The computational rate of B3LYP in DMol3

Hi everyone,

Recently I found that there is B3LYP functional in DMol3 after version6.0.

So I chose it as the applied functional for calculation, with DNP basis set and 4.4 basis file.

For comparing the computational rate between different functionals, I calculate the energy.

For example, the model I used in this test is saturated (TiO2)10 cluster, which had been optimized first.

Functional SCF cycles Comput. Time (min)

BLYP     29      2.35

B3LYP    124     152.10

The results showed that the computational time used for B3LYP is much longer than that of BLYP functional.

However, in our background knowledge, the difference of calculation rate between BLYP and B3LYP should not be quite large.

I am wondering if i had wrong setup? or if there is other reason to let the calculation quite slow?

The following is my input using B3LYP functional.

All the setup is the same as using BLYP functional.

Thanks for your reply!

Sincerely,

Cherry

-----Input for Calculation with B3LYP functional-----

# Task parameters

Calculate                     energy

Symmetry                      off

Max_memory                    12000

File_usage                    smart

# Electronic parameters

Spin_polarization             restricted

Charge                        0

Basis                         dnp

basis_version                 basfile_v4.4

Pseudopotential               none

Functional                    b3lyp umesh=xcoarse minu=-3

Aux_density                   octupole

Integration_grid              fine

Occupation                    thermal 0.0020

Cutoff_Global                 5.2000 angstrom

Scf_density_convergence       1.0000e-008

Scf_charge_mixing             2.0000e-001

Scf_iterations                250

Scf_diis                      6 pulay

# Print options

Print                         eigval_last_it