I am having particular trouble getting an optimized structure after deleting orbitals (using $DEL) using B3LYP/6-311G** calculations (thru G09 RevD) on monocyclic structures. The calculation seems to always fail with this error:

- Code: Select all
`WARNING| EIGENVALUE 48 TOO LARGE. REPLACED BY 25.000000`

WARNING| EIGENVALUE 47 TOO LARGE. REPLACED BY 25.000000

WARNING| EIGENVALUE 46 TOO LARGE. REPLACED BY 25.000000

WARNING| EIGENVALUE 45 TOO LARGE. REPLACED BY 25.000000

WARNING| EIGENVALUE 44 TOO LARGE. REPLACED BY 25.000000

HESSIAN DOES NOT HAVE THE DESIRED LOCAL STRUCTURE

TAKING SIMPLE RFO STEP

SEARCHING FOR LAMDA THAT MINIMIZES ALONG ALL MODES

*****************************************

*** UNABLE TO DETERMINE LAMDA IN FmD114 **

*****************************************

and for reference, eigenvaleus 44-48 are listed in my output as:

- Code: Select all

44 45 46 47 48

28.49815 67.56663 72.29492 78.67816 79.09973

I have tried an even smaller basis set to see if that could cause this, but the same error persists.

What is going wrong here? How do I fix this error??