The Stokes operator, named after George Gabriel Stokes, is an unbounded linear operator used in the theory of partial differential equations, specifically in the fields of fluid dynamics and electromagnetics.
If we define
P
σ
as the Leray projection onto divergence free vector fields, then the Stokes Operator
A
is defined by
A
:=
−
P
σ
△
,
where
△
≡
∇
2
is the Laplacian. Since
A
is unbounded, we must also give its domain of definition, which is defined as
D
(
A
)
=
H
2
∩
V
, where
V
=
{
u
→
∈
(
H
0
1
(
Ω
)
)
n
|
div
u
→
=
0
}
. Here,
Ω
is a bounded open set in
R
n
(usually n = 2 or 3),
H
2
(
Ω
)
and
H
0
1
(
Ω
)
are the standard Sobolev spaces, and the divergence of
u
→
is taken in the distribution sense.
For a given domain
Ω
which is open, bounded, and has
C
2
boundary, the Stokes operator
A
is a self-adjoint positive-definite operator with respect to the
L
2
inner product. It has an orthonormal basis of eigenfunctions
{
w
k
}
k
=
1
∞
corresponding to eigenvalues
{
λ
k
}
k
=
1
∞
which satisfy
0
<
λ
1
<
λ
2
≤
λ
3
⋯
≤
λ
k
≤
⋯
and
λ
k
→
∞
as
k
→
∞
. Note that the smallest eigenvalue is unique and non-zero. These properties allow one to define powers of the Stokes operator. Let
α
>
0
be a real number. We define
A
α
by its action on
u
→
∈
D
(
A
)
:
A
α
u
→
=
∑
k
=
1
∞
λ
k
α
u
k
w
k
→
where
u
k
:=
(
u
→
,
w
k
→
)
and
(
⋅
,
⋅
)
is the
L
2
(
Ω
)
inner product.
The inverse
A
−
1
of the Stokes operator is a bounded, compact, self-adjoint operator in the space
H
:=
{
u
→
∈
(
L
2
(
Ω
)
)
n
|
div
u
→
=
0
and
γ
(
u
→
)
=
0
}
, where
γ
is the trace operator. Furthermore,
A
−
1
:
H
→
V
is injective.
