>>> END DO
.########################################################.
.# Some internal inconsistency of the code was detected #.
.########################################################.Offline
#6 2021-08-08 11:08:24
- cymantren
- Member
- Registered: 2017-09-18
- Posts: 37
Re: Some internal inconsistency of the code was detected
Dear Ignacio,
regarding to your statement from the 2020-05-24 20:19:22, if there exists a IC or MECP at all, this question is crucial for me. Of course it would be good, if one had an idea, if such points exits at all, regarding to an special molecule, e.g. ZnB2.
I only know the theorem of Longuett Higgins, that, if such points exists, the wavefunction should change its sign.
However I am not quite sure, how to apply this in practice. For Instance, should one calculate several geometries, according to a loop, till the first geometry is again reached and if the wavefunction changes its sign, then probably there will be an IC/MECP ?
Is this the correct strategy ?
How does one recognizes the sign of the wavefunction in OPENMOLCAS ?
Can you please help me with this topic ?
Thank you ?
Kindly regards
Alfred
Offline
#7 2021-08-09 08:38:46
- Ignacio
- Administrator
- From: Uppsala
- Registered: 2015-11-03
- Posts: 1,085
Re: Some internal inconsistency of the code was detected
For your error, keywords like INACT and RAS2 must give the number of orbitals in each irrep, and you have more than 1 because you have symmetry element.
As for the sign of the wavefunction, you'd have to compute overlaps between wavefunctions (with RASSI) and see whether they're close to +1 or -1. But it may be easier to just compute more than 1 state and see if they're close in energy.
Offline
#8 2021-08-11 09:14:01
- cymantren
- Member
- Registered: 2017-09-18
- Posts: 37
Re: Some internal inconsistency of the code was detected
Dear Ignacio,
thank you for your replies to my questions.
OK, in Cs-sym. one have more than one sym. element and has to be adjusted
Yes, it seems more easy to compute with e.g SA-2 more than one state, but I thought calculating a loop would be more systematic. If the wfn changes its sign, I know there must be a CI - but I don't know where ?! Maybe using overlaps of the RASSI-wavefunctions, will also give me the corresponding geometry.
I have no yet experience with such kind of calculations and when I detected such points with a MECP calculation, it was also good luck, if the program has found such points.
Of course the TM-elements, like Zn and Cr almost have degenerate energy levels, so that in connection with Boron, it seems possible, that some levels will have almost the same energy. Of course this is not very reliable,because the electrons are shifted.
So summing up, it is possible to use MECP, either as a CI, or ISC, to find out these points. The latter -ISC- works with 2 different symmetries.
It is also possible to calculate an ISC,but again, I look for a more systematic way of detecting such points.
If the the two states have the same energy, is it also justified to say, that this is a Jahn-Teller Effect ?
I found out, that it is also possible to calculate an ISC with OPENMOLCAS. (Regarding to B-Mn-B, it worked, but there was no "Happy landing",because it was simply a try. )
Usually I look for an IC, not ISC:
ISC with MECP:
coord = BMnB.xyz
basis=6-31G**
Constraints
a = EDiff
value
a =0.000
End of Constraints
>>> DO WHILE
&SEWARD
&SCF
UHF
ZSpin=1
&ALASKA
>>> COPY $WorkDir/$Project.RunFile $WorkDir/RUNFILE2
&SEWARD
&SCF
UHF
ZSpin=3
&ALASKA
&SLAPAF
>>> END DO
The problem is to find out such points in a more systematic way, e.g. loops, with an angle of a=[0,360] plotted via the energies.
What is your opinion to this problem, namly to find out JT-geometries in a more systematic way ? I once read, that these geometries are not an exception in a 3-atomic molecule.
In advance, Very Thank you for your reply.
Best regards
Alfred
Offline
#9 2021-08-11 10:27:07
- Ignacio
- Administrator
- From: Uppsala
- Registered: 2015-11-03
- Posts: 1,085
Re: Some internal inconsistency of the code was detected
Looking for a sign change in a loop will only tell you if there's an odd number of crossing points inside the loop. If your loop does not surround a crossing point you'd see nothing. Just computing the overlaps will also not tell you where the crossing point is, it will only help you find out whether the sign changes or not.
If you want to locate a crossing point, you can locate them both with the same (CI) or different symmetries using the "EDiff" constraint. See https://molcas.gitlab.io/OpenMolcas/sph … onstraints.
Please create a new topic if you have more questions not related to "Some internal inconsistency of the code was detected".
Offline
#10 2021-08-12 09:11:27
- cymantren
- Member
- Registered: 2017-09-18
- Posts: 37
Re: Some internal inconsistency of the code was detected
Dear Ignacio,
yes,of course you are right, that now the headline is not identical with its content and that it is sensible to create a new topic.
I will think about it.
Best regards
Alfred
Offline