awdwadwad
- MNDO
- AM1
- PM3
- Length: Å
- Energy: eV
- Temperature: Kelvin
- Time: femtosecond
- Force: eV/Å
- Velocity: Å/femtosecond
# Perform Single Point calculation with batch of systems, see Test1.py
# For furthur usage, see examples in tests/test*
import torch
from seqm.seqm_functions.constants import Constants
from seqm.Molecule import Molecule
from seqm.ElectronicStructure import Electronic_Structure
torch.set_default_dtype(torch.float64)
if torch.cuda.is_available():
device = torch.device('cuda')
else:
device = torch.device('cpu')
# Specify SEQM parameters and create molecule object:
species = torch.as_tensor([[8,6,1,1],
[8,6,1,1],
[8,8,6,0]], # zero-padding for batching
dtype=torch.int64, device=device)
coordinates = torch.tensor([
[
[0.00, 0.00, 0.00],
[1.22, 0.00, 0.00],
[1.82, 0.94, 0.00],
[1.82, -0.94, 0.00]
],
[
[0.00, 0.00, 0.00],
[1.22, 0.00, 0.00],
[1.82, 0.94, 0.00],
[1.82, -0.94, 0.00]
],
[
[0.00, 0.00, 0.00],
[1.23, 0.00, 0.00],
[1.82, 0.94, 0.00],
[0.0,0.0,0.0] # zero-padding for batching
]
], device=device)
const = Constants().to(device)
elements = [0]+sorted(set(species.reshape(-1).tolist()))
seqm_parameters = {
'method' : 'AM1', # AM1, MNDO, PM#
'scf_eps' : 1.0e-6, # unit eV, change of electric energy, as nuclear energy doesnt' change during SCF
'scf_converger' : [2,0.0], # converger used for scf loop
# [0, 0.1], [0, alpha] constant mixing, P = alpha*P + (1.0-alpha)*Pnew
# [1], adaptive mixing
# [2], adaptive mixing, then pulay
'sp2' : [False, 1.0e-5], # whether to use sp2 algorithm in scf loop,
#[True, eps] or [False], eps for SP2 conve criteria
'elements' : elements, #[0,1,6,8],
'learned' : [], # learned parameters name list, e.g ['U_ss']
#'parameter_file_dir' : '../seqm/params/', # file directory for other required parameters
'pair_outer_cutoff' : 1.0e10, # consistent with the unit on coordinates
'eig' : True
}
molecules = Molecule(const, seqm_parameters, coordinates, species).to(device)
### Create electronic structure driver:
esdriver = Electronic_Structure(seqm_parameters).to(device)
### Run esdriver on molecules:
esdriver(molecules)
print(' Total Energy (eV):\n', molecules.Etot)
print('\n Electronic Energy (eV): ', molecules.Eelec)
print('\n Nuclear Energy (eV):\n', molecules.Enuc)
print('\n Heat of Formation (ev):\n', molecules.Hf)
print('\n HOMO-LUMO gap (eV):\n', molecules.e_gap)
print('\n Orbital energies (eV):\n', molecules.e_mo)
print('\n Charges:\n', molecules.q)
print('\n Dipole:\n', molecules.d)
print('\n Forces (eV/A):\n', molecules.force)- ./Test1_SinglePointProp.ipynb : single point calculation
- ./Test2_GeomOpt.ipynb : geometry optimization
- ./Test3_BOMD.ipynb : BOMD (NVE, NVT, Langevin)
- ./Test4_XL-BOMD.ipynb : XL-BOMD
- ./Test5_KSA-XL-BOMD.ipynb : KSA-XL-BOMD
- ./Test6_NormModes_Methanol.ipynb : normal modes analysis of methanol via semi-numerical method
- ./Test7_low_rank_KSA_scf.ipynb : low-rank KSA for density matrix SCF optimization
- ./Fullerenes : fullerenes test set
- ./GramicidinS/KSA-XL-BOMD_GS_dp.ipynb : Langevin MD of doubly protonated Gramicidin S ( [GS + H2]2+ )
- filename.id.xyz : MD trajectory
- Number of atoms
- Additional info line : Step number, Temperature (K), Kinetic energy (eV), Potential energy (eV), HOMO-LUMO gap (eV) KSA error, real time stamp (s)
- Block of XYZ coordinates, velocities, forces: Element X Y Z Vx Vy Vz Fx Fy Fz
- filename.id.Info.txt : extended info log
- Step number
- Orbital energies
- Dipole (x, y, z)
- Electronic entropy contribution (KSA-XL-BOMD only)
- Fermi occupancies (KSA-XL-BOMD only)
- Rank-m Krylov subspace approximation error (KSA-XL-BOMD only)
seqm : seqm module
├── basics.py : collections of classes for perform basic operations
├── MolecularDynamics.py : geometry optimization and NVE and Langevin Molecular Dyanmics
├── XLBOMD.py : XL-BOMD
├── params : MNDO/AM1/PM3 parameters
└── seqm_functions
├── cal_par.py : compute dipole/qutrupole charge separation and additive terms rho1 and rho2
├── canon_dm_prt.py : canonical density matrix perturbation algorithm for first-order response (for KSA-XL-BOMD)
├── constants.py : store some constant parameters
├── data_loader.py : load and prepare dataset from numpy array, not updated
├── diag.py : diagonalization functions, where pseudo_diag is not used
├── diat_overlap.py : get overlap integrals
├── energy.py : compute various energy terms
├── fermi_q.py : Fermi operator expansion (for KSA-XL-BOMD)
├── fock.py : construct Fockian matrix
├── G_XL_LR.py : generates two-electron part of Fockian (for KSA-XL-BOMD)
├── hcore.py : construct Hcore
├── pack.py : functions to deal with the padding in batch of matrix
├── parameters.py : load parameters from structured csv files as in ./params
├── read_xyz.py : read structures .xyz files
├── save_xyz.py : save structures to .xyz files
├── scf_loop.py : perform SCF procedure
├── SP2.py : single particle density matrix expansion algorithm SP2
├── spherical_pot_force.py : apply spherical potential (plane bottom with parabolic walls)
├── two_elec_two_center_int_local_frame.py : compute two electron two center integrals in local frame for each pair of atoms
└── two_elec_two_center_int.py : rotate and get two electron two center integrals in global frame
basics.py
class Parser : prepare data in the form for other parts, similar to ./seqm/seqm_functions/data_loader
class Pack_Parameters : combine parameters provided (like from ML) with other required ones loaded from ./params/
class Hamiltonian : assemble functions in seqm/seqm_functions, perform SCF and construct Fockian
class Energy : get energies based on Hamiltonian (total energy, heat of formation, nuclear energies, etc)
class Force : use torch.autograd.grad to get gradient of total energy on coordinates to get force
ElectronicStructure.py
class Electronic_Structure : calls SCF/XL-BOMD drivers and updates molecular properties
Molecule.py
class Molecule : molecule objects for storing electronic/structural properties
MolecularDynamics.py
class Geometry_Optimization_SD : geometry optimization with steepest descend
class KSA_XL_BOMD : NVE/NVT with Langevin thermostat for KSA-XL-BOMD
class Molecular_Dynamics_Basic : NVE MD
class Molecular_Dynamics_Langevin : NVT with Langevin thermostat
class XL_BOMD : NVE/NVT with Langevin thermostat for XL-BOMD
XLBOMD.py
class EnergyXL : get energies based on XL-BOMD
class ForceXL : get force with XL-BOMD
- Atoms in molecules are sorted in descending order based on atomic number
- 0 padding is used for atoms in molecules
NH3 ==> [7,1,1,1], H2O ==> [8,1,1,0]
- indexing for atoms works with or without padding atom
[[7,1,1,1],[[7,1,1,0],[7,1,1,1]] ==> index [0,1,2,3,4,5,6,8,9,10,11]
[[7,1,1,1],[[7,1,1,0],[7,1,1,1]] ==> index [0,1,2,3,4,5,6,7,8,9,10]
- Benchmark convergence criteria eps for SCF and SP2, eps_sp2 will affect number of iterations for SCF
- in general set converger=[2], combining adaptive mixing and pulay seems takes fewest iterations
- when on GPU, use SP2, on CPU don't use SP2, but benchmark to check
- For molecules with degeneracy, use convergers=0 or 2, and turn on
seqm.seqm_functions.diag.CHECK_DEGENERACY = True