Let's consider a linear transformation $T$ of sequences:

Suppose $\{c_{jk}\}$ is a matrix of complex numbers. For each sequence $\alpha=\{a_n\}$ made up of complex numbers, $\beta=T\alpha$ is a sequence, as well, such that $b_n=\sum_{k=1}^\infty c_{nk}a_k$.

I need a necessary and sufficient condition that $\{b_n\}$ converges whenever $\{a_n\}$ converges.

For example, according to Silverman-Toeplitz theorem, if $\lim_{j\to\infty}c_{jk}=0$, $\lim_{j\to\infty}\sum_{k=1}^\infty c_{jk}=1$ and $\sum_{k=1}^\infty\lvert c_{jk}\rvert$ is uniformly bounded according to different $j$'s, $\{b_n\}$ converges to the same number where $\{a_n\}$ converges.

However, I don't need such a condition that both sequences converges to a same number. For example, $$\sum_{n=1}^\infty d_na_n$$ converges whenever $\{a_n\}$ converges, by virtue of Cauchy's convergence test, so set $c_{jk}=d_k$, we've got a transformation, not of Toeplitz-type.


I would say that the matrix $(c_{jk})$ does what you want if and only if (i) $\sum_{k=1}^\infty \vert c_{jk}\vert$ is uniformly bounded with respect to $j$, (ii) $\lim_{j\to\infty} c_{jk}$ exists for every $k$ and (iii) $\lim_{j\to\infty} \sum_{k=1}^\infty c_{jk}$ exists.

Condition (ii) and (iii) are obviously necessary: to check (ii), consider for fixed $k$ the sequence $a=(0,0,\dots ,1,0\dots)$ where the $1$ is at place $k$); and for (iii) consider the constant sequence $a_n\equiv 1$. That (i) is also necessary follows from the Banach-Steinhaus theorem (as in the proof of Silverman-Toeplitz). Conversely, if (i) and (ii) holds, then it is not hard to prove that the matrix $(c_{jk})$ maps sequences $(a_n)$ tending to $0$ to Cauchy (and hence convergent) sequences; and if moreover (iii) holds this gives the result for all convergent sequences (by substracting the limit).

  • $\begingroup$ It seems good. For (i): the space $V$ of convergent sequence is a closed and bounded subspace of $\mathcal l^\infty$, therefore complete. Let $\Lambda_n\alpha=\sum_{k=1}^\infty c_{nk}a_k$. Since $\{\Lambda_1\alpha,\Lambda_2\alpha,\dotsb\}$ is bounded for each $\alpha$, by virtue of Banach-Steinhaus theorem, it's uniformly bounded. Am I right? Since I'm a novice to real analysis, I'm not sure whether my inference is right. $\endgroup$ – Yai0Phah Jun 1 '13 at 15:01
  • $\begingroup$ In addition, such a matrix could be decomposed as the summation of a Silverman-Toeplitz matrix and copies of a sequence whose sum absolutely converges. $\endgroup$ – Yai0Phah Jun 1 '13 at 15:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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