I have a question on the proof of Thm.3.50 in Rudin's Principles of Mathematical Analysis. This theorem is about the convergence of the Cauchy product of two infinite series. For convenience, I write the necessary details of the theorem and proof.



(a) $\sum_{n = 0}^{\infty}a_n$ converges absolutely,

(b) $\sum_{n = 0}^{\infty}a_n = A$,

(c) $\sum_{n = 0}^{\infty}b_n = B$

(d) $c_n = \sum_{k = 0}^{n}a_kb_{n - k}.$

Then, \begin{align*} \sum_{n = 0}^{\infty}c_n = AB. \end{align*}


Put \begin{align*} A_n = \sum_{k = 0}^{n}a_k, \ \ B_n = \sum_{k = 0}^{n}b_k, \ \ C_n = \sum_{k = 0}^{n}c_k, \ \ \beta_n = B_n - B. \end{align*} Then we can rewtrite $C_n$ as \begin{align*} C_n = A_nB + a_0\beta_n + a_1\beta_{n-1} + \cdots + a_n\beta_0. \end{align*} Put \begin{align*} \gamma_n = a_0\beta_n + a_1\beta_{n-1} + \cdots + a_n\beta_0 \\ \end{align*} so that \begin{align*} C_n = A_nB + \gamma_n. \end{align*} Our aim here is to show $C_n \to AB$. Since $A_nB \to AB$, it suffices to show \begin{align*} \gamma_n \to 0. \end{align*} Put \begin{align*} \alpha = \sum_{n = 0}^{\infty}|a_n|. \end{align*} By (c), we have that $\beta_n \to 0$, or that for any $\varepsilon > 0$, there is some integer $N$ such that $|\beta_n| \leq \varepsilon$ for $n \geq N$. Hence, for sufficiently large n \begin{align} |\gamma_n| &\leq |\beta_0a_n + \cdots + \beta_Na_{n - N}| + |\beta_{N + 1}a_{n - N - 1} + \cdots + \beta_na_0| \\ &\leq |\beta_0a_n + \cdots + \beta_Na_{n - N}| + \varepsilon\alpha. \tag{1} \end{align} Keeping $N$ fixed, and letting $n \to \infty$, we get \begin{align} \limsup_{n \to \infty} |\gamma_n| \leq \varepsilon\alpha, \tag{2} \end{align} since $a_n \to 0$ as $n \to \infty$. Since $\varepsilon$ is arbitrary, we obtain $\lim_{n \to \infty}\gamma_n = 0.$ (end)


The quetion I have here is about (2). I see that, by triangular inequality and the fact that $|a_n| \to 0$, the term $|\beta_0a_n + \cdots + \beta_Na_{n - N}|$ goes to zero with $N$ fixed. Because we take the limit of (1), it seems to me that we obtain \begin{align} \lim_{n \to \infty} |\gamma_n| \leq \varepsilon\alpha \tag{3} \end{align} instead of (2). I notice that we get (3) if (2) holds since $|\gamma_n|$ is nonnegative, but I feel we can derive (3) directly from taking the limit of (1). Am I wrong in some points?

Any comment would be appriciated. Thank you!


If $|t_n| \leq s_n$ and $\lim s_n=s$ we cannot say that $\lim t_n$ exists. What we can say is $\lim \sup |t_n| \leq \lim \sup s_n=\lim s_n=s$ since $\lim \sup$ and $\lim $ are the same for a convergent sequence $(s_n)$.

Example: Let $t_n=0$ for $n$ even and $t_n=1$ for $n$ odd. Let $s_n=1$ for all $n$. Then $\lim t_n$ does not exist even thoug $0 \leq t_n \leq s_n$ and $s_n \to 1$.

  • $\begingroup$ Thank you so much for your quick reply! The example you provided also helps a lot. $\endgroup$
    – mn24
    Dec 28 '20 at 7:05

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.