1
$\begingroup$

Let $H$ be an $n\times n$ Hilbert matrix, $$h_{ij}=(i+j-1)^{-1}.$$

The matrix $p$-norm corresponding to the p-norm for vectors is: $\left \| A \right \| _p = \sup \limits _{x \ne 0} \frac{\left \| A x\right \| _p}{\left \| x\right \| _p}$, $p\ge 1$.

Is there a known (or what is the) formula for $\left \| H \right \| _p$ in terms of $n$ and $p$?

I saw a related post for $n=\infty$. https://math.stackexchange.com/questions/837832/norm-of-hilbert-matrix-is-it-equal-to-pi

After reading the post Spectral norm for a truncated Hilbert matrix I guess the answer to my query is no... so a big challenge.

$\endgroup$

1 Answer 1

2
$\begingroup$

I think the answer is no, indeed. Even in the particular case $p=2$ the formula is not known (to my best knowledge). Concerning $\|H\|_{2}$, is known $\|H\|_{2}\leq\pi$. More precisely, we have the following upper bound $$\|H\|_{2}\leq 2w_n\arcsin\frac{1}{w_n}$$ where $$w_n=2\left[\binom{2n}{n}\right]^{-1/2n},$$ see P. Otte: Upper bounds for the spectral radius of the $n\times n$ Hilbert matrix, Pacific J. Math. 219 No. 2 (2005) 101-109.

In addition, quite precise asymptotic formula for $\|H\|_{2}$ is known as $n\to\infty$. Namely, $$\|H\|_{2}=\pi-\frac{\pi^{5}}{2\ln^{2} n}+O\left(\frac{\ln\ln n}{\ln^{3} n}\right),$$ see N.G. de Bruijn and H.S. Wilf, On Hilbert’s inequality in n dimensions, Bull. Amer. Math. Soc. 68 (1962), 70–73.

$\endgroup$
1
  • $\begingroup$ My belief is that, if $n$ and $p$ are fixed, the norm should be evaluated... though maybe very hard. $\endgroup$
    – M. Lin
    Apr 8, 2015 at 19:21

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.