Here is what can get this system to converge

Here is what can get this system to converge:

1) do first a Hartree-Fock calculation, then use the HF molecular orbitals to start the DFT B3LYP calculation

2) Your symmetry is very close to Cs. A bit of manipulation gets you there.

Here is the modified input following the two suggestions that I have just mentioned.

Cheers, Edo

=======================
start 26dnt

title "2,6-Dinitrotoluene by B3LYP-DFT"

echo

scratch_dir /tmp

memory total 1192 mb noverify

geometry units angtroms print xyz autosym

C 0.34506906 -0.00778468 -1.2005

C 1.73054057 0.01370338 -1.2005

C -0.34746133 -0.01897714 0.00000000

C 2.42281255 0.02427087 0.00000000

C 0.34506906 -0.00778468 1.2005

C 1.73054057 0.01370338 1.2005

C -1.85610551 -0.04194054 0.00000000

N -0.38971017 -0.01774911 -2.4739

N -0.38971017 -0.01774911 2.4739

O 0.21997195 -0.02576942 3.53030374

O -1.60712515 0.05406579 2.4739

O 0.21997195 -0.02576942 -3.53030374

O -1.60712515 0.05406579 -2.4739

H 2.28599047 0.02248751 -2.16262574

H 3.53369701 0.04120014 0.00000000

H 2.28599047 0.02248751 2.16262574

H -2.24192611 0.99997900 0.00000000

H -2.21807403 -0.57077608 0.90765694

H -2.21807403 -0.57077608 -0.90765694

end

driver

linopt 0

maxiter 200

end

basis spherical

H library d-aug-cc-pvdz

O library d-aug-cc-pvdz

N library d-aug-cc-pvdz

C library d-aug-cc-pvdz

end

dft

xc hfexch

iterations 200

vectors input atomic output hf.mos

end

task dft ignore

dft

xc b3lyp

vectors input hf.mos output b3lyp.mos

end

task dft

ecce_print 26dnt.ecce.out

task dft optimize

task dft energy

charge 1

dft

xc b3lyp

mult 2

odft

end

task dft energy