Summary of options¶

The behavior of qcm can be controlled by many global parameters that can be set via a call to qcm.set_global_parameter(). There are four kinds of global parameter, depending on the type of variable involved:

Boolean parameter. They are all false by default. Calling qcm.set_global_parameter(name) with the name of the parameter sets that parameter to True.
integer-valued parameter. Calling qcm.set_global_parameter(name, val) with the name of the parameter sets that parameter’s value to val.
real-valued parameter. Same calling prototype as above.
char-valued parameter. Same calling prototype as above.

The following tables list the different global parameters, their default value and a short description.

Boolean options¶

name	default	description
CSR_sym_store	false	stores CSR matrices fully for openMP application
check_lanczos_residual	false	checks the Lanczos residual at the end of the eigenvector computation
continued_fraction	false	Uses the continued fraction solver for the Green function instead of the band Lanczos method
dual_basis	false	uses the dual basis for wavevector computations
no_degenerate_BL	false	forbids band lanczos to proceed when the eigenstates have degenerate energies
nosym	false	does not take cluster symmetries into account
one_body_solution	false	Only solves the one-body part of the problem, for the Green function
periodic	false	considers the cluster(s) as periodic
periodized_averages	false	computes lattice averages using the periodized Green function
print_Hamiltonian	false	Prints the Hamiltonian on the screen, if small enough
print_all	false	prints dependent parameters as well
strip_anomalous_self	false	sets to zero the anomalous part of the self-energy
verb_ED	false	prints ED information and progress
verb_Hilbert	false	prints progress information on bases and operators in the Hilbert space
verb_integrals	false	prints information and progress about integrals
verb_warning	false	prints warnings
zero_dim	false	sets the spatial dimension to zero, on any model

Integer-valued options¶

name	default	description
Davidson_states	1	Number of states requested in the Davidson-Liu algorithm
GK_min_regions	8	minimum number of regions in the Gauss-Kronrod method
PRIMME_algorithm	1	PRIMME algorithm to solve ground state (if qcm_wed was compiled with PRIMME): 1 (PRIMME_DYNAMIC - default), 2 (PRIMME_DEFAULT_MIN_TIME), 3 (PRIMME_DEFAULT_MIN_MATVECS), 4 (PRIMME_Arnoldi), 5 (PRIMME_GD), 6 (PRIMME_GD_plusK), 7 (PRIMME_GD_Olsen_plusK), 8 (PRIMME_JD_Olsen_plusK), 9 (PRIMME_RQI), 10 (PRIMME_JDQR), 11 (PRIMME_JDQMR), 12 (PRIMME_JDQMR_ETol), 13 (PRIMME_STEEPEST_DESCENT), 14 (PRIMME_LOBPCG_OrthoBasis), 15 (PRIMME_LOBPCG_OrthoBasis_Window). See PRIMME documentation for more information
PRIMME_preconditionning	0	Choose of preconditionner to solve ground state (if qcm_wed was compiled with PRIMME): 0 (No preconditionning), 1 (Jacobi preconditionner)
cuba2D_mineval	1024	minimum number of integrand evaluations in CUBA (2D)
cuba3D_mineval	16000	minimum number of integrand evaluations in CUBA (3D)
dim_max_print	64	Maximum dimension for printing vectors and matrices
kgrid_side	32	number of wavevectors on the side in a fixed wavevector grid
max_dim_full	256	Maximum dimension for using full diagonalization
max_iter_BL	600	Maximum number of iterations in the band Lanczos procedure
max_iter_CF	400	Maximum number of iterations in the continuous fraction Lanczos procedure
max_iter_QN	60	maximum number of iterations in the quasi-Newton method
max_iter_lanczos	600	Maximum number of iterations in the Lanczos procedure
print_precision	8	precision of printed output
seed	0	seed of the random number generator

Real-valued options¶

name	default	description
Qmatrix_vtol	1e-10	minimum value of a Qmatrix contribution
Qmatrix_wtol	1e-05	maximum difference in frequencies in Q-matrix
accur_Davidson	1e-05	maximum norm of residuals in the Davidson-Liu algorithm
accur_OP	0.0001	accuracy of lattice averages
accur_Q_matrix	1e-05	tolerance in the normalization of the Q matrix
accur_SEF	5e-08	Accuracy of the Potthoff functional
accur_band_lanczos	1e-12	energy difference tolerance for stopping the BL process
accur_continued_fraction	0.01	value of beta below which the simple Lanczod process stops
accur_deflation	1e-07	norm below which a vector is deflated in the band Lanczos method
accur_lanczos	1e-12	tolerance of the Ritz residual estimate in the Lanczos method
band_lanczos_minimum_gap	1e-05	gap between the lowest two states in BL below which the method fails
cutoff_scale	1e+12	high-frequency cutoff in integrals
eta	0.0001	value of the imaginary part of the frequency in Chern number/Berry phase computations
large_scale	20	high-frequency region for imaginary frequency axis integrals
minimum_weight	0.01	minimum weight in the density matrix
small_scale	0.5	low-frequency region for imaginary frequency axis integrals
temperature	0	Temperature of the system.

Char-valued options¶

name	default	description
Ground_state_method	L	Desired method to compute the ground state: L (Lanczos method - default), M (modified Lanczos method), D (Davidson method), P (use external PRIMME eigensolver - need qcm_wed to be compiled with PRIMME)
Hamiltonian_format	S	Desired Hamiltonian format: S (CSR matrix), O (individual operators), F (factorized), N (none = on the fly), E (Eigen CSR matrix), P (MPI Distributed matrix with PETSc)
periodization	G	periodization scheme: G, S, M, C or N (None)