Schur test explained

In mathematical analysis, the Schur test, named after German mathematician Issai Schur, is a bound on the

L2\toL2

operator norm of an integral operator in terms of its Schwartz kernel (see Schwartz kernel theorem).

Here is one version.[1] Let

X,Y

be two measurable spaces (such as

Rn

). Let

T

be an integral operator with the non-negative Schwartz kernel

K(x,y)

,

x\inX

,

y\inY

:

Tf(x)=\intYK(x,y)f(y)dy.

If there exist real functions

p(x)>0

and

q(y)>0

and numbers

\alpha,\beta>0

such that

(1)    \intYK(x,y)q(y)dy\le\alphap(x)

x

and

(2)    \intXp(x)K(x,y)dx\le\betaq(y)

for almost all

y

, then

T

extends to a continuous operator

T:L2\toL2

with the operator norm

\Vert

T\Vert
L2\toL2

\le\sqrt{\alpha\beta}.

Such functions

p(x)

,

q(y)

are called the Schur test functions.

In the original version,

T

is a matrix and

\alpha=\beta=1

.[2]

Common usage and Young's inequality

A common usage of the Schur test is to take

p(x)=q(y)=1.

Then we get:

\Vert

2
T\Vert
L2\toL2

\le \supx\in\intY|K(x,y)|dy \supy\in\intX|K(x,y)|dx.

This inequality is valid no matter whether the Schwartz kernel

K(x,y)

is non-negative or not.

A similar statement about

Lp\toLq

operator norms is known as Young's inequality for integral operators:[3]

if

\supx(\int

rdy)
Y|K(x,y)|

1/r+\supy(\int

rdx)
X|K(x,y)|

1/r\leC,

where

r

satisfies
1
r=1-(1
p-1
q)
, for some

1\lep\leq\leinfty

, then the operator

Tf(x)=\intYK(x,y)f(y)dy

extends to a continuous operator

T:Lp(Y)\toLq(X)

, with

\Vert

T\Vert
Lp\toLq

\leC.

Proof

Using the Cauchy–Schwarz inequality and inequality (1), we get:

\begin{align}

2=\left|\int
|Tf(x)|
Y

K(x,y)f(y)dy\right|2 &\le\left(\intYK(x,y)q(y)dy\right)\left(\intY

K(x,y)f(y)2
q(y)

dy\right)\\ &\le\alphap(x)\intY

K(x,y)f(y)2
q(y)

dy. \end{align}

Integrating the above relation in

x

, using Fubini's Theorem, and applying inequality (2), we get:

\VertT

2
f\Vert
L2

\le\alpha\intY\left(\intXp(x)K(x,y)dx\right)

f(y)2
q(y)

dy \le\alpha\beta\intYf(y)2dy=\alpha\beta\Vert

2.
f\Vert
L2

It follows that

\VertT

f\Vert
L2

\le\sqrt{\alpha\beta}\Vert

f\Vert
L2
for any

f\inL2(Y)

.

See also

References

  1. [Paul Richard Halmos]
  2. [I. Schur]
  3. Theorem 0.3.1 in: C. D. Sogge, Fourier integral operators in classical analysis, Cambridge University Press, 1993.