Hello. I’m trying to run RAS-2SF but there is a problem.
When I used 6-31G* basis set, there were no problems.
However, for the basis sets that contatin the polarization effect like6-31+G* or 6-311+G*, RAS calculation takes lots of steps than before so it is still not converged.
Here are the rem section of input and RAS calculation process.
In this situation, how can I resolve this problem? Should I just wait until the calculation is finished? Or don’t the basis sets that contain the polarization effect work well for RAS calculations?
Just as a point of clarification, I think you mean diffuse functions (“+”), not polarization functions (“*”). Can you please provide the $molecule? It’s not obvious to me why this happens.
Yes, as you said I intended to say diffuse functions, not polarization functions.
The $molecule that you requested is here.
$molecule
0 1
C -3.4189946735 1.9389485476 -0.6818928456
C -2.0667392405 1.6571706833 -0.9140585237
C -1.6232208331 0.3991900991 -1.3045940474
C -2.5927832173 -0.6552278079 -1.3197602040
C -3.9624392788 -0.3625349968 -1.1009810619
C -4.3609144481 0.9552153043 -0.7956257127
C -2.2084441557 -2.0196405073 -1.4808572977
C -3.1898687311 -2.9978948293 -1.5059536780
C -4.5422582460 -2.6883672635 -1.3449009687
C -4.9306376681 -1.3883217886 -1.1225164866
C -6.3557569989 -1.0943710426 -0.8629466001
N -6.6839239283 0.2218227966 -0.5532619007
C -5.7704742547 1.2651561354 -0.4807719587
C -0.2211041729 0.0746204990 -1.6140803440
C 0.1711928402 -1.2997971745 -1.6343069758
C -0.7811160313 -2.3554695867 -1.5549267075
C 1.5453597643 -1.6234317846 -1.7506675310
C 1.9610776130 -2.9691947770 -1.6945309495
C 1.0280658233 -3.9732269761 -1.5887516877
C -0.3342402006 -3.6660081043 -1.5399567874
C 0.7366487415 1.0335317603 -1.8965701209
C 2.0893041571 0.7006392151 -2.0469817851
C 2.5060965112 -0.6016619672 -1.9235221005
C 3.9472211943 -0.9173595377 -1.9791169707
N 4.2989923181 -2.2618342275 -1.8940992899
C 3.3964513317 -3.3082080763 -1.7325155032
O -7.2103275582 -1.9674418321 -0.8959430684
O -6.1421872631 2.3857200968 -0.1705287920
O 4.8019939584 -0.0475023528 -2.0745590334
O 3.8077948304 -4.4533620944 -1.6221608223
O -1.1842723574 2.6352133796 -0.5436006351
C -1.4592013896 3.8502539210 -2.6033086373
C -1.2123109431 4.9958847180 -3.3505734112
C -0.4877871180 6.0440215843 -2.7948637377
C -0.0167294439 5.9348785773 -1.4916599760
C -0.2471533707 4.7978172901 -0.7100359031
C -0.9868534908 3.7610251898 -1.2979108702
C 5.7117307335 -2.6380875478 -1.9300843580
C -8.0680902174 0.5569685023 -0.2210477431
H -3.7027132446 2.9383101225 -0.3729938688
H -2.9254369165 -4.0386353147 -1.6338328113
H -5.2887580396 -3.4737507765 -1.3568376487
H 1.3607613240 -5.0036617323 -1.5419184415
H -1.0326653474 -4.4875993804 -1.4563646858
H 0.4704700782 2.0721548094 -1.9803898840
H 2.8210245471 1.4799064653 -2.2241288578
H -2.0182132347 3.0230080791 -3.0270477401
H 0.5560079886 6.7474031022 -1.0560560419
H 5.8820776978 -3.3200971794 -2.7656807423
H 6.3040111194 -1.7342541297 -2.0496029577
H 5.9811955246 -3.1373742943 -0.9985564515
H -8.1224014954 0.8804182465 0.8214033148
H -8.6801663541 -0.3283378697 -0.3735547036
H -8.4067221155 1.3718558346 -0.8634251385
H -0.2836927601 6.9401457266 -3.3715500349
H -1.5835458064 5.0594959422 -4.3685222558
C -2.0904438954 0.7039107914 2.0452878175
C -0.7384543107 1.0389161084 1.8936917849
C 0.2210647570 0.0810739860 1.6135814463
C -0.1686519997 -1.2940239297 1.6377474158
C -1.5421761461 -1.6198985179 1.7552761560
C -2.5047985693 -0.5994962462 1.9255636758
C 0.7856392704 -2.3481094233 1.5612116303
C 0.3412225796 -3.6595157067 1.5498242035
C -1.0205065295 -3.9691443555 1.5996721552
C -1.9553746720 -2.9665982023 1.7030004236
C -3.3900921421 -3.3082036416 1.7424986141
N -4.2945767056 -2.2630719738 1.9011491717
C -3.9453124065 -0.9177052816 1.9823518774
C 1.6225450035 0.4074528817 1.3030252859
C 2.5941143108 -0.6450757629 1.3213467731
C 2.2123265763 -2.0097655065 1.4862702101
C 3.9632456432 -0.3504248731 1.1018690277
C 4.9334095834 -1.3742808912 1.1265689178
C 4.5474586894 -2.6743972668 1.3527373071
C 3.1956280670 -2.9860488523 1.5144675599
C 2.0637547165 1.6651111689 0.9087859464
C 3.4154746385 1.9487208679 0.6758393482
C 4.3592673273 0.9671629588 0.7926876844
C 5.7683171497 1.2789323310 0.4773827105
N 6.6837719537 0.2376049949 0.5531789470
C 6.3579632966 -1.0784134077 0.8660195144
O -3.7992885913 -4.4544706468 1.6357570610
O -4.8016549170 -0.0491818756 2.0758565094
O 6.1379473181 2.3992976625 0.1639306220
O 7.2142957861 -1.9496310921 0.9023103856
O 1.1797821976 2.6407085561 0.5353758254
C 1.4508079914 3.8604339783 2.5929278233
C 0.9784938473 3.7670608088 1.2878160708
C 0.2347068283 4.7999845219 0.6982899362
C 0.0000021903 5.9374958145 1.4779771146
C 0.4708600075 6.0507265981 2.7809109110
C 1.1996246015 5.0064052215 3.3382485056
C -5.7065623528 -2.6419599901 1.9389748079
C 8.0674201361 0.5745380433 0.2206342193
H -2.8236022349 1.4823068121 2.2203488826
H -0.4744114142 2.0782641870 1.9746637986
H 1.0411537537 -4.4800323171 1.4683602684
H -1.3512781978 -5.0003228665 1.5556786194
H 5.2954484378 -3.4583253006 1.3670959158
H 2.9331991446 -4.0269096677 1.6455281200
H 3.6972914063 2.9477171528 0.3640299924
H -0.5759810106 6.7470184941 1.0410698593
H 0.2633433644 6.9470254509 3.3560973246
H 1.5708537112 5.0732212964 4.3559914107
H -5.9750179524 -3.1462002889 1.0098402560
H -6.3006410083 -1.7387557650 2.0542672346
H -5.8756093039 -3.3203272782 2.7778161667
H 8.1214367645 0.8957750610 -0.8225122175
H 8.4043953137 1.3914590065 0.8612999396
H 8.6810636652 -0.3093119288 0.3752822705
H 2.0130807799 3.0361164250 3.0180407378
$end
I’m running some tests but I wonder if the RAS code has trouble with linear dependencies. For 6-31+G* basis, I find
Smallest overlap matrix eigenvalue = 5.08E-10
Linear dependence detected in AO basis
whereas for 6-31G*:
Smallest overlap matrix eigenvalue = 9.69E-05
Therefore, can you try your job WITH diffuse functions but adding the following to $rem: BASIS_LIN_DEPTH_THRESH = 14. That will prevent Q-Chem from throwing out the numerical linear dependencies, by setting the threshold for doing so to a very small value (default value is 6). This may adversely affect SCF convergence (a little), which is to be expected and it’s why we discard those functions, but if the behavior with RAS changes significantly then this tells me that the RAS code doesn’t handle that well.
As you adviced the solution, I added the option ‘BASIS_LIN_DEP_THRESH = 14’ to the rem section.
(My QChem version is 5.2 so ‘BASIS_LIN_DEPTH_THRESH’ is not recognized in this system)
The original molecule is very big so I tested for small molecule that is composed of 30 atoms to save the times. When I use the initial rem section for 6-31+G* basis set above, it showed similar phenomenon that takes lots of steps in RAS calculation and finally not converged. However, after ‘BASIS_LIN_DEP_THRESH = 14’ is added, the calculation is successfully finished. Here are the results.
It’s very interesting as only one additional option changed a lot. So I have two questions in this context.
Exactly what does ‘BASIS_LIN_DEPTH_THRESH’ operate in this calculation? and what is the difference between ‘THRESH’ and ‘BASIS_LIN_DEPTH_THRESH’? Actually, I don’t understand the concept of ‘THRESH’ well.
I run other RAS calculation that is set as ‘RAS_SPIN_MULT = 5’ for finding quintet states. In this case, only one state is present so it is converged well without ‘BASIS_LIN_DEPTH_THRESH = 14’. In this situation, should I re-calculate this calculation to match the rem section? I mean, if this option affect the results, it is better to re-calculate all calculations to keep the consistency. What do you think of? Is it necessary or not?
BASIS_LIN_DEPTH_THRESH has been around for 20+ years, should certainly be in Q-Chem 5.x.
THRESH controls integral screening and shell-pair drop tolerances. In my opinion, it is set too low by default, at least for large molecules where linear dependencies may be a problem. I routinely use THRESH = 12 and some of my students use THRESH = 14. For large systems, I’ve seen cases where the SCF converges faster with those tighter settings than with the looser defaults (e.g., coronene dimer, 48 C atoms). The reason is that an SCF cycle takes slightly longer with the tighter THRESH, but the tighter THRESH handles linear dependencies better so converges in fewer cycles.
BASIS_LIN_DEPTH_THRESH is the threshold for removing linear dependencies all together, based on small eigenvalues of the overlap matrix. The SCF code handles this properly but it’s a bit confusing on the code end so I don’t trust that all developers handle this properly, hence my guess that this might be the issue (combined with your observation about diffuse functions - those tend to exacerbate linear dependencies).
I would probably recalculate with a consistent $rem section, would be curious to know if this makes a difference. Do you have a small example where different values of BASIS_LIN_DEPTH_THRESH (default vs. tighter value) changes the results of a RAS calculation? If so, please post the inputs here and I will post a bug ticket, because ideally that should be handled automatically.
It is the input file that you requested. The molecule is generated by myself arbitrarily so there are no chemical meanings. I mean, it is just a molecule for the test.
$molecule
0 1
C 1.57083 0.51532 2.73483
C 2.99879 0.42036 2.68005
C 3.73914 0.30173 3.86900
H 4.82064 0.24820 3.82401
C 3.11990 0.25865 5.11612
C 1.72427 0.34698 5.18135
H 1.20671 0.31681 6.13094
C 0.96976 0.47772 4.01657
H -0.10219 0.52796 4.14029
S 3.97714 0.54246 1.16780
C 4.39212 -1.03181 1.18024
H 3.82287 -1.75800 0.61096
O 3.90139 0.13828 6.22079
C 3.45414 0.08774 7.52484
H 4.32607 -0.01190 8.20323
H 2.79050 -0.79126 7.66959
H 2.91136 1.02257 7.78098
C 0.66757 0.66487 1.52915
C -0.56585 1.35176 1.64706
H -0.87795 1.80095 2.57860
C -1.42212 1.50421 0.55337
H -2.35326 2.04308 0.67822
C -1.08100 0.96951 -0.69098
N -1.95048 1.12236 -1.79890
H -2.84219 1.62992 -1.70211
H -1.70466 0.72898 -2.71919
C 0.12477 0.27785 -0.82939
H 0.39816 -0.15272 -1.78491
C 0.98529 0.12688 0.26196
H 1.87576 -0.44425 0.08975
$end
$rem
JOBTYPE = SP
EXCHANGE = HF
BASIS = 6-31+G*
UNRESTRICTED = FALSE
SYM_IGNORE = TRUE
SYMMETRY = FALSE
SCF_ALGORITHM = DIIS
MAX_SCF_CYCLES = 512
SCF_CONVERGENCE = 8
THRESH = 14
BASIS_LIN_DEP_THRESH = 14
MEM_TOTAL = 24000
MEM_STATIC = 4000
$end
@@@
$molecule
0 5
read
$end
$rem
JOBTYPE = SP
EXCHANGE = HF
CORRELATION = RASCI
BASIS = 6-31+G*
UNRESTRICTED = FALSE
SCF_ALGORITHM = DIIS
MAX_SCF_CYCLES = 512
SCF_CONVERGENCE = 8
THRESH = 14
BASIS_LIN_DEP_THRESH = 14
SCF_GUESS = READ
MEM_TOTAL = 24000
MEM_STATIC = 4000
RAS_ACT = 4
RAS_ELEC = 4
RAS_OCC = 62
RAS_ROOTS = 20
RAS_SPIN_MULT = 1
RAS_NFRAG = 2
RAS_NFRAG_ATOMS = [17,13]
RAS_PRINT = 3
RAS_AMPL_PRINT = 10
SET_ITER = 512
$end
If you delete ‘BASIS_LIN_DEP_THRESH = 14’ and run the job, you can see the strange phenomena that I mentioned. But if you don’t edit it and just submit it to the server, the calculation is finished well. The above results in output file correspond to this case.
So if you finished your own tests, it would be pleasure to tell me the results and the conclusions.(It is just bug, or there are some unknown reasons, etc.) I’m proceeding the calculations for the original molecules that I wanted to calculate, adding the option that you recommended. I will leave whether the calculation is completed or not soon. Thank you for advicing me sincerely.
As I said before, the jobs are running now but it looks bad
Before add ‘BASIS_LIN_DEPTH_THRESH = 14’, SP calculations for singlet and quintet are completed well and RAS calculation showed the convergence problem.
But now, SP calculations show some strange situations.
It is still oscillating now and not converged yet.
And it is the case for other molecule which is similar to the one but has a different linker between the monomers.
RECOMMENDATION:
Set to 5 or smaller if you have a poorly behaved SCF and you suspect linear dependence in you
basis set. Lower values (larger thresholds) may affect the accuracy of the calculation.
The SCF behave bad now so I will try set this value as 3. Or can you recommend other solutions?
This is with BASIS_LIN_DEP_THRESH=14 that you have convergence problems? If so, it’s probably the result of the linear dependencies that are not projected out, which I had thought might somewhat affect SCF convergence but I guess they dramatically affect convergence. Given that the RAS code seems not to be able to handle the reduction in the number of orbitals that comes with projecting out the linear dependencies (for which I will post a bug ticket), I see two possible options:
(1) Try a different SCF_ALGORITHM (maybe GDM) in conjunction with BASIS_LIN_DEP_THRESH=14. This is likely to be very slow, may or may not converge.
(2) Use the default SCF_ALGORITHM and the default BASIS_LIN_DEP_THRESH, which you know will converge the SCF, then do the RAS as a separate job where you read in the previous SCF solution (SCF_GUESS = READ) and set BASIS_LIN_DEP_THRESH = 14.
(1) I set SCF_ALGORITHM = GDM & BASIS_LIN_DEP_THRESH = 14. But as you said, SCF convergence problem occured while calculating for the quintet state.
---------------------------------------
Cycle Energy RMS Gradient
---------------------------------------
1 -3410.7027501269 6.49E-02 Descent step
2 -3410.5107290385 3.95E-02 Descent step
3 -3410.5295807226 2.20E-02 Normal BFGS step
4 -3410.5407606342 1.71E-02 Normal BFGS step
5 -3410.5424929777 1.32E-02 Normal BFGS step
6 -3410.5965796063 8.28E-02 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
7 -3543.7259106730 1.47E+01 Descent step
8 -3549.0168240732 5.83E+03 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
9 -3618.2986729937 2.18E+01 Descent step
10 -3627.1025152337 9.38E+03 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
11 -3777.9466934028 2.71E+01 Descent step
12 -3792.8346949490 1.20E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
13 -3940.8217990483 3.61E+01 Descent step
14 -3955.6601201734 1.59E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
15 -4410.5591453882 2.21E+02 Descent step
GDM::BFGS: rebuilding Hinv to be positive definite
16 -4449.8950719396 5.52E+01 Descent step
17 -4462.8328447650 2.64E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
18 -4639.0229433077 5.04E+01 Descent step
19 -4681.2932460467 2.68E+04 Line search: understep
20 -4848.8888085941 2.79E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
21 -6927.6448515812 7.61E+02 Descent step
22 -7208.5989144640 4.74E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
23 -8446.9066847368 6.00E+01 Descent step
24 -8479.7103907401 3.16E+04 Line search: understep
25 -8739.2351776110 3.12E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
26 -10785.6078526574 1.18E+02 Descent step
27 -10842.8417273695 4.49E+04 Line search: understep
28 -11219.2690954293 4.37E+04 Line search: understep
GDM::BFGS: rebuilding Hinv to be positive definite
29 -15486.6624167050 1.64E+02 Descent step
30 -11948.9287619079 1.17E+04 Line search: overstep
31 -11893.3749619658 1.17E+04 Line search: overstep
32 -11887.7573216195 1.17E+04 Line search: overstep
33 -11887.0901005278 1.17E+04 Line search: overstep
34 -11887.0449902574 1.17E+04 Line search: overstep
35 -11887.0384093086 1.17E+04 Line search: overstep
36 -11887.0405652980 1.17E+04 Line search: overstep
37 -11887.0395587171 1.17E+04 Line search: overstep
38 -11887.0404594997 1.17E+04 Line search: overstep
39 -11887.0400360832 1.17E+04 Line search: overstep
40 -11887.0405718694 1.17E+04 Line search: overstep
41 -11887.0398751656 1.17E+04 Line search: overstep
42 -11887.0402740697 1.17E+04 Line search: overstep
43 -11887.0407863642 1.17E+04 Line search: overstep
44 -11887.0405217940 1.17E+04 Line search: overstep
45 -9281.3451186503 4.90E+04 Line search: overstep
46 -9281.3454202349 4.90E+04 Line search: overstep
47 -9281.3449121351 4.90E+04 Line search: overstep
48 -9281.3451705659 4.90E+04 Line search: overstep
49 -9281.3451531763 4.90E+04 Line search: overstep
50 -9281.3446475426 4.90E+04 Line search: overstep
51 -9281.3452270115 4.90E+04 Line search: overstep
52 -9281.3443463380 4.90E+04 Line search: overstep
53 -9281.3454434881 4.90E+04 Line search: overstep
54 -9281.3452541465 4.90E+04 Line search: overstep
55 -9281.3452791879 4.90E+04 Line search: overstep
56 -9281.3451838205 4.90E+04 Line search: overstep
57 -9281.3451145227 4.90E+04 Line search: overstep
58 -9281.3454879383 4.90E+04 Line search: overstep
59 -9281.3453500802 4.90E+04 Line search: overstep
60 -10164.6995870034 4.60E+04 Line search: overstep
61 -10164.6999759907 4.60E+04 Line search: overstep
62 -10164.6998834992 4.60E+04 Line search: overstep
63 -10164.7000929201 4.60E+04 Line search: overstep
64 -10164.6995922262 4.60E+04 Line search: overstep
65 -10164.6992089609 4.60E+04 Line search: overstep
66 -10164.6995597864 4.60E+04 Line search: overstep
67 -10164.7000751133 4.60E+04 Line search: overstep
68 -10164.6993737752 4.60E+04 Line search: overstep
69 -10164.6996568824 4.60E+04 Line search: overstep
70 -10164.6995845223 4.60E+04 Line search: overstep
71 -10164.6996860663 4.60E+04 Line search: overstep
72 -10164.7000382478 4.60E+04 Line search: overstep
73 -10164.6997192065 4.60E+04 Line search: overstep
74 -10164.7001619721 4.60E+04 Line search: overstep
75 -10501.5410063351 4.38E+04 Line search: overstep
76 -10501.5411474416 4.38E+04 Line search: overstep
77 -10501.5420132140 4.38E+04 Line search: overstep
78 -10501.5418170775 4.38E+04 Line search: overstep
79 -10501.5411902601 4.38E+04 Line search: overstep
80 -10501.5411484102 4.38E+04 Line search: overstep
81 -10501.5414508218 4.38E+04 Line search: overstep
82 -10501.5412611003 4.38E+04 Line search: overstep
83 -10501.5410817350 4.38E+04 Line search: overstep
84 -10501.5416754568 4.38E+04 Line search: overstep
85 -10501.5415732200 4.38E+04 Line search: overstep
86 -10501.5419409955 4.38E+04 Line search: overstep
87 -10501.5418174426 4.38E+04 Line search: overstep
88 -10501.5415610308 4.38E+04 Line search: overstep
89 -10501.5419276143 4.38E+04 Line search: overstep
90 -10442.1821419457 4.45E+04 Line search: overstep
91 -10442.1820081408 4.45E+04 Line search: overstep
92 -10442.1817749078 4.45E+04 Line search: overstep
93 -10442.1819083030 4.45E+04 Line search: overstep
94 -10442.1818180168 4.45E+04 Line search: overstep
95 -10442.1813120926 4.45E+04 Line search: overstep
96 -10442.1814675862 4.45E+04 Line search: overstep
97 -10442.1817407990 4.45E+04 Line search: overstep
98 -10442.1816326390 4.45E+04 Line search: overstep
99 -10442.1816465194 4.45E+04 Line search: overstep
100 -10442.1816376904 4.45E+04 Line search: overstep
101 -10442.1814673627 4.45E+04 Line search: overstep
102 -10442.1819388429 4.45E+04 Line search: overstep
103 -10442.1815400655 4.45E+04 Line search: overstep
104 -10442.1820737804 4.45E+04 Line search: overstep
105 -11926.1967482130 5.47E+04 Line search: overstep
106 -11926.1964063133 5.47E+04 Line search: overstep
107 -11926.1971641714 5.47E+04 Line search: overstep
108 -11926.1970221559 5.47E+04 Line search: overstep
109 -11926.1968003223 5.47E+04 Line search: overstep
110 -11926.1978947456 5.47E+04 Line search: overstep
111 -11926.1969638030 5.47E+04 Line search: overstep
112 -11926.1975288252 5.47E+04 Line search: overstep
113 -11926.1973312730 5.47E+04 Line search: overstep
114 -11926.1971033598 5.47E+04 Line search: overstep
115 -11926.1971156383 5.47E+04 Line search: overstep
116 -11926.1967938031 5.47E+04 Line search: overstep
117 -11926.1968125562 5.47E+04 Line search: overstep
118 -11926.1972103948 5.47E+04 Line search: overstep
119 -11926.1971679712 5.47E+04 Line search: overstep
120 -9442.6325492049 4.73E+04 Line search: overstep
121 -9442.6326112086 4.73E+04 Line search: overstep
122 -9442.6328994940 4.73E+04 Line search: overstep
123 -9442.6328050951 4.73E+04 Line search: overstep
124 -9442.6322731683 4.73E+04 Line search: overstep
125 -9442.6327153007 4.73E+04 Line search: overstep
126 -9442.6321481923 4.73E+04 Line search: overstep
127 -9442.6326976354 4.73E+04 Line search: overstep
128 -9442.6324435333 4.73E+04 Line search: overstep
129 -9442.6328266943 4.73E+04 Line search: overstep
130 -9442.6318965415 4.73E+04 Line search: overstep
131 -9442.6326150010 4.73E+04 Line search: overstep
132 -9442.6330595473 4.73E+04 Line search: overstep
133 -9442.6328402544 4.73E+04 Line search: overstep
134 -9442.6324464540 4.73E+04 Line search: overstep
135 -10219.2245769063 5.19E+04 Line search: overstep
136 -10219.2249123538 5.19E+04 Line search: overstep
137 -10219.2244806076 5.19E+04 Line search: overstep
138 -10219.2246835912 5.19E+04 Line search: overstep
139 -10219.2245009551 5.19E+04 Line search: overstep
140 -10219.2241605903 5.19E+04 Line search: overstep
141 -10219.2243908206 5.19E+04 Line search: overstep
142 -10219.2245161692 5.19E+04 Line search: overstep
143 -10219.2245301838 5.19E+04 Line search: overstep
144 -10219.2247283617 5.19E+04 Line search: overstep
145 -10219.2246296788 5.19E+04 Line search: overstep
146 -10219.2246528352 5.19E+04 Line search: overstep
147 -10219.2245707820 5.19E+04 Line search: overstep
148 -10219.2249623994 5.19E+04 Line search: overstep
149 -10219.2242918769 5.19E+04 Line search: overstep
150 -9041.7321751873 4.67E+04 Line search: overstep
151 -9041.7316349233 4.67E+04 Line search: overstep
152 -9041.7319265688 4.67E+04 Line search: overstep
153 -9041.7317587892 4.67E+04 Line search: overstep
154 -9041.7316784198 4.67E+04 Line search: overstep
155 -9041.7319387952 4.67E+04 Line search: overstep
156 -9041.7320318455 4.67E+04 Line search: overstep
157 -9041.7322466756 4.67E+04 Line search: overstep
158 -9041.7314078147 4.67E+04 Line search: overstep
159 -9041.7324134345 4.67E+04 Line search: overstep
160 -9041.7317236895 4.67E+04 Line search: overstep
161 -9041.7320108051 4.67E+04 Line search: overstep
162 -9041.7316697622 4.67E+04 Line search: overstep
163 -9041.7313183556 4.67E+04 Line search: overstep
164 -9041.7320361967 4.67E+04 Line search: overstep
165 -10248.6030163179 4.60E+04 Line search: overstep
166 -10248.6027600849 4.60E+04 Line search: overstep
167 -10248.6027619997 4.60E+04 Line search: overstep
168 -10248.6027172590 4.60E+04 Line search: overstep
169 -10248.6030238057 4.60E+04 Line search: overstep
170 -10248.6033090139 4.60E+04 Line search: overstep
171 -10248.6030545244 4.60E+04 Line search: overstep
172 -10248.6032024855 4.60E+04 Line search: overstep
173 -10248.6028502966 4.60E+04 Line search: overstep
174 -10248.6031582067 4.60E+04 Line search: overstep
175 -10248.6034172560 4.60E+04 Line search: overstep
176 -10248.6034484068 4.60E+04 Line search: overstep
177 -10248.6030356521 4.60E+04 Line search: overstep
178 -10248.6031993116 4.60E+04 Line search: overstep
179 -10248.6032293151 4.60E+04 Line search: overstep
180 -10443.7167950655 4.87E+04 Line search: overstep
181 -10443.7166594798 4.87E+04 Line search: overstep
182 -10443.7168608094 4.87E+04 Line search: overstep
183 -nan NAN Convergence criterion met
---------------------------------------
SCF time: CPU 346286.73 s wall 46660.56 s
Info = 3
Q-Chem warning in module liblas/diagon.C, line 97:
Call to dsyevd failed in Diagonalize
Info = 3
Q-Chem fatal error occurred in module liblas/diagon.C, line 105:
Call to dsyev also failed in Diagonalize
Please submit a crash report at q-chem.com/reporter
(2) I used the default SCF_ALGORITHM and the default BASIS_LIN_DEP_THRESH, and add SCF_GUESS = READ and BASIS_LIN_DEP_THRESH = 14 in the RAS but the problem of inconsistent size for SCF MO coefficient file occured.
Q-Chem warning in module stvman/mkSTV.C, line 318:
Overlap eigenvalue is smaller than square root of threshold
Use BASIS_LIN_DEP_THRESH <= 7 and THRESH = 14 to get rid of this warning
Scale SEOQF with 1.000000e-02/1.000000e-02/1.000000e-01
Standard Electronic Orientation quadrupole field applied
Nucleus-field energy = -0.0000000883 hartrees
Guess MOs from SCF MO coefficient file
Reading MOs from coefficient file
Size of previous SCF MO coefficient file: 6613884 (expected); 6406518 (actual).
Q-Chem warning in module gesman/GuessMan.C, line 786:
Inconsistent size for SCF MO coefficient file. Please check basis and PURECART setting.
Q-Chem fatal error occurred in module libmdc/newfileman.C, line 384:
FileMan error: End of file reached prematurely reading (14544) bytes in file FILE_MO_COEFS
Path: /work/206453.dinger00/qchem1996/53.0
Please submit a crash report at q-chem.com/reporter
(3) Finally, as I said before, I’m trying to calculate utilizing BASIS_LIN_DEP_THRESH = 3 and it looks good now. But it is not correct as lower BASIS_LIN_DEP_THRESH derive the loose results so SCF energies in the singlet state and the quintet state are different from the former calculations. Before, SCF energy of the sinlget state was -3410.7174345920 but this calculation give the value as -3410.6975644867. So I’m not sure this result is reliable although it is converged well.
Okay, so (2) didn’t work precisely because the two calculations have different numbers of virtual MOs, and even though that doesn’t fundamentally matter there’s no user-facing way to circumvent this (at least not one that comes to my mind).
I don’t completely understand what you mean in (3) but the absolute energies will change when linear dependencies are removed, the question is whether this has any effect on observables.
I have submitted a bug ticket and copied the developer, not sure what else to do at this point except to ask whether these calculations really need diffuse functions.
