[molpro-user] Overflow in exponential generator

[Pfeifle, Mark Claudius Gerhard]

Hello,

I keep getting error messages of the following type in Molpro 2012.1 when doing MRCI jobs:

zz=   1194.77140253933       nterm=                 10001
 ? Error
 ? Overflow in exponential generator
 ? The problem occurs in muint2:updui

There seems to be no clear pattern as to when this error occurs. It might be at some rather random point in a geometry optimization or in a single-point energy step of a numerical frequency job.
Can anybody clarify what this means exactly and if there is a fix to this?

I also noticed the following lines in the *.log file, not sure if this is related to my problem:

Orbital overlap <old|new>:  < 1.1| 1.1> = 0.698248   < 2.1| 2.1> = 0.608532   < 3.1| 3.1> = 0.503882   < 4.1| 4.1> = 0.741411
                             < 5.1| 5.1> = 0.534934   < 6.1| 6.1> = 0.966994   < 7.1| 7.1> = 0.948818   < 8.1| 8.1> = 0.957069
                             < 9.1| 9.1> = 0.960224   <10.1|10.1> = 0.964671   <11.1|11.1> = 0.979189   <12.1|12.1> = 0.970835
                             <13.1|13.1> = 0.960944   <14.1|14.1> = 0.968968   <15.1|15.1> = 0.946732   <16.1|16.1> = 0.961805
                             <17.1|17.1> = 0.950729   <18.1|18.1> = 0.934844   <19.1|19.1> = 0.982848   <20.1|20.1> = 0.940276
                             <21.1|21.1> = 0.939871   <22.1|22.1> = 0.965210

As I understand, these numbers should be usually very close to 1. I suspect that somehow wrong orbitals are read during the geometry optimization.

This is my input file:

***,TS-opt
FILE,2,record2
memory,700,M
angstrom
geometry={
          C1
          O2, C1, RA
          O3, O2, RB, C1, AA
          H4, C1, RC, O2, AB, O3, DA
          H5, C1, RD, O2, AC, O3, DB
          C6, C1, RE, O2, AD, O3, DC
          O7, C6, RF, C1, AE, O2, DD
          H8, C6, RG, C1, AF, O2, DE
          H9, C6, RH, C1, AG, O2, DF
         }

readvar,coo_inp.act

basis=cc-pVDZ

{CASSCF
closed,18
occ,   22
wf,40,1,0
}

put,molden,orbitals-init.molden

MRCI

{OPTG,NUMERICAL,VARIABLE=ENERGD0,SAVEACT=coo_opt.act,SAVEXYZ=coo_opt.xyz,MAXIT=100
ROOT,2}

put,molden,orbitals-final.molden

---

Contents of coo_inp.act:

RA    =     1.44555409 ANGSTROM
 RB    =     1.32218102 ANGSTROM
 AA    =   109.60072498 DEGREE
 RC    =     1.10042061 ANGSTROM
 AB    =   107.79260503 DEGREE
 DA    =   -60.48504956 DEGREE
 RD    =     1.09924989 ANGSTROM
 AC    =   104.58318190 DEGREE
 DB    =  -177.54442173 DEGREE
 RE    =     1.53891741 ANGSTROM
 AD    =   112.23215675 DEGREE
 DC    =    64.16930776 DEGREE
 RF    =     1.38230926 ANGSTROM
 AE    =   114.24523883 DEGREE
 RG    =     1.10850568 ANGSTROM
 AF    =   109.40023469 DEGREE
 DE    =     1.13756982 DEGREE
 RH    =     1.10746338 ANGSTROM
 AG    =   111.96709591 DEGREE
 DF    =  -115.24016010 DEGREE
 DD    =   121.44849573 DEGREE

Note: record2 contains orbitals from a previous single-point CASSCF job.

In this case, the crash happens at task no. 76 of the initial numerical Hessian calculation, in iteration no. 2 of the CASSCF step.

Best regards,
Mark

----------------------------------------------------
Dr. Mark Pfeifle
Chemical Sciences & Engineering Division - Fundamental Interactions
Argonne National Laboratory
9700 S Cass Ave
Lemont, IL, 60439
mpfeifle at anl.gov<mailto:mpfeifle at anl.gov>
+1 (630) 252-3717



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20170721/7d9d8fa7/attachment-0001.html>