The TURBOMOLE Interface
Introduction
This interface is designed to access the basic functionality of the
TURBOMOLE program package. It is capable of calculating energies
and gradients at the HF, DFT, MP2 and CC2 levels
of theory. In order to use the interface, the environment must be set
up as required for a TURBOMOLE standard run (PATH and TURBODIR). The interface passes
commands to the TURBOMOLE define module to create the necessary input
files for a TURBOMOLE run and then starts the corresponding modules
(dscf, ridft, etc.). Alternatively the control- and related files can
be provided manually and passed to the interface (see
read_control and restart).
Serial and MPI-parallel versions are supported. For parallel
runs, the environment variables PARA_ARCH (=MPI) and PARNODES (number
of CPUs to be used) have to be set before running ChemShell.
Command Line Arguments
| Argument |
Argument type |
Mandatory |
Default |
To specify |
| basis= |
keyword |
no |
undefined |
Required basis set (turbomole internal library) |
| basisspec= |
Tcl List
|
no |
sto-3g all |
Required basis set (ChemShell library) |
| charge= |
integer |
no |
0 |
total charge |
| accuracy= |
keyword |
no |
medium |
Accuracy of the grid (DFT): low, medium, high,
veryhigh 1 |
| conv= |
integer |
no |
undefined |
Overrides SCF convergence implied by "accuracy". Convergence
is set to 10-conv au. |
| list_option= |
Output keyword
|
no |
medium |
how much output to generate |
| jobname= |
string |
no |
turbomole |
name to use as root for file names |
| scratchdir= |
string |
no |
/scratch |
where to store integral files |
| maxscratch= |
integer |
no |
undefined |
max. size of integral files in MB 2 |
| hamiltonian= |
keyword |
no |
hf |
choice of QM Hamiltonian: hf, mp2, cc2, lda, blyp, bp86, pbe, tpss,
b3lyp, b3lyp_G, bhlyp, pbe0, tpssh. |
| read_control= |
Boolean
|
no |
no |
whether to use pre-existing control- and
related files 3 |
| restart= |
Boolean
|
no |
no |
If restart=yes, then no define will be run. An existing
control file will be used. |
| symmetry= |
Boolean
|
no |
no |
whether to use symmetry 4 |
| use_ri= |
Boolean
|
no |
See 5 |
whether to use the RI approximation. |
| ri_memory= |
integer |
no |
200 |
RI-memory
in MB 5 |
| disp3= |
Boolean
|
no |
no |
Use Grimme's D3 dispersion correction. |
Options for excited state calculations
TDHF/TDDFT and CC2 excited state energies and gradients can be requested using the
following options:
| Argument |
Argument type |
Mandatory |
Default |
To specify |
| excited |
Boolean
|
no |
no |
Perform an excited state calculation. |
| estate |
integer |
no |
1 |
The excited state of interest (1 is the 1st excited state, etc.) |
| eroots |
integer |
no |
= estate |
Number of excitations to calculate. Only
needs to be set if you want to calculate further states above estate
(e.g. to avoid root flipping). |
| etriplet |
Boolean
|
no |
no |
Calculate triplet excitations instead of singlet excitations.
(Ignored for unrestricted calculations) |
| tda |
Boolean
|
no |
no |
Use the Tamm-Dancoff approximation/CI singles (TDHF/TDDFT only). |
| rpa_memory |
integer |
no |
200 |
Memory for response calculations in MB (TDHF/TDDFT only). |
| Argument |
Argument type |
Mandatory |
Default |
To specify |
| iter |
integer |
no |
not spec. |
iter=0 causes a normal QM calculation
and (R)ESP charges fitted to the obtained density afterwards. iter=1 skips the
QM calculation and calculates the electrostatic interaction with bq-charges
from the (R)ESP charges. |
| esp_npoint |
integer |
no |
-1 |
Number of points to fit the potential. Negative
values mean twice the number of QM atoms, see
get_esp_points |
| esp_type |
keyword |
no |
shell |
mmatoms leads to a fit of the potential at the
nearest MM atoms, shell leads to a fit at a shell around the QM atoms, see
get_esp_points |
| esp_method |
keyword |
no |
esp |
resp causes RESP charges to be calculated, see
fit_esp_charges |
Examples
Single point MP2 energy of a charged species:
energy coords=start.c energy=e theory=turbomole : { hamiltonian=mp2 ri_memory=1000 basis=TZVPP charge=-3 }
print_matrix matrix=e
Parallel geometry optimisation, using an already existing control file:
hdlcopt coords=start.c result=final.c theory=turbomole : { hamiltonian=bp86
ri_memory=1000 nproc=4 read_control=yes } 3
-
The different levels of accuracy correspond to the TURBOMOLE DFT integration
grid size and SCF convergence. Low: Grid m3, SCF convergence 10-6 au;
medium: m3, 10-7 au; high: m4, 10-8 au;
veryhigh: m5, 10-9 au.
-
The size of integral files in case of the non RI-methods is determined
automatically via a dscf statistics run in the default case.
-
Using a predefined control file via read_control=yes requires only the
input of the Hamiltonian and optional memory, CPU and disk requirements
(see example). Other keywords will be ignored.
-
Symmetry=yes will use the define desy command to determine the symmetry.
It is not possible to guarantee that symmetry will not change during an
optimisation so the current default is needed for robust general
operation.
-
The RI-option will be automatically enabled for functionals without HF-Exc
(RI-Coulomb approximation) and for MP2 and CC2.
|
