I am curious about why most of the continuous transforms, for instance fourier/ Laplace are based on integral operator?
OR,
In other words, why integral operator provides the "best" correlation between two functions?
Thanks
I am curious about why most of the continuous transforms, for instance fourier/ Laplace are based on integral operator?
OR,
In other words, why integral operator provides the "best" correlation between two functions?
Thanks
From the point of Functional Analysis, we consider the operator's space(or domain & range). Taking Fourier Transform as an example, we can see it like this: $T:C(R^n)\rightarrow{C(R^n)}$ or $T:L^2(R^n)\rightarrow{L^2(R^n)}$. So we enlarge the space(or domain & range) through Lebesgue Integral and we will find more beautiful and smooth in $L^2(R^n)$ than $C(R^n)$. Actually, we can make a larger(or more suitable) space(or domain & range) by using integral operator.
We can also see this question from the Representation Theory. If we look this question carefully, we will find that $T:L^2(R^n)\rightarrow{L^2(R^n)}$ is a transform from a space into its dual space. It is a way that studying a space by studying its dual space. So we can see this question from the Representation Theory:
(1)Fourier Transform: $T:L^2(R^n)\rightarrow{L^2(R^n)}$. It is a Representation Theory about group $(R^n,+)$.
(2)Fourier Series: It is about the transform of period functions to series and converse. First, we can see that period functions is defined on $R^n/Z^n$ or $n$ dimensional torus $T^n$ which is still a group by seen as the product $S$. Second, series can be seen a function defined on group $Z^n$. So transform is $T:L^2(T^n)\rightarrow{L^2(Z^n)}$ and $T:L^2(Z^n)\rightarrow{L^2(T^n)}$. It is a Representation Theory about $T^n$ and $Z^n$.
From above, we know the integral operator may be related to the Topological Group Representation Theory and this will use some integral or precisely, Haar Measure.
The book about this: "A First Course in Harmonic Analysis" #http://en.bookfi.org/book/443200#