1
$\begingroup$

In a different thread, we stumbled upon the following question:

Given a continuous finite measure $w(x)dx$ on some interval $(a,b)$, $-\infty \leq a <b \leq \infty$, and the set of respective orthogonal polynomials $\{p_n \}_{n=0}^{\infty}$. We follow Szego and define the kernel as $$ K_N (t,x) = \sum\limits_{\nu =0}^N \bar{p_{\nu} }(t) p_{\nu}(x).$$

Question 1: Since for every polynomial $q$ of small enough a degree, we have $\int\limits_a^b K_N(t,x)q(t) \, dt = q(x)$, can say that $K_N(t,x)$ converges in $w^*$ topology? If so, that what is its limit?

Edit: In Aronszajn (1950), section 9, it seems clear that $K_n$ themselves converge in $L^2$ norm.

Question 2: If (1) is true, does $K_N$ converges in a stronger sense? pointwise, almost everywhere etc.

$\endgroup$

1 Answer 1

1
$\begingroup$

It seems that both questions receive some answers in this paper.

In th $L^2$ norm, Section 9 seems to stipulate convergence and a limit $K$.

For the uniform and pointwise convergence, we have on p. 374, Theorem II, that since the orthogonal projections converge, we have pointwise convergence of $K_n(x,y)$ to $K$. Also, at least for some polynomial sets (Jacobi and Legendre specifically), we have uniform convergence of the projection, which implies uniform convergence of $K_n$.

Edit: After further searching, I've found out that this is the Christoffel-Darboux kernel. Here is a broad survey about this family of kernels.

$\endgroup$
2

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.