计算化学公社

标题: CASSCF L510报错 [打印本页]

作者
Author: nuke 时间: 2017-12-6 10:29
标题: CASSCF L510报错
本帖最后由 nuke 于 2017-12-6 10:29 编辑

利用Exploring-chemistry-with-electric-structure-methods 第二版第九章的Advanced Exercise 9.6学习conical intersection。
其中9_06a.gjf和9_06b.gjf顺利运行，但是9_06c.gjf在用g09w计算的时候会报错，错误信息如下：

ITN= 59 MaxIt= 64 E= -154.6853255067 DE= 1.50D-05 Acc= 1.00D-08 Lan=  0
ITN= 60 MaxIt= 64 E= -154.6853108721 DE= 1.46D-05 Acc= 1.00D-08 Lan=  0
ITN= 61 MaxIt= 64 E= -154.6852966375 DE= 1.42D-05 Acc= 1.00D-08 Lan=  0
ITN= 62 MaxIt= 64 E= -154.6852828683 DE= 1.38D-05 Acc= 1.00D-08 Lan=  0
ITN= 63 MaxIt= 64 E= -154.6852696183 DE= 1.33D-05 Acc= 1.00D-08 Lan=  0
ITN= 64 MaxIt= 64 E= -154.6852569291 DE= 1.27D-05 Acc= 1.00D-08 Lan=  0
Convergence failure -- run terminated.
Error termination via Lnk1e in C:\G09W\l510.exe at Wed Dec 06 10:03:07 2017.
Job cpu time:  0 days  0 hours  0 minutes 16.0 seconds.
File lengths (MBytes):  RWF=    5 Int=    0 D2E=    0 Chk=    1 Scr=    1

不知这个问题如何解决谢谢！-----------------------------------------------------------------
我在ubuntu下也用g16测试了一下，9_06c.gjf可以通过，但9_06d.gjf，依旧是l510的错误。
Davidson Disk Diagonalization is being used:
MVecIn=    4 MSekIn= 4 NVConv= 2 NVConT= 0 MConv= 2 WhenSc=  2 IFact= 2 FinIt=F
NRoot=    4 Conv= 0.00D+00 ConvT= 0.00D+00 MaxItI=          0 Dim=       120 Max sub= 200
Iteration    1 internal iteration    1 dimension is    4
MCAXMl: NEAAct < 3 not allowed
Error termination via Lnk1e in /home/nuke/Documents/g16/l510.exe at Tue Dec  5 21:28:08 2017.
Job cpu time:    0 days  0 hours  0 minutes  0.3 seconds.
Elapsed time:    0 days  0 hours  0 minutes  0.3 seconds.
File lengths (MBytes):  RWF=    6 Int=    0 D2E=    0 Chk=    1 Scr=    1
Error: segmentation violation
rax 0000000000000000, rbx ffffffffffffffff, rcx 00002ade1cdae767
rdx 00000000000008e5, rsp 00007ffc4647d608, rbp 00007ffc4647db80
rsi 000000000000000b, rdi 00000000000008e5, r8  0000000000000020

---------------------------------------------------
我查了一下有的说是CASSCF参数选取问题，为了例子能在windows下运行，我把9_06c.gjf的参数改为
#T CASSCF(6,8,StateAverage,NRoot=3)/3-21G Guess=Read Geom=Allcheck Test
但依旧报错，
ITN= 55 MaxIt= 64 E= -154.7185562196 DE=-5.93D-07 Acc= 1.00D-08 Lan=  0
ITN= 56 MaxIt= 64 E= -154.7185618596 DE=-5.64D-06 Acc= 1.00D-08 Lan=  0
ITN= 57 MaxIt= 64 E= -154.7185700335 DE=-8.17D-06 Acc= 1.00D-08 Lan=  0
ITN= 58 MaxIt= 64 E= -154.7185773385 DE=-7.31D-06 Acc= 1.00D-08 Lan=  0
ITN= 59 MaxIt= 64 E= -154.7185842049 DE=-6.87D-06 Acc= 1.00D-08 Lan=  0
ITN= 60 MaxIt= 64 E= -154.7185905191 DE=-6.31D-06 Acc= 1.00D-08 Lan=  0
ITN= 61 MaxIt= 64 E= -154.7185963397 DE=-5.82D-06 Acc= 1.00D-08 Lan=  0
ITN= 62 MaxIt= 64 E= -154.7186017178 DE=-5.38D-06 Acc= 1.00D-08 Lan=  0
ITN= 63 MaxIt= 64 E= -154.7186067000 DE=-4.98D-06 Acc= 1.00D-08 Lan=  0
ITN= 64 MaxIt= 64 E= -154.7186113236 DE=-4.62D-06 Acc= 1.00D-08 Lan=  0
Convergence failure -- run terminated.
Error termination via Lnk1e in C:\G09W\l510.exe at Wed Dec 06 09:59:14 2017.
Job cpu time:  0 days  0 hours 13 minutes 21.0 seconds.
File lengths (MBytes):  RWF=    48 Int=    0 D2E=    0 Chk=    1 Scr=    1

在下面附上四个书上对应的输入文件：
输入文件为：
9_06a.gjf：
%Chk=cas_gs
#T RHF/3-21G Test

Ground State C2V butadiene orbitals

0,1
C    .000000 .740576       .570074
C    .000000 -.740576       .570074
C    .000000 1.551071       -.509337
C    .000000 -1.551071    -.509337
H    .000000 1.175386       -1.517998
H    .000000 2.620804       -.392120
H    .000000 1.199468       1.545694
H    .000000 -1.199468    1.545694
H    .000000 -2.620804    -.392120
H    .000000 -1.175386    -1.517998

9_06b.gjf：
%Chk=cas_gs
#T CASSCF(4,4,StateAverage,NRoot=3)/3-21G Guess=(Alter,Read) Test

Ground state for butadiene

0,1
C    .000000 .740576       .570074
C    .000000 -.740576       .570074
C    .000000 1.551071       -.509337
C    .000000 -1.551071    -.509337
H    .000000 1.175386       -1.517998
H    .000000 2.620804       -.392120
H    .000000 1.199468       1.545694
H    .000000 -1.199468    1.545694
H    .000000 -2.620804    -.392120
H    .000000 -1.175386    -1.517998

17,18
18,29
19,31
20,32
21,35

! Weight all three states equally.
0.3333333 0.3333333 0.3333333

--Link1--
%Chk=cas_gs
#T CASSCF(4,4,StateAverage,NRoot=3)/4-31G Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

--Link1--
%Chk=cas_gs
#T CASSCF(4,4,StateAverage,NRoot=3)/6-31G(d) Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

--Link1--
%Chk=cas_gs
#P CASSCF(4,4,StateAverage,NRoot=3)/6-31+G(d,p) Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

9_06c.gjf：
%Chk=cas_gs
#T CASSCF(4,6,StateAverage,NRoot=3)/3-21G Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

--Link1--
%Chk=cas_gs
#T CASSCF(4,6,StateAverage,NRoot=3)/4-31G Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

--Link1--
%Chk=cas_gs
#T CASSCF(4,6,StateAverage,NRoot=3)/6-31G(d) Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

--Link1--
%Chk=cas_gs
%NoSave
#P CASSCF(4,6,StateAverage,NRoot=3)/6-31+G(d,p) Guess=Read Geom=Allcheck Test

0.3333333 0.3333333 0.3333333

9_06d.gjf：
%Chk=cas_buta
#T UHF/STO-3G Test Guess=Mix Pop=NaturalOrbitals NoSymm

Conical Intersection: UHF/STO-3G

0 1
6          .000000    .000000    .702859
6          .000000    .000000 -.702859
6          -.631078 1.104055 1.471080
6          -.723500 -1.265744 -.892781
1          .458274 -.840810 1.204515
1          -.240104    .872255 -1.270306
1       -1.608859 1.448944 1.172244
1          -.027452 1.808388 2.026115
1          -.272487 -2.190375 -.587305
1       -1.783159 -1.274584 -1.06726

--Link1--
%Chk=cas_buta
#T CASSCF(4,4,UNO,NRoot=2,StateAverage,NoFullDiag)/STO-3G Test
Guess=Read Geom=Check NoSymm

Conical Intersection: State-averaged CAS with UNO option
                  to get starting orbitals

0 1

! Weight the two states equally
0.5    0.5

--Link1--
%Chk=cas_buta
#T CASSCF(4,4,StateAverage,NRoot=2)/4-31G Geom=Check
Guess=Read NoSymm Test

Conical Intersection: Converge with extended basis set

0 1

0.5    0.5

--Link1--
%Chk=cas_buta
#T CASSCF(4,4)/4-31G Opt=Conical Test
Geom=Check Guess=Read NoSymm IOp(1/8=5)

Conical Intersection optimization

0 1

--Link1--
%Chk=cas_buta
%NoSave
#P CASSCF(4,4,StateAverage,NRoot=2)/4-31G Test Geom=Check Guess=Read

Energies

0 1

0.5    0.5

作者
Author: 一颗赛艇 时间: 2017-12-6 18:00
高斯的casscf迭代循环最大64次，而且好像也没法改。我的建议是换molcas

作者
Author: sobereva 时间: 2017-12-6 18:19

一颗赛艇发表于 2017-12-6 18:00
高斯的casscf迭代循环最大64次，而且好像也没法改。我的建议是换molcas

IOp(5/7=x)可以改最大迭代次数

作者
Author: zjxitcc 时间: 2017-12-6 19:18
高斯CASSCF可以改迭代圈数，就直接用scf(maxcycle=500)就可以了，它虽然是SCF的圈数，也对CASSCF生效。前提是你得确定你的活性空间是对的，并且能量有收敛的趋势，否则加大圈数也是白搭。另外，如果活性空间不超过(10,10)，能量有收敛的趋势，但100圈内没收敛或者有略微的振荡，还可以考虑用fulldiag。

作者
Author: nuke 时间: 2017-12-6 22:00

感谢指点，这个命令确实很管用。
不过9_06d.gjf 继续报错还是L510的。大概意思是
MCAXMl: NEAAct < 3 not allowed
Error termination via Lnk1e in /home/nuke/Documents/g16/l510.exe at Wed Dec  6 08:34:26 2017.
MCAXMl这个单词百度也很难找到。。。

out文件的最后30行如下
NBasis= 26 NCore= 13 NVal= 4 NVirt= 9 IFrame=1 Acc=  1.00D-08.
  Memory needed for Incore Integrals:                323920
  Integrals KEPT IN MEMORY
IBUJAK length=                   7722
Integral file not found: evaluate integrals
FoFCou: FMM=F IPFlag=          0 FMFlag=          0 FMFlg1=          0
      NFxFlg=          0 DoJE=F BraDBF=F KetDBF=F FulRan=T
      wScrn=  0.000000 ICntrl=    600 IOpCl=  0 I1Cent=          0 NGrid=          0
      NMat0= 1 NMatS0=    1 NMatT0= 0 NMatD0= 1 NMtDS0= 0 NMtDT0= 0
Symmetry not used in FoFCou.
In Davids. NSec=       21
IMaxHF,MaxMI,MSekIn,IFactt,WhenSc,nroot 5  200 4 2 2 2
Store on disk first 4 roots
      (  1)    Eigenvalue -152.7681863
(    1) 0.9997529 (    3)-0.0200730 (    8)-0.0070605 (    6)-0.0051093 (    2)-0.0023783 (    13)-0.0020552 (    10)-0.0017577
(    5)-0.0011004 (    15)-0.0007444 (    7)-0.0006106 (    16) 0.0001947 (    9)-0.0001779 (    4) 0.0000407 (    17) 0.0000297
(    12) 0.0000167 (    11) 0.0000136 (    18)-0.0000097 (    14) 0.0000096 (    21) 0.0000093 (    19) 0.0000062 (    20)-0.0000041
      (  2)    Eigenvalue -152.2847430
(    2) 0.9640987 (    7) 0.2522511 (    3) 0.0732649 (    8) 0.0265305 (    4) 0.0260930 (    11) 0.0071070 (    5)-0.0051855
(    10) 0.0048150 (    1) 0.0041229 (    6) 0.0025678 (    12)-0.0017794 (    9)-0.0012285 (    13) 0.0010707 (    16)-0.0005509
(    14)-0.0004307 (    20)-0.0003389 (    15) 0.0002153 (    17) 0.0002113 (    18)-0.0000772 (    19) 0.0000538 (    21)-0.0000280
Davidson Disk Diagonalization is being used:
MVecIn=    4 MSekIn= 4 NVConv= 2 NVConT= 0 MConv= 2 WhenSc=  2 IFact= 2 FinIt=F
NRoot=    4 Conv= 0.00D+00 ConvT= 0.00D+00 MaxItI=          0 Dim=       21 Max sub= 200
Iteration    1 internal iteration    1 dimension is    4
MCAXMl: NEAAct < 3 not allowed
Error termination via Lnk1e in /home/nuke/Documents/g16/l510.exe at Wed Dec  6 08:34:26 2017.
Job cpu time:    0 days  0 hours  0 minutes  0.3 seconds.
Elapsed time:    0 days  0 hours  0 minutes  0.3 seconds.
File lengths (MBytes):  RWF=    6 Int=    0 D2E=    0 Chk=    1 Scr=    1

作者
Author: sobereva 时间: 2017-12-6 23:51

nuke 发表于 2017-12-6 22:00
感谢指点，这个命令确实很管用。
不过9_06d.gjf 继续报错还是L510的。大概意思是
MCAXMl: NEAA ...

MCAXMl用百度不可能查到，要查也要用google
不过这属于开发者层面的信息，是变量名，搜一般也搜不到（除非CCL等地方有人讨论过），非要弄明白得读源代码，但更好的做法是直接给官方客服写信

作者
Author: nuke 时间: 2017-12-7 15:09

sobereva 发表于 2017-12-6 23:51
MCAXMl用百度不可能查到，要查也要用google
不过这属于开发者层面的信息，是变量名，搜一般也搜不到（ ...

是呀又学了一招多谢：）
我把所有CASSCF(4,4)改为CASSCF(6,6)这个问题就没有了。
但是又出现新的报错。。依旧是L510错误。
不知是不是机器配置的问题。
ulimit -S
chmod -R 770  g16 之类的命令都用了，也没有解决。

下面是运行9_06d.gjf之后输出结果的最后20行：
      NO OF BASIS FUNCTIONS =    175 NO TO BE DELETED = 0
CI Matrix Elements calculated here
      NO. OF CONFIGURATIONS IN REFERENCE SPACE = 1
                                 SECONDARY SPACE =  175
                                 TERTIARY SPACE =  175
      NO. OF ORBITALS = 6
      NO. OF ELECTRONS = 6
      NO. OF WEIGHTS = 19
      REFERENCE STATE CONFIGURATIONS ARE:    0
      NO. OF CORE ORBITALS = 0
      OPTION:  NON-DIAGONAL HOLE LINE INTERACTIONS INCLUDED
Len28=    94025 LenMCI=    51992.
  MCSCF will prepare Gradient Dif. and Deriv. Cps.
Enter MCSCF program.
NBasis= 48 NCore= 12 NVal= 6 NVirt= 30 IFrame=0 Acc=  1.00D-08.
End of file reading weights.
Error termination via Lnk1e in /home/nuke/Documents/g16/l510.exe at Thu Dec  7 01:50:59 2017.
Job cpu time:    0 days  0 hours  0 minutes  0.4 seconds.
Elapsed time:    0 days  0 hours  0 minutes  0.4 seconds.
File lengths (MBytes):  RWF=    6 Int=    0 D2E=    0 Chk=    1 Scr=    1

应该是9_06d.gjf中的倒数第二个--Link1--出错了

新的 9_06d.gjf如下
%Chk=cas_buta
#T UHF/STO-3G Test Guess=Mix Pop=NaturalOrbitals NoSymm

Conical Intersection: UHF/STO-3G

0 1
6          .000000    .000000    .702859
6          .000000    .000000 -.702859
6          -.631078 1.104055 1.471080
6          -.723500 -1.265744 -.892781
1          .458274 -.840810 1.204515
1          -.240104    .872255 -1.270306
1       -1.608859 1.448944 1.172244
1          -.027452 1.808388 2.026115
1          -.272487 -2.190375 -.587305
1       -1.783159 -1.274584 -1.06726

--Link1--
%Chk=cas_buta
#T CASSCF(6,6,UNO,NRoot=2,StateAverage,NoFullDiag)/STO-3G Test
Guess=Read Geom=Check NoSymm

Conical Intersection: State-averaged CAS with UNO option
                  to get starting orbitals

0 1

! Weight the two states equally
0.5    0.5

--Link1--
%Chk=cas_buta
#T CASSCF(6,6,StateAverage,NRoot=2)/4-31G Geom=Check
Guess=Read NoSymm Test

Conical Intersection: Converge with extended basis set

0 1

0.5    0.5

--Link1--
%Chk=cas_buta
#T CASSCF(6,6)/4-31G Opt=Conical Test
Geom=Check Guess=Read NoSymm IOp(1/8=5)

Conical Intersection optimization

0 1

--Link1--
%Chk=cas_buta
%NoSave
#P CASSCF(6,6,StateAverage,NRoot=2)/4-31G Test Geom=Check Guess=Read

Energies

0 1

0.5    0.5

作者
Author: 函数与激情 时间: 2017-12-7 16:23
高斯的CASSCF经常会出现L510报错，后来我就弃了用Molcas了

作者
Author: zjxitcc 时间: 2017-12-7 17:17

nuke 发表于 2017-12-7 15:09
是呀又学了一招多谢：）
我把所有CASSCF(4,4)改为CASSCF(6,6)这个问题就没有了。
但是又出现新的报错 ...

读取权重因子错误。权重因子要求每个因子符合F10.8，高斯手册里写了，检查是否不符合该格式。书上的例子，如果做不出来，要检查自己是否漏抄了一步，或者哪些地方没看清楚。

作者
Author: nuke 时间: 2017-12-8 00:25
提醒的极是，倒数第二个link1里少了权重因子。不知道是不是书上的笔误，太坑初学者了。
--Link1--
%Chk=cas_buta
#T CASSCF(6,6)/4-31G Opt=Conical Test
Geom=Check Guess=Read NoSymm IOp(1/8=5)

Conical Intersection optimization

0 1

0.5 0.5

作者
Author: Accelerator 时间: 2017-12-8 01:16
感觉不仅Molcas, ORCA的CASSCF也比Gaussian的易用

作者
Author: Abandon-fmt 时间: 2019-7-8 13:15
最近我也在学习CASSCF计算conical intersection，请问怎么从输出文件中找到conical intersection？

欢迎光临计算化学公社 (http://bbs.keinsci.com/)