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 2022-05-25 04:34:06
- xyin
- Member
- Registered: 2022-05-12
- Posts: 6
How to solve the low Reference weight？
Dear Molcas users:
In my system, I set the active space to cas(14,14), which I think already contains all pairs of pi orbitals, but the reference state is still very small, only about 0.51684.
According to the manual, i think the active space lacks an important orbital, however, I can't find it. Should I keep increasing the active space?
Also, I saw the instructions forChecking the list of large perturbative contributions (small denominators combined with large RHS values; check the output) and also the occupation number of the CASPT2 orbitals. Can this information tell us what the missing orbitals are?
Any hints would be very helpful! Thank you!!
Here is my input:
&RASSCF
Title = 1413-defTZVP-S0
FILEORB = 36-deftzvp_uhf_uno_asrot2gvb41_s.INPORB
Symmetry = 1
Charge = 0
Spin = 1
RAS2 = 14
nActE1 = 14 0 0
CiRoot = 5 5 1
Rlxroot= 1
&CASPT2
Multistate = 5 1 2 3 4 5
Imaginary Shift = 0.2
IPEA=0Here is part of my output:
Compute H0 matrices for state 1
--------------------------------------------------------------------------------
With new orbitals, the CI array of state 1
--------------------------------------------------------------------------------
CI COEFFICIENTS LARGER THAN 0.50D-01
Occupation of active orbitals, and spin coupling
of open shells. (u,d: Spin up or down).
SGUGA info is (Midvert:IsyUp:UpperWalk/LowerWalk)
Conf SGUGA info Occupation Coefficient Weight
1 ( 1:1: 1/ 1) 22222220000000 0.896759 0.804177
240 ( 3:1: 22/ 7) u22222du00000d 0.060379 0.003646
247 ( 3:1: 1/ 8) 22222022000000 -0.057418 0.003297
248 ( 3:1: 2/ 8) 2222202ud00000 -0.066496 0.004422
420 ( 3:1: 6/ 14) 22220220020000 -0.050547 0.002555
424 ( 3:1: 10/ 14) 22220220002000 -0.054110 0.002928
681 ( 3:1: 15/ 23) 22022220000200 -0.078021 0.006087
771 ( 3:1: 21/ 26) 20222220000020 -0.086661 0.007510
807 ( 3:1: 1/ 28) 02222222000000 -0.055729 0.003106
813 ( 3:1: 7/ 28) 0222222u00d000 -0.055319 0.003060
Constructing G3/F3
memory avail: 29.422625160 GB
memory used: 4.394899080 GB
Sym: 1, #Tasks: 196
--------------------------------------------------------------------------------
H0 matrices have been computed.
********************************************************************************
CASPT2 EQUATION SOLUTION
--------------------------------------------------------------------------------
Computing the S/B matrices
--------------------------
Construct S matrices
Construct B matrices
Find transformation matrices to eigenbasis of block-diagonal part of H0.
Eliminate linear dependency. Thresholds for:
Initial squared norm : 0.1000E-09
Eigenvalue of scaled S: 0.1000E-07
Condition numbers are computed after diagonal scaling and after removal of
linear dependency. Resulting sizes, condition numbers, and times:
CASE(SYM) NASUP NISUP NINDEP COND NR CPU (s)
VJTU (1) 2744 46 2729 0.40E+09 63
VJTIP (1) 105 1081 105 0.14E+06 0
VJTIM (1) 91 1035 91 0.42E+04 0
ATVX (1) 2744 408 2729 0.38E+09 63
AIVX (1) 392 18768 391 0.23E+07 1
VJAIP (1) 14 441048 14 18. 0
VJAIM (1) 14 422280 14 18. 0
BVATP (1) 105 83436 105 0.17E+06 0
BVATM (1) 91 83028 91 0.55E+04 0
BJATP (1) 14 3838056 14 13. 0
BJATM (1) 14 3819288 14 13. 0
Total nr of CASPT2 parameters:
Before reduction: 320544554
After reduction: 320518976
Computing the right-hand side (RHS) elements
--------------------------------------------
Using conventional MKRHS algorithm
The contributions to the second order correlation energy in atomic units.
-----------------------------------------------------------------------------------------------------------------------------
IT. VJTU VJTI ATVX AIVX VJAI BVAT BJAT BJAI TOTAL RNORM
-----------------------------------------------------------------------------------------------------------------------------
1 -0.007522 -0.012194 -0.023466 -0.280750 -0.153993 -0.053235 -0.434634 -1.960362 -2.926157 0.026003
2 -0.007791 -0.012451 -0.023763 -0.283598 -0.154832 -0.053463 -0.435415 -1.961336 -2.932651 0.005745
3 -0.007803 -0.012461 -0.023781 -0.283819 -0.154865 -0.053483 -0.435483 -1.961299 -2.932993 0.001359
4 -0.007803 -0.012459 -0.023767 -0.283718 -0.154822 -0.053472 -0.435430 -1.961244 -2.932715 0.000332
5 -0.007802 -0.012459 -0.023767 -0.283710 -0.154820 -0.053471 -0.435426 -1.961243 -2.932698 0.000090
6 -0.007803 -0.012459 -0.023768 -0.283714 -0.154822 -0.053472 -0.435429 -1.961247 -2.932713 0.000020
7 -0.007803 -0.012459 -0.023768 -0.283714 -0.154822 -0.053472 -0.435430 -1.961247 -2.932714 0.000005
8 -0.007803 -0.012459 -0.023768 -0.283714 -0.154822 -0.053472 -0.435429 -1.961247 -2.932713 0.000001
-----------------------------------------------------------------------------------------------------------------------------
FINAL CASPT2 RESULT:
Correlation energy /Case, /Symm, and sums:
VJTU -0.00780260 -0.00780260
VJTIP -0.01114800 -0.01114800
VJTIM -0.00131127 -0.00131127
ATVX -0.02376766 -0.02376766
AIVX -0.28371399 -0.28371399
VJAIP -0.08165358 -0.08165358
VJAIM -0.07316839 -0.07316839
BVATP -0.04762963 -0.04762963
BVATM -0.00584221 -0.00584221
BJATP -0.18145329 -0.18145329
BJATM -0.25397609 -0.25397609
BJAIP -1.28407060 -1.28407060
BJAIM -0.67717606 -0.67717606
Summed: -2.93271337 -2.93271337
Reference energy: -931.9417673901
E2 (Non-variational): -2.9327133671
Shift correction: -0.0146913759
E2 (Variational): -2.9474047430
Total energy: -934.8891721330
Residual norm: 0.0000003182
Reference weight: 0.51684
Contributions to the CASPT2 correlation energy
Active & Virtual Only: -0.0772395017
One Inactive Excited: -0.7269459655
Two Inactive Excited: -2.1285278999
++ Denominators, etc.
--------------------------------------------------------------------------------------------------------------
Report on small energy denominators, large coefficients, and large energy contributions.
The ACTIVE-MIX index denotes linear combinations which gives ON expansion functions
and makes H0 diagonal within type.
DENOMINATOR: The (H0_ii - E0) value from the above-mentioned diagonal approximation.
RHS VALUE : Right-Hand Side of CASPT2 Eqs.
COEFFICIENT: Multiplies each of the above ON terms in the first-order wave function.
Thresholds used:
Denominators: 0.3000
Coefficients: 0.0250
Energy contributions: 0.0050
CASE SYMM ACTIVE-MIX NON-ACTIVE INDICES DENOMINATOR RHS VALUE COEFFICIENT CONTRIBUTION
VJTU 1 Mu1.0001 In1.063 0.29801604 -0.00019482 0.00054100 -0.00000011
VJTU 1 Mu1.0001 In1.064 0.29862795 0.00192546 -0.00521200 -0.00001004
VJTU 1 Mu1.0001 In1.065 0.29119750 -0.00086287 0.00128861 -0.00000111
VJTU 1 Mu1.0001 In1.066 0.28239780 -0.00000333 -0.00003271 0.00000000
VJTU 1 Mu1.0001 In1.067 0.20649505 -0.00049922 0.00104040 -0.00000052
VJTU 1 Mu1.0002 In1.067 0.27727754 -0.00149909 0.00427165 -0.00000640
ATVX 1 Mu1.0001 Se1.082 0.17309653 -0.00000850 0.00010989 -0.00000000
ATVX 1 Mu1.0002 Se1.082 0.24357274 0.00003858 -0.00011066 -0.00000000
ATVX 1 Mu1.0003 Se1.082 0.26376109 -0.00002746 0.00009133 -0.00000000
ATVX 1 Mu1.0001 Se1.083 0.19017446 0.00000305 0.00001209 0.00000000
ATVX 1 Mu1.0002 Se1.083 0.26065067 0.00000895 -0.00002753 -0.00000000
ATVX 1 Mu1.0003 Se1.083 0.28083902 -0.00000243 0.00001580 -0.00000000
ATVX 1 Mu1.0001 Se1.084 0.20072498 -0.00001141 0.00004780 -0.00000000
ATVX 1 Mu1.0002 Se1.084 0.27120119 0.00000407 -0.00001779 -0.00000000
ATVX 1 Mu1.0003 Se1.084 0.29138955 0.00001241 -0.00001107 -0.00000000
ATVX 1 Mu1.0001 Se1.085 0.20679027 0.00003422 -0.00005125 -0.00000000
ATVX 1 Mu1.0002 Se1.085 0.27726648 0.00000748 -0.00002755 -0.00000000
ATVX 1 Mu1.0003 Se1.085 0.29745484 -0.00000368 0.00002617 -0.00000000
ATVX 1 Mu1.0001 Se1.086 0.21915436 -0.00015999 0.00031678 -0.00000005
ATVX 1 Mu1.0002 Se1.086 0.28963057 0.00002305 -0.00002200 -0.00000000
ATVX 1 Mu1.0001 Se1.087 0.23324696 -0.00000327 -0.00005057 0.00000000
ATVX 1 Mu1.0001 Se1.088 0.22754391 0.00006061 -0.00015034 -0.00000001
ATVX 1 Mu1.0002 Se1.088 0.29802012 -0.00001854 0.00003912 -0.00000000
ATVX 1 Mu1.0001 Se1.089 0.23549626 0.00000270 -0.00007686 -0.00000000
ATVX 1 Mu1.0001 Se1.090 0.25468256 0.00015777 -0.00028444 -0.00000004
ATVX 1 Mu1.0001 Se1.091 0.24762479 0.00000124 -0.00001393 -0.00000000
ATVX 1 Mu1.0001 Se1.092 0.26433438 0.00025977 -0.00048163 -0.00000013
ATVX 1 Mu1.0001 Se1.093 0.26992284 -0.00159938 0.00288276 -0.00000461
ATVX 1 Mu1.0001 Se1.094 0.27060013 -0.00019994 0.00033916 -0.00000007
ATVX 1 Mu1.0001 Se1.095 0.27848273 -0.00016967 0.00031276 -0.00000005
ATVX 1 Mu1.0001 Se1.096 0.28441576 -0.00004381 0.00006268 -0.00000000
ATVX 1 Mu1.0001 Se1.097 0.28951388 0.00004381 -0.00007834 -0.00000000
ATVX 1 Mu1.0001 Se1.098 0.29807393 -0.00054168 0.00108759 -0.00000059
--
(Skipping property calculation,
use PROP keyword to activate)
********************************************************************************
CASPT2 MULTI-STATE COUPLINGS SECTION
Hamiltonian Effective Couplings
-------------------------------
| 1 >
< 1 | -9.34889172133045E+02
< 2 | -1.44516943616683E-02
< 3 | 1.32260584237542E-02
< 4 | -3.03190823400640E-03
< 5 | 3.28160507115732E-03
Time spent for multi-state couplings for root 1:
----------------- CPU TIME -------- WALL TIME
< 1 | 0.000 0.000
< 2 | 23100.840 1214.150
< 3 | 35556.840 1892.180
< 4 | 31698.370 1666.220
< 5 | 32604.220 1725.720
CASPT2 TIMING INFO FOR STATE 1
cpu time (s) wall time (s)
------------- -------------
Group initialization 24327.86 1884.49
- Fock matrix build 13392.95 1056.10
- integral transforms 10904.87 809.98
State initialization 22874.35 1169.67
- density matrices 22874.23 1169.66
CASPT2 equations 12249.10 3682.84
- H0 S/B matrices 0.33 0.46
- H0 S/B diag 406.52 30.13
- H0 NA diag 3.18 0.49
- RHS construction 4.00 9.06
- PCG solver 11834.38 3639.00
- scaling 22.78 28.78
- lin. comb. 887.32 122.80
- inner products 1193.66 135.15
- basis transforms 961.77 85.80
- sigma routines 8704.85 3200.62
- array collection 0.00 0.00
Properties 0.00 0.00
Gradient/MS coupling 122960.27 6502.51
Total time 182411.58 13239.51Last edited by xyin (2022-05-25 04:35:21)
Offline
#2 2022-05-25 07:49:33
- Ignacio
- Administrator
- From: Uppsala
- Registered: 2015-11-03
- Posts: 1,011
Re: How to solve the low Reference weight？
What makes you think the reference weight is "very small"? You should expect it to decrease as your total number of electrons increases. A reference weight of ~0.8 is common for small organic molecules, but you won't see such values in larger molecules.
The common estimate is (https://books.google.se/books?id=40DqCA … &q&f=false):
w = (1+alpha)^(-N/2)
where N is the number of correlated electrons (i.e. not including those in the frozen core orbitals), and alpha is a constant of around 0.015 (related to the average correlation energy per electron pair).
Your value looks fine if you have around 90 electrons, but it could also be reasonable for anything between 65 and 130 electrons (it just depends on which value you use for alpha, which is arbitrary). More importantly, there are no large contributions or coefficients in your "Denominators etc." list, so I see no reason to suspect problems
Offline
#3 2022-05-26 03:27:10
- xyin
- Member
- Registered: 2022-05-12
- Posts: 6
Re: How to solve the low Reference weight？
What makes you think the reference weight is "very small"? You should expect it to decrease as your total number of electrons increases. A reference weight of ~0.8 is common for small organic molecules, but you won't see such values in larger molecules.
The common estimate is (https://books.google.se/books?id=40DqCA … &q&f=false):
w = (1+alpha)^(-N/2)
where N is the number of correlated electrons (i.e. not including those in the frozen core orbitals), and alpha is a constant of around 0.015 (related to the average correlation energy per electron pair).
Your value looks fine if you have around 90 electrons, but it could also be reasonable for anything between 65 and 130 electrons (it just depends on which value you use for alpha, which is arbitrary). More importantly, there are no large contributions or coefficients in your "Denominators etc." list, so I see no reason to suspect problems
Thank you so much for your reply!
Offline