We investigate the initial states both in the one dimensional
coordinate and momentum representation.
Let x be the position variable.
Its operator
is the multiplication with x , i.e.
Let p be the momentum variable. In atomic units p = k and this allows us to use the k wave number instead of the p momentum throughout this paper. The operator of the wave number k is
Moreover, x and k are conjugated variables, i.e.
The wave number space wave function is the Fourier transform of the coordinate space wave function:
![\begin{eqnarray}
\varphi(k) =
{\cal F} [ \psi(x) ] (k) &=&
\frac{1}{ \sqrt{2 \...
... \sqrt{2 \pi} }
\int_{-\infty}^{\infty} \varphi(k) e^{i k x} dk .\end{eqnarray}](img18.gif)
The form of the position and wave number operator in wave number space is the following:

For a free particle the energy and wave number operators commute,
, hence the eigenfunctions
of the
Hamilton operator which are the
solutions of the free space stationary Schrödinger equation
are eigenfunctions of
as well.
Hence k is a ''good quantum number'', the solutions of the
free space stationary Schrödinger equation can be uniquely
labelled with k values.
For the sake of comparison it is good to choose the
parameters of all our initial wave packets so that
their initial position indeterminacy values are all equal with a
common
.(This is not possible for the two
completely localised cases
VE
and VF
because in these cases one
can not choose
freely,
its value is for x localisation and
for k localisation.)
Some of our initial wave functions have compact support (bounded support).
Definition. A function has compact support if it is zero outside some finite interval.
If a coordinate wave function has compact support this means that it is localised to the given interval, i.e. there is no chance to find the quantum particle outside this interval. This kind of localisation can be achieved in principle by enclosing the particle into an infinite potential box. A finite potential box is not enough because the quantum particle always has a non-zero probability density in the forbidden region, i.e. there is a non-zero chance to find it at any far distance. In practice the localisation to a finite x interval can be achieved with a high enough finite potential box. The core electrons are e.g. well localised in a neutral atom or the valence electrons are well localised in a piece of metal.
If a wave function has compact support in k it means that its
spectrum is limited.
As we will see in
Section(III)
the dynamics in the x and k
spaces are different because the
Hamilton operator is not
symmetric in x and k in free space.
If a function does not have compact support than it is said to have infinite support. If a wave function has infinite support in x this means that the quantum particle has a non-zero probability of being found at any x value.
There is an important theorem for wave functions with compact support:
Theorem 1. If a wave function has compact support in the x coordinate (k wave number) space than it has infinite support in the conjugated k (x ) space.
Proof:
is the Fourier transform of
.The Fourier transform of a compact supported function
has always infinite support.
The wave number space wave function
corresponding to a
compact supported
can however decrease fast if
is chosen adequately.
This can be formulated as the following
theorem [4]:
Theorem 2.
If a function f(x) has compact support in x and it
is infinitely smooth then its Fourier
transform g(k) and all of its derivatives
decrease faster than any polynomial of 1 / |k| .
A function is infinitely smooth if it has continuous derivatives
of all orders on all points, i.e. it is infinitely
differentiable at all points.
At first sight it is not obvious that there exist infinitely
differentiable compact supported functions at all
because the simple compact supported functions (e.g. square or
triangle) are not infinitely differentiable at the edges of the support
interval.
But there is a whole class of such functions: these are the so
called ''good functions'' or ''test functions'' of distribution
theory [4].
These functions are infinitely differentiable also at the edges
of the support interval because they join smoothly the
function.
An example of a ''good function'' is the one we use at
VD.