Script mode of DFTpy¶
DFTpy is a set of python modules. However, it can be executed in the old way by using the dftpy script which is generated at installation time. Here’s a quick guide to the script’s configuration dictionary, or config.
Warning
PP is a mandatory input (i.e., no default is avaliable for it).
Note
Defaults work well for most arguments.
When Options is empty, it can accept any value.
JOB¶
Control of the running job.
- task
The task to be performed.
Options : Optdensity, Calculation, Propagate, Casida, Diagonalize, Inversion
Default : Optdensity
- calctype
The property to be calculated.
Options : Energy, Potential, Both, Force, Stress
Default : Energy
PATH¶
Specify the path of needed files.
- pppath
The path of pseudopotential.
Options :
Default : ./
- cellpath
The path of input structure.
Options :
Default : ./
MATH¶
Some methods and techniques that make DFTpy really fast.
- linearii
Linear-scaling method to deal with Ion-Ion interactions (PME).
Options : True, False
Default : True
- linearie
Linear-scaling method to deal with Ion-Electron interactions (PME).
Options : True, False
Default : True
- twostep
A two-step method for performing density optimizations. ‘True’ is equivalent to multistep = 2.
Options : True, False
Default : False
- multistep
A multi-step method for performing density optimizations.
Options : 1,2,…
Default : 1
- reuse
Except in the first step, the initial density is given by the optimized density of the previous step.
Options : True, False
Default : True
PP¶
Control of the running job.
e.g. :
Al = Al_lda.oe01.recpot
CELL¶
Information about the input structure.
- cellfile
The file of input structure.
Options :
Default : POSCAR
- elename
The name of atom.
Options :
Default : Al
- zval
The charge of atomic species.
Options :
Default : None
- format
The format of structure file.
Options : pp, vasp, xsf, snpy, …
Default : None
Note
Only snpy format support parallel read and write
GRID¶
Control the grid.
- ecut
The kinetic energy cutoff.
Options :
Default : 600
Unit : eV
- spacing
The spacing (or gap) separating nearest real space grid points. If set this, ecut is disabled.
Options :
Default : None
Unit : Angstrom
- gfull
Determines oif the number of grid points in the reciprocal and real space grids are equal. If ‘False’ only use half grid, which will be faster.
Options : True, False
Default : False
Note
gfull=False’ implies that the the number of points of reciprocal space is only half of real space.
- nr
The number of grid points in the direction of the three lattice vectors.
Options :
Default : None
e.g. :
nr = 32 32 32
- maxprime
The max prime of guess best number of grid points for FFT
Options : 3, 5, 7, 11, 13, 17,…, 97
Default : 13
- scale
The minimum scale for guess the best number of grid points
Options :
Default : 0.99
- cplx
The type of real space value
Options : True, False
Default : False
DENSITY¶
Control the charge density.
- nspin
non/spin-polarized calculation
Options : 1, 2
Default : 1
- magmom
Total electronic magnetization.
Options :
Default : 0
- densityini
The initial density is given by homogeneous electron gas (HEG) or read from densityfile. If set Read, must given the densityfile.
Options : HEG, Read
Default : HEG
- densityfile
The charge density for initial density, only works when if densityini set Read.
Options :
Default : None
- densityoutput
The output file of final density. The default is not output the density.
Options :
Default : None
EXC¶
Control the exchange-correlation functional.
- xc
The kind of exchange-correlation functional. If not LDA, must have pylibxc installed. It has higher priority than libxc
Options : LDA, PBE
Default : None
- libxc
The type of exchange-correlation functionals with pylibxc. See available functionals.
Options :
Default : lda_x lda_c_pz
- x_str
See libxc’s available exchange functionals (deprecated).
Options :
Default : None
- c_str
See libxc’s available correlation functionals (deprecated).
Options :
Default : None
KEDF¶
Control the kinetic energy density functional (KEDF). DFTpy features most KEDFs, from GGAs to nonlocal to nonlocal with density dependent kernel.
- kedf
The type of KEDF. GGA functionals are available with keywords GGA and LIBXC_KEDF.
Options : TF, GGA, LIBXC_KEDF, vW, x_TF_y_vW, WT, SM, FP, MGP, MGPA, MGPG, LMGP, LMGPA, LMGPG
Default : WT
- x
The ratio of TF KEDF.
Options :
Default : 1
- y
The ratio of vW KEDF.
Options :
Default : 1
- alpha
The alpha parameter typical in nonlocal KEDF \(\rho^{\alpha}\).
Options :
Default : 0.8333333333333333
- beta
The beta parameter typical in nonlocal KEDF \(\rho^{\beta}\).
Options :
Default : 0.8333333333333333
- sigma
A parameter used to smooth with a Gaussian convolution FFTs of problematic functions (e.g., invfft of \({G^2\rho(G)}\) ).
Options :
Default : None
- nsp
The number of \({k_{f}}\) points for splining LWT like nonlocal KEDFs. There are three options to achieve the same goal, the priority is nsp -> delta -> ratio. Default is using ratio.
Options :
Default : None
- interp
The interpolation method for LWT KEDF’s kernel from the kernel table.
Options :
Default : hermite
- kerneltype
The kernel for LWT KEDF.
Options : linear, newton, hermite
Default : WT
- symmetrization
The symmetrization way for MGP KEDF. See MGP_paper.
Options : None, Arithmetic, Geometric
Default : None
- lumpfactor
The kinetic electron for LWT KEDF.
Options :
Default : None
- neta
The max number of discrete \(\eta\) for LWT KEDF.
Options :
Default : 50000
- etamax
The max value of eta for kernel in LWT KEDF.
Options :
Default : 50
- order
The order for the interpolation of the kernel in LWT KEDF. ‘0’ means using the value of nearest-neighbor point is used.
Options : 1, 2, 3, 4, 5
Default : 3
- ratio
The ratio of \({k_{f}}\) for spline in LWT KEDF. There are three options to do same thing, the priority is nsp -> delta -> ratio. Default is using ratio.
Options :
Default : 1.2
- maxpoints
The max number of integration points for the evaluation of the MGP kernel.
Options :
Default : 1000
- delta
The gap of spline
Options :
Default : None
- kdd
- The kernel density dependent for LWT KEDF:
1 : The origin LWT KEDF.
2 : Considers the \(\rho^{\beta}(r')\omega(\rho(r),r-r')\) term in the potential.
3 : Also considers the derivative of kernel which is neglected in LWT. See LMGP_paper.
Options : 1, 2, 3
Default : 3
- rho0
The ‘average’ density used for the definition of the Fermi momentum. Default is None, which means it calculated based on the total charge and system volume.
Options :
Default : None
- k_str
Functional type for GGA/LIBXC_KEDF
Options : LKT, DK, LLP, LLP91, OL1, OL, OL2, T92, THAK, B86A, B86, B86B, DK87, PW86, PW91O, PW91, PW91k, LG94, E00, P92, PBE2, PBE3, PBE4, P82, TW02, APBE, APBEK, REVAPBEK, REVAPBE, VJKS00, LC94, VT84F, SMP, TF, VW, X_TF_Y_VW, TFVW, STV, PBE2M, PG
Default : revAPBEK
Warning
GGA invokes DFTpy’s implementation. LIBXC_KEDF invokes libxc’s implementation (discouraged).
- params
Parameters for GGA KEDF functionals
Options :
Default : None
- kfmin
Lower limit of kf
Options :
Default : None
- kfmax
Upper limit of kf
Options :
Default : None
- rhomax
Maximum/cutoff density
Options :
Default : None
- ldw
local density weight
Options :
Default : None
- temperature
The temperature of TF KEDF.
Options :
Default : None
Unit : eV
- temperature0
The temperature of TF KEDF (analytical approximation).
Options :
Default : None
Unit : eV
KEDF2¶
Control the kinetic energy density functional (KEDF). DFTpy features most KEDFs, from GGAs to nonlocal to nonlocal with density dependent kernel.
copy
Options :
Default : KEDF
active
Options : True, False
Default : False
- x
The ratio of TF KEDF.
Options :
Default : 0
- y
The ratio of vW KEDF.
Options :
Default : 0
- kedf
The type of KEDF. GGA functionals are available with keywords GGA and LIBXC_KEDF.
Options : TF, GGA, LIBXC_KEDF, vW, x_TF_y_vW, WT, SM, FP, MGP, MGPA, MGPG, LMGP, LMGPA, LMGPG
Default : WT
- alpha
The alpha parameter typical in nonlocal KEDF \(\rho^{\alpha}\).
Options :
Default : 0.8333333333333333
- beta
The beta parameter typical in nonlocal KEDF \(\rho^{\beta}\).
Options :
Default : 0.8333333333333333
- sigma
A parameter used to smooth with a Gaussian convolution FFTs of problematic functions (e.g., invfft of \({G^2\rho(G)}\) ).
Options :
Default : None
- nsp
The number of \({k_{f}}\) points for splining LWT like nonlocal KEDFs. There are three options to achieve the same goal, the priority is nsp -> delta -> ratio. Default is using ratio.
Options :
Default : None
- interp
The interpolation method for LWT KEDF’s kernel from the kernel table.
Options :
Default : hermite
- kerneltype
The kernel for LWT KEDF.
Options : linear, newton, hermite
Default : WT
- symmetrization
The symmetrization way for MGP KEDF. See MGP_paper.
Options : None, Arithmetic, Geometric
Default : None
- lumpfactor
The kinetic electron for LWT KEDF.
Options :
Default : None
- neta
The max number of discrete \(\eta\) for LWT KEDF.
Options :
Default : 50000
- etamax
The max value of eta for kernel in LWT KEDF.
Options :
Default : 50
- order
The order for the interpolation of the kernel in LWT KEDF. ‘0’ means using the value of nearest-neighbor point is used.
Options : 1, 2, 3, 4, 5
Default : 3
- ratio
The ratio of \({k_{f}}\) for spline in LWT KEDF. There are three options to do same thing, the priority is nsp -> delta -> ratio. Default is using ratio.
Options :
Default : 1.2
- maxpoints
The max number of integration points for the evaluation of the MGP kernel.
Options :
Default : 1000
- delta
The gap of spline
Options :
Default : None
- kdd
- The kernel density dependent for LWT KEDF:
1 : The origin LWT KEDF.
2 : Considers the \(\rho^{\beta}(r')\omega(\rho(r),r-r')\) term in the potential.
3 : Also considers the derivative of kernel which is neglected in LWT. See LMGP_paper.
Options : 1, 2, 3
Default : 3
- rho0
The ‘average’ density used for the definition of the Fermi momentum. Default is None, which means it calculated based on the total charge and system volume.
Options :
Default : None
- k_str
Functional type for GGA/LIBXC_KEDF
Options : LKT, DK, LLP, LLP91, OL1, OL, OL2, T92, THAK, B86A, B86, B86B, DK87, PW86, PW91O, PW91, PW91k, LG94, E00, P92, PBE2, PBE3, PBE4, P82, TW02, APBE, APBEK, REVAPBEK, REVAPBE, VJKS00, LC94, VT84F, SMP, TF, VW, X_TF_Y_VW, TFVW, STV, PBE2M, PG
Default : revAPBEK
Warning
GGA invokes DFTpy’s implementation. LIBXC_KEDF invokes libxc’s implementation (discouraged).
- params
Parameters for GGA KEDF functionals
Options :
Default : None
- kfmin
Lower limit of kf
Options :
Default : None
- kfmax
Upper limit of kf
Options :
Default : None
- rhomax
Maximum/cutoff density
Options :
Default : None
- ldw
local density weight
Options :
Default : None
- temperature
The temperature of TF KEDF.
Options :
Default : None
Unit : eV
- temperature0
The temperature of TF KEDF (analytical approximation).
Options :
Default : None
Unit : eV
OUTPUT¶
Control the output.
- time
Output the time information of all parts.
Options : True, False
Default : True
- stress
Output the stress information of all terms.
Options : True, False
Default : True
- electrostatic_potential
Output of electrostatic potential
Options :
Default : None
OPT¶
Control the charge density optimization.
- method
The density optimization method.
Options : TN, LBFGS, CG-HS, CG-DY, CG-CD, CG-LS, CG-FR, CG-PR
Default : CG-HS
- algorithm
The direct minimization method : Energy (EMM) or Residual (RMM).
Options : EMM, RMM
Default : EMM
- vector
The scheme to deal with search direction.
Options : Orthogonalization, Scaling
Default : Orthogonalization
- c1
The wolfe parameters c1
Options :
Default : 0.0001
- c2
The wolfe parameters c2
Options :
Default : 0.2
- maxls
The max steps for line search.
Options :
Default : 10
- econv
The energy convergence for last three steps (a.u./atom).
Options :
Default : 1e-06
Unit : a.u./atom
- maxfun
The max steps for function calls. For TN density optimization method its the max steps for searching direction.
Options :
Default : 50
- maxiter
The max steps for optimization
Options :
Default : 100
- xtol
Relative tolerance for an acceptable step.
Options :
Default : 1e-12
- h0
The initial approximation for the inverse Hessian needed by LBFGS.
Options :
Default : 1
PROPAGATOR¶
Control of the propagator. DFTpy has an implementation of hydrodynamic TDDFT. This is essentially TDDFT with one orbital only, defined as \({\psi(r,t)=\sqrt{\rho(r,t)}e^{iS(r,t)}}\), and determined by the following time-dependent Schroedinger equation,
\({-\frac{1}{2} \nabla^2 \psi(r,t) + v_s(r,t) \psi(r,t) = i\frac{d}{dt}\psi(r,t)}\),
where \({v_s = v_{xc} + v_H + v_{T_s} - v_{vW} + v_{dyn}}\), See paper.
- propagator
The type of propagator.
Options : crank-nicholson, taylor
Default : crank-nicholson
- order
The order used for the Taylor expansion propagator.
Options :
Default : 20
- linearsolver
The linear solver used for the Crank-Nicolson propagator. The solvers with a name end with _scipy are from the SciPy package and should be used in serial calculations only.
Options : bicg, bicgstab, cg, bicg_scipy, bicgstab_scipy, cg_scipy, cgs_scipy, gmres_scipy, lgmres_scipy, minres_scipy, qmr_scipy
Default : cg
- tol
The relative tolerance for the linear solver used for the Crank-Nicolson propagator.
Options :
Default : 1e-10
- maxiter
The max amount of iteration steps for the linear solver used for the Crank-Nicolson propagator.
Options :
Default : 100
- atol
The absolute tolerance for the linear solver used for the Crank-Nicolson propagator.
Options :
Default : None
TD¶
Control the TDDFT parameters that lie outside the propagator class.
- outfile
The prefix of the output files.
Options :
Default : td_out
- timestep
The time step in atomic units.
Options :
Default : 0.001
- tmax
The total amount of time in atomic units.
Options :
Default : 1
- max_pc
The max amount of the predictor-corrector steps.
Options :
Default : 1
- tol_pc
The relative tolerance for the predictor-corrector.
Options :
Default : 1e-08
- atol_pc
The absolute tolerance for the predictor-corrector.
Options :
Default : 1e-10
- direc
The direction of the initial kick. 0, 1, 2 stands for x-, y-, z-direction, respectively.
Options : 0 or x, 1 or y, 2 or z
Default : 0
- strength
The strength of the initial kick in atomic units.
Options :
Default : 0.001
- max_runtime
Max amount of running time in seconds before the program saves the intermediate result and quitting.
Options :
Default : 0
- restart
Restart the propagation from a previously saved intermediate result.
Options : True, False
Default : False
- restart_input
Input file for restart data.
Options :
Default : restart_data.npy
- save_interval
The number of time step interval between two saves of states.
Options :
Default : 1000
correction
Options : True, False
Default : False
- vector_potential
If true, use vector potential for the initial kick.
Options : True, False
Default : False
- propagate_vector_potential
Only matters if vector_potential is true. If true, propagate the vector potential.
Options : True, False
Default : True
omega
Options :
Default : 0
z_split
Options :
Default : 0
CASIDA¶
Control of the CASIDA.
- numeig
Number of eigenstates used in constructing casida matrix.
Options :
Default : None
- diagonalize
If true, diagonalize the Hamiltonian before construct the Casida matrix. If false, read the eigenstates from a saved file.
Options : True, False
Default : True
- tda
Use Tamm-Dancoff approximation.
Options : True, False
Default : False
NONADIABATIC¶
- nonadiabatic
Name of nonadiabatic Pauli potential.
Options :
Default : None
- cutoff
If :math:’k_F’ is smaller than the cutoff, treat it equal to the cutoff.
Options :
Default : 0.01
rho_cutoff
Options :
Default : 0.001
k
Options :
Default : 2
NONADIABATIC2¶
copy
Options :
Default : NONADIABATIC
active
Options : True, False
Default : False
- nonadiabatic
Name of nonadiabatic Pauli potential.
Options :
Default : None
- cutoff
If :math:’k_F’ is smaller than the cutoff, treat it equal to the cutoff.
Options :
Default : 0.01
rho_cutoff
Options :
Default : 0.001
k
Options :
Default : 2
INVERSION¶
Control of the INVERSION.
- rho_in
Input file for the density.
Options :
Default : None
- v_out
Output file for the potential.
Options :
Default : None