A normed space is separable if it has a countabledense subset. i.e. it contains a countable subset whose closure is the entire space.
e.g.
The finite dimensional complex spaces are separable Hilbert spaces, with the inner product being
An infinite-dimensional analogue of the above example is the space with the inner product
The space of all complex-valued measurable function on such that , modulo the equivalence relation of being equal almost everywhere, denoted as , is a prime example of a separable Hilbert space. The inner product is defined as We shall first check that is a vector space, and is a well-defined inner product.
For any , we have so for all , hence is a vector space. Moreover, So the inner product is well-defined. Now we verify that is complete.
Indeed, is a Hilbert space if and only if . For , is only a Banach space.
Orthonormal Basis
Bessel's Inequality
Let be a Hilbert space, and let be an orthonormal set in . For any , there holds
Characterisation of Orthonormal Basis in Hilbert Space
The following properties of a countable orthonormal set in a Hilbert space are equivalent:
For all , . This is called the Parseval’s identity
Proof The Parseval’s identity is a direct result of the Pythagorean theorem.
Theorem
A Hilbert space has an orthonormal basis if and only if it is separable.
Proof Suppose is a separable Hilbert space and is a countable subset of whose closure equals . We will inductively define an orthonormal basis such that for all . This will imply that , which will mean that is an orthonormal basis.
Without loss of generality, assume that . First set . Then suppose for some , is an orthonormal set such that . If for every , then is an orthonormal basis of and the process should be stopped. Otherwise, let be the smallest positive integer such that Define as follows: Clearly , for all , our choice of guarantees there is no division by , and completing the induction.
Same Gram-Schmidt Procedure in Linear Algebra
This is actually the same Gram-Schmidt procedure in linear algebra.
e.g. Consider the space , the set is an orthonormal basis, and for each function , the corresponding th coefficient is the th Fourier coefficient: This can be seen by the following two facts:
The Parseval’s identity holds for :
The Fourier series converges to in the norm:
Indeed, the mapping makes (unitarily) “equivalent” to , we shall seriously define this equivalence, and prove this in the next section. (See the proposition.)
Unitary Mappings
Unitary Mapping
A mapping between two Hilbert spaces is called unitary if it is a vector space isomorphism and preserves the inner product, i.e. for all . We call two Hilbert spaces unitarily equivalent if there exists a unitary mapping between them.
Theorem
Any two separable Hilbert spaces are unitarily equivalent if and only if they have the same dimension.
Proof
Proposition
is unitarily equivalent to under the mapping , where .
Suppose we are given a pre-Hilbert space . Then there exists a unique (up to unitarily equivalence) Hilbert space , such that is dense, and is the restriction of to .
Proof
Fatou’s Theorem
Fatou's Theorem
A bounded holomorphic function on the unit disk has radial limits at almost every .