Let K:T×T→C be a covariance function such that the associated RKHS HK is separable where T⊂R. Then there exists a family of vector functions
Ψ(t,x)=(ψn(t,x),n≥1)∀t∈T |
and a Borel measure μ on T such that ψn(t,x)∈L2(T,μ) in terms of which K is representable as:
K(s,t)=∫T∞∑n=1ψn(s,x)¯ψn(t,x)dμ(x) |
The vector functions Ψ(s,.),s∈T and the measure μ may not be unique, but all such (Ψ,.),.) determine K and its reproducing kernel Hilbert space (RKHS) HK uniquely and the cardinality of the components determining K remains the same. [1, ]
Remark
K(s,t)=∫TΨ(s,x)¯Ψ(t,x)dμ(x) |
which includes the tri-diagonal triangular covariance with μ absolutely continuous relative to the Lebesgue measure.
2. The following notational simplification of (25) can be made. Let n=R×Z+=S⊗P, where P is the power set of integers Z, and let P = u @ o where o is the counting measure. Then
Ψ(t,n)=(ψn(t,x),n∈Z) |
Hence
|Ψ∗(t)|2L2=∫T|ψn(t,x)|2dμ(x) |
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.