September 27, 2004 16:46 WSPC/Book Trim Size for 9.75in x 6.5in GlobalAttractors/chapter 5
176 Global Attractors of Non-autonomous Dissipative Dynamical Systems
Remark 5.2 Unlike Theorems 5.1 and 5.2 in Theorem 5.3 we do not demand the
continuity of the function V , but instead of that we impose certain condition on the
velocity of decreasing of the function V along trajectories of the system (X, T
1
, π).
In additional, when T
1
= R
+
the completeness of the space is not necessary (in the
applications there are examples, when Y is a fortiori incomplete).
Theorem 5.4 Let h(X, T
1
, π), (Y, T
2
, σ), hi be a non-autonomous dynamical
system, T
1
= R
+
, (X, T
1
, π) be asymptotically compact and there is a continuous
function V : E → R satisfying the following conditions:
1. for all c ∈ R the set {x ∈ E| V (x) ≤ c} is bounded;
2. along trajectories of the system (X, T
1
, π) the function V is non-increasing, i.e.
V (xt) ≤ V (x) for all x ∈ E and t ≥ 0;
3. there exists r > 0 such that V (xt) < V (x), if xτ ∈ E
r
for all τ ∈ [0, t] and
t > 0.
Then the non-autonomous dynamical system h(X, T
1
, π), (Y, T
2
, σ), hi is locally k-
dissipative.
Proof. First let us notice that as V is continuous, then in a point p ∈ X there is
C
p
∈ R and δ
p
> 0 such that V (x) ≤ C
p
for any x ∈ B(p, δ
p
). According to the
condition 2. V (xt) ≤ V (x) ≤ C
p
for all x ∈ B(p, δ
p
) and t ≥ 0. According to the
condition 1. the set M
p
:= Σ
+
(B(p, δ
p
)) = ∪{π
t
B(p, δ
p
)|t ≥ 0} is bounded and as
(X, T
1
, π) is asymptotically compact, then there is a nonempty compact K
p
such
that the equality
lim
t→+∞
β(π
t
M
p
, K
p
) = 0 (5.6)
takes place. Define K := {x ∈ X | |x| ≤ r}, where r > 0 is the number from the
condition 3. of the theorem. As Y is compact and as the fiber bundle (X, h, Y ) is
locally trivial, then the set K is bounded. Let us notice that ω
p
∩ K 6= ∅ for any
point p ∈ X. In fact, if this were not so, we should have a point p
0
∈ X such that
ω
p
0
∩ K = ∅. According to the equality (5.6) Σ
+
p
0
:= ∪{π
t
p
0
| t ≥ 0} is relatively
compact and, hence, ω
p
0
6= ∅ is compact and invariant. The function µ : R
+
→ R
defined by the equality µ(t) := V (p
0
t) is continuous, bounded and monotone non-
increasing and, hence, there is lim
t→+∞
V (p
0
t) = c
0
. As V is continuous, we have
ω
p
0
⊆ V
−1
(c
0
). Let x ∈ ω
p
0
, then V (xt) = V (x) = c
0
for all t > 0. The last
contradicts condition 3. of the theorem. This contradiction shows that ω
p
∩K 6= ∅
for all p ∈ X. Let us show that as a matter of fact ω
p
⊆ K for all p ∈ X. Indeed, if
it is not so, then there will be t
0
> 0 such that xτ ∈ ω
p
0
\K for all τ ∈ [0, t
0
] and,
hence, V (xt
0
) < V (x). On the other hand, reasoning as we do above, we will have
V (x) = c
0
for some c
0
∈ R and all x ∈ ω
p
0
. This contradicts the last inequality.
Thus Ω ⊆ K, where Ω :=
∪{ω
p
|p ∈ X}, and, as Ω is invariant and as (X, T
1
, π)