Recall that
given a
-dimensional vector subspace of a vector space
of dimension
one can always find an
-dimensional vector subspace
of
(see Exercise 3.3.20.9) satisfying
The subspace
is called the complementary subspace of
in
We now define an important class of linear transformations
on an inner product space, called orthogonal projections.
Remark 5.3.2
The map
is well defined due to the following reasons:
-
implies that for every
we can find
and
such that
-
implies that the expression
is unique for every
The next proposition states that the map defined above is a linear
transformation from
to
We omit the proof, as it follows directly
from the above remarks.
Remark 5.3.5
- The projection map
depends on the complementary subspace
- Observe that for a fixed subspace
there are infinitely many choices
for the complementary subspace
- It will be shown later that if
is an inner product space with inner product,
then the subspace
is unique if we put an additional condition that
We now prove some basic properties about projection maps.
THEOREM 5.3.6
Let
and
be complementary subspaces of a vector space
Let
be a projection operator of
onto
along
Then
- the null space of
- the range space of
-
The condition
is equivalent to
The next result uses the Gram-Schmidt orthogonalisation process to get
the complementary subspace in such a way that the vectors in different subspaces
are orthogonal.
DEFINITION 5.3.8 (Orthogonal Subspace of a Set)
Let
be an inner product space. Let
be a non-empty subset of
.
We define
EXAMPLE 5.3.9
Let
.
-
. Then
.
-
, Then
.
- Let
be any subset of
containing a non-zero real number. Then
.
THEOREM 5.3.10
Let
be a subset of a finite dimensional inner product space
with inner product
Then
-
is a subspace of
- Let
be equal to a subspace
. Then the subspaces
and
are complementary. Moreover, if
and
then
and
Proof.
We leave the prove of the first part for the reader. The prove of the
second part is as follows:
Let
and
Let
be a basis of
By Gram-Schmidt orthogonalisation process, we get an
orthonormal basis, say,
of
Then,
for any
So,
Also, for any
by definition of
So,
That is,
height6pt width 6pt depth 0pt
DEFINITION 5.3.12 (Self-Adjoint Transformation/Operator)
Let
be an inner product space with inner product
A linear transformation
is called
a self-adjoint operator if
for every
Remark 5.3.14
- By Proposition 5.3.3, the map
defined above
is a linear transformation.
-
- Let
with
and
for some
and
Then we know that
whenever
Therefore,
for every
Thus, the orthogonal projection operator is a self-adjoint operator.
- Let
and
Then
for all
Therefore, using Remarks 5.3.14.2
and 5.3.14.3, we get
for every
- In particular,
as
. Thus,
, for every
.
Hence, for any
and
we have
Therefore,
and the equality holds
if and only if
Since
we see that
That is,
is the vector nearest to
This can also be stated
as: the vector
solves the following minimisation problem:
Subsections
A K Lal
2007-09-12