Restarting MCSCF calculation with different basis

[ankit7540]

I am trying to rerun a RASSCF calculation with two basis sets for N2. The first calculation goes from HF-> MP2-> RASSCF with the basis cc-pVTZ (this is as suggested on my previous question (viewtopic.php?f=9&t=1074). I obtain the output along with the SIRIUS.RST and SIRIFC files.

Now I would like to restart the same with aug-cc-pVTZ basis.

I tried, passing SIRIUS.RST to the new input, as following,

Code: Select all

*ORBITAL INPUT
.MOSTART
NEWORB

and received the error.

Code: Select all

                   .--------------------------------------------.
                   | Starting in Wave Function Section (SIRIUS) |
                   `--------------------------------------------'


Input ERROR in connection with ".MOSTART" of type    9
     NSYM, NBAS from input       :  8        19   11   11    5   19   11   11    5
     NSYM, NBAS from MOSTART file:  8        13    7    7    3   13    7    7    3

I was hoping the program could use orbital overlap scheme (so called projecting orbitals in MCSCF from prev results) to guess the initial occupation for the new larger basis based on the previous results. Any ways to achieve this in DALTON.

The two output files are attached.

Thanks.

[taylor]

I don't think you can "restart" with a different basis. I think the word is intended to be interpreted literally, as in picking up a calculation at the point where it has stopped (or died) and continuing it from there, possibly with a minor change to the input parameters.

I can imagine that projecting the previous calculation MOs onto the new atomic basis would give you a better starting guess and save maybe a few MCSCF iterations, but I am sceptical that the saving would be that great in terms of the overall computer time.

Best regards
Pete

[ankit7540]

Thank you.