Environmental Engineering Reference
In-Depth Information
a
0
and intensity
s
gn
; i.e.,
a
=
a
0
+
ξ
gn
. Thus the stochastic dynamics can be expressed
as (
4.26
) with
f
(
A
)
=
a
0
A
(
A
c
−
A
)
−
kA
,
g
(
A
)
=
A
(
A
c
−
A
)
.
(4.29)
This model is well suited to investigate the possible emergence of noise-induced
transitions as an effect of random environmental fluctuations in the growth rate
of the dynamics. In what follows we normalize
A
with respect to
A
c
and use the
state variable
A
=
A
/
A
c
. Moreover, to simplify the notation we drop the prime
superscript.
The steady-state pdf of
A
can be calculated with Eq. (
2.83
) from Chapter 2:
1
s
gn
k
1
s
gn
(
a
0
1
s
gn
(
a
0
Ce
−
−
k
)
−
1
A
)
−
−
k
)
−
1
p
(
A
)
=
A
(1
−
,
(4.30)
1
−
A
where
C
is the normalization constant. It can be shown that when
k
=
0 the normal-
ization constant
C
is not finite and
p
(
A
)
0),
as the noise intensity (i.e.,
s
gn
) increases, the dynamics undergo a noise-induced tran-
sition from a system with only one stable state (for the deterministic system or for
low values of
s
gn
) to bistable dynamics. Figure
4.21
(a) shows the modes of
A
calcu-
lated with Eq. (
3.48
): As
s
gn
exceeds a critical value
s
c
=
=
δ
(
A
−
1). In all the other cases (
k
=
k
), a noise-induced
transition occurs, in that multiple equilibria emerge and the probability distribution
of
A
becomes bimodal [Fig.
4.21
(b)]. Thus the multiplicative noise is able to in-
duce a new stable state
A
(
a
0
−
=
0 that did not exist in the underlying deterministic
system.
4.4.2 Stochastic genetic model
Logistic harvest model (
4.29
) can be generalized to account for possible inputs of
A
occurring at constant rate
β
.Knownasa
genetic model
(
Horsthemke and Lefever
,
1984
), this process finds a number of applications to population biology and dy-
namic ecology. For example, the genetic model could describe the dynamics of a
population affected by logistic growth, state-dependent emigration or harvest, and
state-independent immigration (
). If, for the sake of simplicity,
A
is normalized with
respect to
A
c
in a way that it ranges between 0 and 1, these dynamics are expressed
by
β
d
A
d
t
=
aA
(1
−
A
)
+
β
−
kA
.
(4.31)
Dividing both sides of (
4.31
)by
k
, normalizing the model's parameters
a
=
a
/
k
,
β
=
β/
k
, rescaling the time variable (
t
=
kt
), and dropping the prime superscripts,
we obtain
d
A
d
t
=
aA
(1
−
A
)
+
β
−
A
.
(4.32)
Search WWH ::
Custom Search