Why can't the ratio test be used for geometric series?

The usual proof of the ratio test is to compare the series to a geometric series. If $$\lim_{n\to\infty} \left| \frac{a_{n+1}}{a_n} \right| = \alpha < 1,$$ then we have $$ |a_{n+1}| < |a_n| \alpha $$ for all sufficiently large $n$. It then follows from an induction argument that $$ |a_{n+k}| < |a_n| \alpha^k $$ for $n$ sufficiently large, which means that $$ \sum_{j=1}^{\infty} |a_j| = \sum_{j=1}^{n-1} |a_j| + \sum_{j=n}^{\infty} |a_j| \le \sum_{j=1}^{n-1} |a_j| + \sum_{k=0}^{\infty} |a_n| \alpha^k.$$ The first series has only finitely many terms and is therefore finite, and the second series is geometric and therefore converges by some other argument. From this, it follows that if $$ \lim_{n\to\infty} \left| \frac{a_{n+1}}{a_n} \right| < 1, \qquad\text{then}\qquad \sum_{j=1}^{\infty} a_j $$ converges (by the limit comparison test, for example). The other inequality in the ratio test can be argued by noting that the general term does not go to zero, and the uncertainty at 1 can be argued by considering (for example) the harmonic and alternating harmonic series.

The key point is that we prove that the ratio test implies the convergence of series by comparison to a convergent geometric series. Since the proof of the ratio test relies on this convergence, it is circular to argue that a geometric series converges by the ratio test (unless, of course, you have another proof for the ratio test that doesn't use the convergence of geometric series).

While one could use the ratio test to establish the convergence of a geometric series (there is nothing stopping us!), it is typically poor style to rely on circular arguments as it can (potentially) lead to overlooking important hypotheses or exceptional cases. This is particularly important in a pedagogical setting, when students may not be entirely cognizant of the line of reasoning that lead up to a result (it is hard to keep track of all of the lemmata, theorems, and proofs that lead up to a result if it is the first time that you have had to deal with them).

Moreover, I don't see why one would want to use the ratio test to show that a geometric series converges. Basically no computation is needed to show that a geometric series converges, while a couple of computational steps are needed in order to invoke the ratio test.


We can't know for sure (unless context from the notes you've omitted says something on the topic), but the various comments appearing in this topic suggest an explanation.

Two major activities mathematicians (and people who use mathematics) engage in are:

  • Using mathematical tools to do calculations and prove things
  • Designing and validating the mathematical tools used in the above bullet

The ratio test is an example of such a mathematical tool, and is perfectly applicable to geometric series. It may even be the preferable tool for deciding when a geometric series converges, simply to cut down on the number of things one needs to remember.

However, if you happen to be in the process of validating the ratio test, it would not be valid to use the ratio test to justify any of the facts you need — such as when geometric series converge — in its validation. You would have to first derive the facts about geometric series in some other fashion.

As for your notes, the most likely explanations are:

  • You are reading notes about how to validate the ratio test
  • You have misunderstood the notes
    • Possibly due to the author failing to actually convey his meaning
  • The author of these notes was confused

To elaborate on that last point, people often get stuck in the "tool building" mindset. A fair amount of mathematics education involves validating the tools one is already familiar with, which requires temporarily suspending the use of those tools so that we can see how they can be built from more basic tools — and sometimes people learn the wrong lesson and think that's something you always have to do.

So, when facts about convergence of geometric series in the proof of the ratio test, this sometimes leads to people mistakenly thinking that any time the topic of convergence of geometric series comes up, one must suspend their use of the ratio test and argue from more basic tools.

Tags:

Analysis

Sequences And Series

Real Analysis

Power Series

Geometric Series

Related

Could the orbit of $0$ through a polynomial be dense in $\mathbb R$? $\cos(\theta_n) \to \cos(\theta)$ and $\sin(\theta_n) \to \sin(\theta)$. How to show that $\theta_n \to \theta$? Why log-of-sum-of-exponentials $f(x)=\log\left(\sum_{i=1}^n e^ {x_i}\right)$ is a convex function for $x \in\mathbb R^n$ Infinite 'hex': is a win always achievable? What is the general solution to equations of form $f(k\cdot x)-lf(x)=1, k,l \in \Bbb N$ ? Prove that $\log_{10} x$ is the solution when $k = 10,l =1$. Is it possible to determine if you were on a Möbius strip? Does $\operatorname{tr} (A)=0$ imply $\operatorname{tr} (A^3)=0$? Prove $\ln(1+x)\geq x-\frac{x^2}{2}$ Probability that the sum of $k$ distinct integers selected from $1, 2, \dots, n$ is divisible by $n$ Criterion for deciding whether matrix is diagonalizable What is the negation of the implication statement Conditional expectation as a Radon-Nikodym derivative.

Recent Posts