command,t=,dt=,ns=,ng=,grsize=,print=; PULSE,options
The total propagation time (in au) is set by t; dt sets the
timestep and ns
the number of steps, where two of the three have to be provided.
ng sets the number of grid points in one
dimension (default = 0) and grsize the grid size in bohr
(default = 10 bohr).
Setting ng switches on the calculation of quantum
currents (see below). The option print
determines the
level of output (0=normal output, 1=object linear in matrices,
2=matrices as well,
debug).
The subcommand PULSE determines the envelope
used, and takes several options depending on the envelope
selected. Possibilities are:
All runs produce a file ($input).dat (or molpro.dat) which contains time-dependent properties, like the components of the field, components of the dipole moment, total energy, orbital occupation numbers, orbital energies and total number of electrons. The Molpro output file specifies the order of the data in the file. For very long runs the size of the file is restricted by printing only every twentieth set of data.
In case of an unrestricted run three files are produced: ($input).dat, ($input)a.dat, ($input)b.dat. The first file contains the
generic data about the field, total energy, dipole moments and the
expectation value of the total spin operator. The other two files
contain the orbital energies and occupation numbers for the
and
electrons respectively.
TDKS and TDUKS use the options provided by the KS and UKS comands. At the moment TDKS/TDUKS suffers from numerical instabilities when using strong fields. Divergence of the energy is observed, possibly due to the use of quadrature for the evaluation of the potential.
Quantum currents will be calculated when choosing . A cubic
grid will be computed of size grsize with a total of
gridpoints. The imaginary part of the density will be summed at
every timestep and after the dynamics the total current will be
evaluated at every gridpoint. The user can extract the required
data from this array by printing out parts of it, or by integrating
over point, but this requires actual coding, as this has not been
implemented sufficiently neat. It is also straightforward to
evaluate the currents at every timestep or at selected timesteps.
This is not done automatically, because it slows down the dynamics
considerably.