Environmental Engineering Reference
In-Depth Information
−1
x
Figure 6.8. Spatiotemporal patterns of
φ
obtained as solutions of the 1D version of
(
6.11
) calculated for
A
1
=
A
2
=
1. Figure taken from
Buceta et al.
(
2002a
).
respect to the rate of convergence to equilibrium in each of the two states, we can
replace the noise term in (
6.13
) with its average value. We obtain the (deterministic)
equation
∂φ
(
r
,
t
)
2
k
0
)
2
=
f
1
[
φ
(
r
,
t
)]
P
+
f
2
[
φ
(
r
,
t
)](1
−
P
)
−
D
(
∇
+
φ
(
r
,
t
)
.
(6.14)
∂
t
Thus, in this case, we can investigate pattern formation by referring to the aver-
age dynamics, using the methods presented for the deterministic case [Eqs. (
5.18
)
and (
5.19
)]. Although for
P
=
0and
P
=
1 the potential function defined as
f
(
V
(
)] has only one stable state, for inter-
mediate values of
P
bistability may emerge. When
A
1
=
φ
)
=−
φ
)
=−
[
Pf
1
(
φ
)
+
(1
−
P
)
f
2
(
φ
A
2
=
1and
P
=
0
.
5, the
local dynamics are expressed by the function
f
3
. In this system, stable
patterns emerge, as shown in Fig.
6.8
for the case of a 1D domain.
We need to stress again that, even though in this example patterns emerge as a result
of the random switching between two deterministic dynamics, these patterns are not
necessarily noise induced. In fact,
Buceta and Lindenberg
(
2002a
) showed that the
same patterns would emerge even when the random forcing
=
φ
−
φ
ξ
dn
in (
6.13
) is replaced
with a deterministic periodic function driving the switching between the two states,
1 and 2. Thus, in this system, dichotomous noise is used as a random mechanism to
drive the repeated alternation between two deterministic states. However, deterministic
Search WWH ::
Custom Search