[The Algebraic Limit Theorem] Let \(\lim a_n = a\) and \(\lim b_n = a\). Then,
  1. \(\lim (ca_n) = ca\) for all \(c \in \mathbf{R}\);
  2. \(\lim (a_n + b_n) = a + b\);
  3. \(\lim (a_nb_n) = ab\);
  4. \(\lim (a_n/b_n) = a\);


For the definitions of sequences and what it means to for a sequence to converge, see this, for subsequences see this.
For the “show the limit” template and an example, see this.

Problem Discussion

We’ll follow a similar approach to proving (i) and (ii) in here and here For (iii) we want to show that \(\lim (a_nb_n) = ab\). So we want to find \(N\) such that for any \(\epsilon > 0\),

$$ \begin{align*} \lvert a_nb_n - ab \rvert < \epsilon. \end{align*} $$

The trick here to add and subtract a term \(ab_n\) and then use the triangle inequality:

$$ \begin{align*} \lvert a_nb_n - ab \rvert &= \lvert a_nb_n + ab_n - ab_n - ab \rvert \\ &\leq \lvert a_nb_n - ab_n \rvert + \lvert ab_n - ab \rvert \\ &= |b_n| \lvert a_n - a \rvert + |a|\lvert b_n - b \rvert. \end{align*} $$

Using the same trick from (ii), we can make \(\lvert a_n - a \rvert\) be as small as we want and similarly we can make \(\lvert b_n - b \rvert\) be as small as we want. The idea is to bound these two terms to values such that the total will be exactly \(\epsilon\) which is what we want to prove. So, given that \(\lim b_n = b\), then we know that for some \(n \geq N_1\),

$$ \begin{align*} |a|\lvert b_n - b \rvert \leq \frac{\epsilon}{2}. \\ \lvert b_n - b \rvert \leq \frac{\epsilon}{2} \frac{1}{|a|}. \\ \end{align*} $$

Similarly, given that \(\lim b_n = b\), then we know that for some \(n \geq N_2\),

$$ \begin{align*} |b_n|\lvert a_n - a \rvert \leq \frac{\epsilon}{2} \\ \lvert a_n - a \rvert \leq \frac{\epsilon}{2} \frac{1}{|b_n|}. \end{align*} $$

We still need to bound the term \(|b_n|\) since it’s not a constant. One trick is to use the convergent sequences theorem which says that all convergent sequences are bounded by some term \(M\). This means that we can replace the bound above with the following:

$$ \begin{align*} \lvert a_n - a \rvert < \frac{\epsilon}{2} \frac{1}{M}. \end{align*} $$

If we plug in everything back, we get

$$ \begin{align*} \lvert a_nb_n - ab \rvert &= \lvert a_nb_n + ab_n - ab_n - ab \rvert \\ &= \lvert a(b_n - b) + b_n(a_n - a) \rvert \\ &\leq \lvert a(b_n - b) \rvert + \lvert b_n(a_n - a) \rvert \quad \quad \text{(Triangle Inequality)}\\ &= |a|\lvert b_n - b \rvert + |b_n|\lvert a_n - a \rvert \quad \quad \text{(\(|ab| = |a||b|\))} \\ &< |a|\big(\frac{1}{|a|}\frac{\epsilon}{2}\big) + M\big(\frac{1}{M}\frac{\epsilon}{2}\big). \\ &= \frac{\epsilon}{2}+ \frac{\epsilon}{2} = \epsilon. \\ \end{align*} $$

Finally we can choose \(N\) to be the maximum of \(N_1\) and \(N_2\) to guarantee the above bounds.

Formal Proof

Let \(\epsilon > 0\) be arbitrary. We are given that \(\lim a_n = a\). This means that for some \(n \geq N_1\) and for any choice of \(\epsilon\) , the following holds:

$$ \begin{align*} \lvert a_n - a \rvert \leq \epsilon. \end{align*} $$

Since this holds for any \(\epsilon\), let \(\epsilon = \big(\frac{1}{|a|}\frac{\epsilon}{2}\big)\) so,

$$ \begin{align*} \lvert a_n - a \rvert \leq \frac{1}{|a|}\frac{\epsilon}{2}. \end{align*} $$

Similarly, given that \(\lim b_n = b\). This means that for some \(n \geq N_2\), the following holds:

$$ \begin{align*} \lvert a_n - a \rvert \leq \frac{1}{M}\frac{\epsilon}{2}. \end{align*} $$

\(M\) is an upper bound on the sequence \((b_n)\) guaranteed by convergent sequences theorem. Therefore, choose a natural number \(N\) satisfying \(\max(N_1,N_2)\). We now verify that this choice has the desired property. Let \(n \geq N\). Then,

$$ \begin{align*} \lvert a_nb_n - ab \rvert &= \lvert a_nb_n + ab_n - ab_n - ab \rvert \\ &= \lvert a(b_n - b) + b_n(a_n - a) \rvert \\ &\leq \lvert a(b_n - b) \rvert + \lvert b_n(a_n - a) \rvert \quad \quad \text{(Triangle Inequality)}\\ &= |a|\lvert b_n - b \rvert + |b_n|\lvert a_n - a \rvert \quad \quad \text{(\(|ab| = |a||b|\))} \\ &< |a|\big(\frac{1}{|a|}\frac{\epsilon}{2}\big) + M\big(\frac{1}{M}\frac{\epsilon}{2}\big). \\ &= \frac{\epsilon}{2}+ \frac{\epsilon}{2} = \epsilon, \\ \end{align*} $$

as we wanted to show. \(\blacksquare\)

References: