molpro计算分子势能面时提示的关于BASIS LINEARLY DEPENDENT OR WRONG S的问题

Annieyang

计算三原子全局势能面的过程中，分别计算固定SiH的距离和SiH-H键角，改变SiH-H距离。已经计算完了一组，SiH距离为1.5，角度为0~180(30°为间隔)，这组输出文件没有出现问题。但在我改变SiH的距离，设置为2.0，2.5，3.0，等等，都出现了不同的问题。譬如在SiH键长为2.0，SiH-H键角为180°这个计算中，就出现了以下问题，输出文件提示：
Eigenvalues of metric

         1 0.122E-09 0.347E-06 0.143E-05 0.468E-05 0.655E-05 0.146E-04 0.171E-04 0.184E-04
         2 0.655E-05 0.171E-04 0.111E-03 0.127E-03 0.163E-03 0.203E-02 0.259E-02 0.568E-02

ERROR: BASIS LINEARLY DEPENDENT OR WRONG S

The current threshold for the smallest tolerated eigenvalue of S is   1.0D-08
输入文件和已经算出来的那组数据唯一的区别就是更改了SiH的键长，其他数据都没有动。这个问题在以前从未遇到过，不知道怎么处理，希望大神解答一下

Annieyang

例如另一个SiH键长为2.5时，SiH-H角度为60度时，输出文件提示的错误是：
? Error
? Insufficient overlap
? The problem occurs in cihdia
这组数据的输入文件和已经算出结果的输入文件相比，也是只有SiH键长不一样而已，这是为啥呢？

Annieyang

输入文件
*** H-SiH
! FILE,2,/scratch/H-SiH.wfu,NEW;
GPRINT,BASIS;
GPRINT,ORBITALS;
GPRINT,ORBEN;
GPRINT,CIVECTOR;

I=0;

NPOINTS=46;

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! loop over R in Jacobi coordinates
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

DO I=1,NPOINTS;

BRMIN=2.0D+00 BOHRS;

BRMAX=11.00D+00 BOHRS;

STEP=(BRMAX-BRMIN)/(NPOINTS-1.00D+00);

SYMMETRY,X;

GEOMTYP=ZMAT;

BOHR;

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! SiH bond distance
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
rjac=2.5 BOHR;  
! center-of-mass H-SiH distance
BIGR=BRMIN+STEP*(I-1.0D+00) BOHR;
! Jacobi angle
ANGJAC=60.0 DEGS;
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! atomic masses of diatomic
AMUP1=28.085;
AMUP2=1.0079;
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
FAC1=AMUP1/(AMUP1+AMUP2);
FAC2=AMUP2/(AMUP1+AMUP2);
rr1=FAC2*rjac;
rr2=FAC1*rjac;
A=180.0;
SUPANG=(180.0-ANGJAC);
D=180.0;
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! X is the center-of-mass
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

GEOMETRY={
X;
H1,X,BIGR;
Si,X,rr1,H1,ANGJAC;
H2,X,rr2,H1,SUPANG,Si,D;
}

BASIS,AVTZ;

IF (I.EQ.1) THEN

{HF,MAXIT=500;

WF,16,1,0;ORBITAL,IGNORE_ERROR;
};

END IF

{MULTI;
WF,16,1,0;state,3
MAXITER,40;ORBITAL,IGNORE_ERROR;
};

ECASD1(I)=ENERGY(1);
ECASD2(I)=ENERGY(2);
ECASD3(I)=ENERGY(3);

{CI;
WF,16,1,0;state,3
MAXITER,1919,18000;ORBITAL,IGNORE_ERROR;
};

EMRCIQD1(I)=ENERGD(1);
EMRCIQD2(I)=ENERGD(2);
EMRCIQD3(I)=ENERGD(3);

BASIS,AVQZ;

{MULTI;

WF,16,1,0;state,3;ORBITAL,IGNORE_ERROR;
MAXITER,40;
};

ECAST1(I)=ENERGY(1);
ECAST2(I)=ENERGY(2);
ECAST3(I)=ENERGY(3);

{CI;
WF,16,1,0;state,3
MAXITER,1919,18000;ORBITAL,IGNORE_ERROR;
};

EMRCIQT1(I)=ENERGD(1);
EMRCIQT2(I)=ENERGD(2);
EMRCIQT3(I)=ENERGD(3);


!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

r(I)=rjac;
BR(I)=BIGR;
ANG(I)=ANGJAC;


R1(I)=SQRT(BIGR**2+rr1**2-2.0*BIGR*rr1*COS(ANGJAC));
R2(I)=rjac;
R3(I)=SQRT(BIGR**2+rr2**2-2.0*BIGR*rr2*COS(SUPANG));
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!



{TABLE,r,BR,ANG,ECASD1,EMRCIQD1};
{TABLE,r,BR,ANG,ECAST1,EMRCIQT1};
{TABLE,R1,R2,R3,ECASD1,EMRCIQD1};
{TABLE,R1,R2,R3,ECAST1,EMRCIQT1};


{TABLE,R1,R2,R3,ECASD1,ECASD2,ECASD3,EMRCIQD1,EMRCIQD2,EMRCIQD3,ECAST1,ECAST2,ECAST3,EMRCIQT1,EMRCIQT2,EMRCIQT3};
{TABLE,r,BR,ANG,ECASD1,ECASD2,ECASD3,EMRCIQD1,EMRCIQD2,EMRCIQD3,ECAST1,ECAST2,ECAST3,EMRCIQT1,EMRCIQT2,EMRCIQT3};

{TABLE,R1,R2,R3,ECASD1,EMRCIQD1,ECAST1,EMRCIQT1};

{TABLE,r,BR,ANG,ECASD1,EMRCIQD1,ECAST1,EMRCIQT1};

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

END DO
---

wzkchem5

Annieyang 发表于 2020-11-30 21:32
例如另一个SiH键长为2.5时，SiH-H角度为60度时，输出文件提示的错误是：
? Error
? Insufficient overl ...

这个应该是说明MRCISD波函数和CASSCF波函数的差别比较大，也就是说参考态不好。
建议检查一下态平均的态数是否合理

qinjiu

不建议使用ignore_error指令
大概率会遇到初猜不好的问题

zjxitcc

（1）同楼上，不建议使用IGNORE_ERROR；
（2）“ERROR: BASIS LINEARLY DEPENDENT OR WRONG S”这问题我知道，但是目前我也不知道咋解决。此即表示基函数存在线性相关问题，有冗余。可能是由于弥散函数过多所致，但也与几何结构有关。如果用的是aug-基组，可以改用没有aug的基组+F12技术。但是如何像高斯一样去除冗余，我查了molpro手册目前并没有找到。高斯的阈值只有1.0D-06，Molpro是1.0D-08，但是我看了手册没看出怎么改这个阈值。