This is a Julia version of the DoNOF (Donostia Natural Orbital Functionals) software written by Prof. Mario Piris, with the intention of take advantage of the wonderful capabilities of Julia language.
Open a julia prompt, enter to the Pkg REPL by pressing ], and simply add DoNOF:
add DoNOF- Put the following directly on the Julia prompt to run a calculation:
using DoNOF
mol = """
0 1
O 0.0000 0.000 0.121
H 0.0000 0.751 -0.485
H 0.0000 -0.751 -0.485
"""
bset,p = DoNOF.molecule(mol,"cc-pvdz")
p.ipnof = 8
DoNOF.energy(bset,p)- You can also save the input in a file (e.g. example.jl) and run it directly on the command line:
julia example.jl > example.outTip
Available functionals
- p.ipnof = 9 (GNOFm)
- p.ipnof = 8 (GNOF)
- p.ipnof = 7 (PNOF7)
- p.ipnof = 7 + p.ista=1 (PNOF7s)
- p.ipnof = 5 (PNOF5)
- p.ipnof = 4 (PNOF4)
Note
Available optimizers
- For Orbital Optimization:
- p.orb_method = "ADAM"
- p.orb_method = "ADABelief"
- p.orb_method = "YOGI"
- p.orb_method = "Demon"
- For Occupation Optimization:
- p.occ_method = "Softmax"
- p.occ_method = "Trigonometric"
You can give title to the jobs, and the orbitals and occupations will be saved on a jld2 file with that name:
- p.title = "molecule_name"
Orbitals are saved in a FCHK file that you can open in your prefered software (e.g. Avogadro2, IQmol)
-
You should have Julia installed
-
Remember to configure the number of threads for parallelism, for example:
export JULIA_NUM_THREADS=12
- To use GPUs, just add CUDA, cuTENSOR and KernelAbstractions:
using CUDA, cuTENSOR, KernelAbstractions, DoNOF
mol = """
0 1
O 0.0000 0.000 0.121
H 0.0000 0.751 -0.485
H 0.0000 -0.751 -0.485
"""
bset,p = DoNOF.molecule(mol,"cc-pvdz")
p.ipnof = 8
DoNOF.energy(bset,p)