| Back to the main page of this manual | Input reference of CP2K version 7.1 |
CP2K_INPUT /
FORCE_EVAL /
DFT /
XC /
WF_CORRELATION /
IM_TIME&IM_TIME {Logical} |
|
| Activates cubic-scaling RPA, GW and Laplace-SOS-MP2 calculations. | |
| This keyword cannot be repeated and it expects precisely one logical. | |
The lone keyword behaves as a switch to
.TRUE. |
|
Default value:
.FALSE. |
CUTOFF_W {Real} |
|
| Cutoff for screened Coulomb interaction for GW kpoints. | |
| This keyword cannot be repeated and it expects precisely one real. | |
Default value:
5.00000000E-001 |
DO_DBCSR_T {Logical} |
|
| Uses DBCSR tensors for cubic scaling RPA/L-SOS-MP2/GW | |
| This keyword cannot be repeated and it expects precisely one logical. | |
The lone keyword behaves as a switch to
.TRUE. |
|
Default value:
.TRUE. |
DO_KPOINTS {Logical} |
|
| Besides in DFT, this keyword has to be switched on if one wants to do kpoints in. cubic RPA. | |
| This keyword cannot be repeated and it expects precisely one logical. | |
The lone keyword behaves as a switch to
.TRUE. |
|
Default value:
.FALSE. |
EPS_FILTER_IM_TIME {Real} |
|
| Determines threshold for DBCSR based multiply P(it)=(Mocc(it))^T*Mvirt(it). Threshold for Mocc/virt(it)=(abP)*D_ad^occ/virt(it) is determined by EPS_FILTER in WFC_GPW. Default should be kept. | |
| This keyword cannot be repeated and it expects precisely one real. | |
Default value:
1.00000000E-008 |
EXP_KPOINTS {Real} |
|
| For kpoints in low-scaling GW, a Monkhorst-Pack mesh is used. Because the screened Coulomb interaction W(k) diverges at the Gamma point with W(k) ~ k^alpha, we adapt the weights of the Monkhorst-Pack mesh to compute int_BZ k^alpha dk (BZ=Brllouin zone) correctly with the Monkhorst-Pack mesh. You can enter here the exponent alpha. For solids, the exponent is -2 (known from plane waves), for 2d periodic systems -1 and for 1d systems W(k) ~ log(1-cos(a*k)) where a is the length of the unit cell in periodic direction. If you enter 1.0, one of these three functions are picked according to the periodicity. If you enter a value bigger than 2.0, the ordinary Monkhorst-Pack mesh with identical weights is chosen. | |
| This keyword cannot be repeated and it expects precisely one real. | |
Default value:
1.00000000E+000 |
GROUP_SIZE_3C {Integer} |
|
| This keyword is not needed if DO_DBCSR_T enabled. Group size used for cutting the RI index P in (alpha beta P) and for the multiplication M^occ/virt(it)=(alpha beta P)*D^occ/virt. Has to be increased for larger systems, at least linearly with system size, at most quadratically in system size. | |
| This keyword cannot be repeated and it expects precisely one integer. | |
Default value:
1 |
GROUP_SIZE_INTERNAL {Logical} |
|
| If DO_DBCSR_T and this keyword set to .TRUE., group sizes are automatic and keywords GROUP_SIZE_3C & GROUP_SIZE_P are ignored. This should be set to .FALSE. for testing purposes only. | |
| This keyword cannot be repeated and it expects precisely one logical. | |
Default value:
.TRUE. |
GROUP_SIZE_P {Integer} |
|
| This keyword is not needed if DO_DBCSR_T enabled. Group size used for the calculation of P(it)=(Mocc(it))^T*Mvirt(it). Default is to use a single process. A larger group size (for example the node size), could be recommended for larger systems, since the P matrix is replicated in every group. | |
| This keyword cannot be repeated and it expects precisely one integer. | |
Default value:
1 |
KPOINTS {Integer} {Integer} {Integer} |
|
| For periodic calculations, using kpoints for the density response and the Coulomb operator are strongly recommended. For 2d periodic systems (e.g. xy periodicity, please specify KPOINTS N_x 0 N_z. | |
| This keyword cannot be repeated and it expects precisely 3 integers. | |
Default values:
0 0 0 |
MAO {Logical} |
|
| Decide whether to use modified atomic orbitals to reduce the prefactor of the quadratic-scaling step in im. time RPA. The number of MAOsper atom have to be specified in the KIND section for all elements. | |
| This keyword cannot be repeated and it expects precisely one logical. | |
The lone keyword behaves as a switch to
.TRUE. |
|
Default value:
.FALSE. |
MAX_BLOCK_SIZE_SQRT {Integer} |
|
| Square root of maximum block size for splitting blocks of matrix representation of large tensors along the combined dimension. Atomic blocks are split in order to allow for better load balance. | |
| This keyword cannot be repeated and it expects precisely one integer. | |
Default value:
10 |
MEMORY_CUT {Integer} |
|
| Cuts the memory of the M_munu_P^occ/virt matrix by the factor given. The calculation is slowed down when using a high value of MEMORY_CUT. A high value for MEMORY_CUT is recommended for large systems in order not to run out of memory. | |
| This keyword cannot be repeated and it expects precisely one integer. | |
Default value:
1 |
MEMORY_INFO {Logical} |
|
| Decide whether to print memory info on the sparse matrices. | |
| This keyword cannot be repeated and it expects precisely one logical. | |
The lone keyword behaves as a switch to
.TRUE. |
|
Default value:
.FALSE. |
POINTS_PER_MAGNITUDE {Integer} |
|
| Number of evaluation points (x) of the function 2x/(x^2+omega^2) per order of magnitude of the range Rc. Normally, this parameter needs not to be touched. | |
| This keyword cannot be repeated and it expects precisely one integer. | |
Default value:
200 |
|
| Alias names for this keyword: PPM |
STABILIZE_EXP {Real} |
|
| It is observed that EXP(-alpha) gives NaN if alpha is larger than 100-300. Using the STABILIZE_EXP keyword, all EXP(-alpha) are not computed but set to zero for alpha > STABILIZE_EXP. | |
| This keyword cannot be repeated and it expects precisely one real. | |
Default value:
7.00000000E+001 |
| Back to the main page of this manual or the CP2K home page | (Last update: 14.12.2021) |