|
|
Impurity solver For OCA and NCA one should choose OCA. To use NCA method, executable should be set to nca, and solver should be set to OCA. |
|
|
Correlated set of orbitals Accepts Python dictionary of the form: {atom1:cor1,(atom2,atom3):cor2,...}. Here atom2 and atom3 are equivalent. cor1 is either the angular momentum l, or a dictionary of the form {l:(1,1,2,2,3,...)} where the sequence (1,1,2,2,3...) explains how to build a diagonal matrix out of 1..n columns of input self-energy |
|
|
Recompute the chemical potential at each step If the system is Mott insulating, it is advisable to set this variable to False, otherwise small error in the density can lead to closure of the gap. |
|
|
Double counting scheme. default: local limit of type U(nf-1/2), fixn: fixing nf in the equation U(nf0-1/2) to nf0, defined in the impurity problem below, fixEimp: fixing the impurity levels |
|
|
Number of charge self-consistent steps For charge non-self-consistent run use finish=1 |
|
|
In which base is the greens function computed The two methods should give equivalent result. DMFT base is faster and more tested |
|
|
At the very beginning, LDA run is converged and only then DMFT is started It is advisable to converge LDA first (in separate place) and only then use this code. |
|
|
Number of charge self-consistent steps If finish is smaller than max_iterations, the loop will take only finish steps |
|
|
Use tetrahedron method to determin the chemical potential It is advisable not to trast this method yet. |
|
|
Number of iterations in the inner DMFT loop Number of iterations at fixed charge density |
|
|
Run LDA at each step of the outside loop Sometimes we want to freeze ksum directory with the current content and we do not want the Hk and Ok to be recomputed. |
|
|
Run impurity solver at each iteration in the inside loop We can just converge the density keeping self-energy fixed. |
|
|
Is double counting computed from impurity density matrix or charge density matrix It is advisable to use impurity. The charge density matrix can lead to larger errors. |
|
|
The eigenvalues which are outside this interval are ignored Sometimes this can lead to divergence of Delta. In this case, make it large. |