The structure of the algorithm can be seen by derivation of the second iteration term of the Neumann series for (4.24) using the forward
formulation to obtain the ensemble Monte Carlo algorithm. This algorithm does not give the distribution function value at a given point, but instead the
relative number of carriers in a region with volume
around point in the quasi-momentum
space4.7.
The relative number of carriers in determined by the small perturbation
is defined as:
|
(4.37) |
where the domain indicator
has bee introduced as a function which evaluates to one if
and zero otherwise.
In order to use the phase space volume conservation law described in Section 2.1.3, an additional integral over the real
space is added and (4.38) is rewritten as
|
(4.38) |
where is a crystal volume. From (4.37) the second iteration term is obtained:
|
|
|
(4.39) |
|
|
|
|
Considering as
which is the real space trajectory corresponding to the quasi-momentum trajectory (4.31) it follows from the
phase space volume conservation4.8 that
.
Denoting
with
(4.31) can be rewritten in a forward initialization:
|
(4.40) |
In the same manner combining
with
gives
. Denoting
with
(4.31) can be rewritten in a forward initialization:
|
(4.41) |
Finally
. Denoting
as
one obtains for the corresponding
forward initialization:
|
(4.42) |
Integrating out cancels the crystal volume and the final expression is:
|
|
|
(4.43) |
|
|
|
|
To express the time integrals in a forward way the following identity is used:
|
(4.44) |
which is schematically shown in Fig. 4.3.
Figure 4.3:
The same integration area can be covered either vertically a) or horizontally b).
|
Now the second iteration term has the form:
where
stands for an after-scattering wave vector and
denotes an initial wave vector. The quantity
represents a normalized after-scattering distribution. As can be seen from
(4.13) and (4.14) it is normalized to unity. It follows from (4.47), that during Monte Carlo simulation a particle
trajectory is constructed in terms of new quantities
and
.
S. Smirnov: