676. Mixed basis set with gaussian when continuing from a chk-file

 I'm currently working with computations on paramagnetic species, so being able to read the geometry from the chk-file is important in order to avoid having to define the multiplicities of all the segments of the molecule. 

I also can't use def2-tzvp for the entire molecule since the .chk files during freq cals become too big for my current storage allocation (>500 gb). The work is exploratory at the moment, so compromises are OK.

I first run a calc with a small basis set:

%nprocshared=28
%Mem=24gb
%Chk=I1_hs.chk
#P uPBE1PBE/def2svp 5D 7F Punch=(MO) Pop=() guess=(fragment=8,only) nosymm

intermediate 1, pbe0/def-svp.

0 9 -2 1 0 1 0 1 -3 1 -1 1 0 1 4 4 2 6 
S(fragment=1)          -7.953777000    -0.003180000    -2.912460000
O(fragment=1)          -3.008379000    -0.432531000    -1.322496000
...

--Link1--
%nprocshared=28
%mem=24gb
%chk=I1_hs.chk
#P uPBE1PBE/chkbasis 5D 7F Opt=() SCRF=(PCM,Solvent=water)  Punch=(MO) Pop=() guess=read geom=allcheck nosymm

I then want to optimise the structure obtained at def2-svp using a mixed basis set where I use def2-tzvp for all the cobalt centres, and def2-svp for the organic bits.

%nprocshared=28
%Mem=24gb
%oldChk=I1_hs.chk
%chk=I1_hs_tzvp.chk
#P uPBE1PBE/genecp 5D 7F Opt=() SCRF=(PCM,Solvent=water)  Punch=(MO) Pop=() guess=read geom=allcheck nosymm

C H N O 0
def2svp
****

Co 0
def2tzvp
****

Co 0
def2

The basis set definitions are in blue, and the ecp bit is in red.