As is well known, one way of constructing the real numbers is to consider Cauchy sequences and call two of them equivalent if they have the same limit. I got to thinking about the Cauchy sequences themselves (ignoring the equivalence relation).

If we have a Cauchy sequence, we can make it shorter by removing the first element, or adding the first two elements. In fact we can do this arbitrarily often, creating a sequence of sequences.

Let the Cauchy sequence be $(a_n)_n$ and define $b_1 = (a_n)_n$ and $b_{n+1}$ by adding the first two elements of $b_n$ and attaching the remaining elements. Thus $b_2=(a_1+a_2,a_3,a_4,\dots), b_3=(a_1+a_2+a_3,a_4,\dots)$ etc.

I am having a hard time determining what the limit of $(b_n)_n$ would be as $n$ tends to infinity. Am I correct in surmising that the limit of $(b_n)_n$ generally does not exist, and even if it exists, the limit is not a sequence? Is it therefore correct to state that the set of Cauchy sequences is not complete, even though the set of their equivalence classes (= the real numbers) obviously is complete?


You are mixing up two systems of enumeration different things.

Your $a_n$ are numbers, given for each $n\geq1$, and the $n$ enumerates these numbers in the list $(a_n)_{n\geq1}$. Assume that $\lim_{n\to\infty} a_n=\alpha\in{\mathbb R}$. On the other hand, your $b_n$ are not numbers, but sequences, and the $n$ enumerates these sequences. I'd write it in the following way: $$\eqalign{ {\bf b}_1&=(a_1,a_2,a_3,a_4,\ldots),\cr {\bf b}_2&=(a_1+a_2,a_3,a_4,a_5,\ldots),\cr {\bf b}_3&=(a_1+a_2+a_3,a_4,a_5,\ldots),\cr &\vdots\cr {\bf b}_j&=(a_1+a_2+\ldots+a_j,a_{j+1},a_{j+2},a_{j+3},\ldots)\qquad(j\geq1).\cr}$$ Let ${\bf b}_{j.k}$ be the $k^{\rm th}$ element of the sequence ${\bf b}_j$. Then it is easy to see that for every $j\geq1$ one has $\lim_{k\to\infty}{\bf b}_{j.k}=\alpha$.

That's all one can say about the sequences ${\bf b}_j$, $\>j\geq1$. In particular; this has nothing to do with the completeness of ${\mathbb R}$.

  • $\begingroup$ Thank you! This helps a lot. Sorry for using $n$ twice. Still - suppose we have the sequence consisting of all ones. Then the first element of $b_j$ grows without bound as $j\to\infty$. Is the limit then even a sequence? Or is it not even possible to talk about the limit of a sequence of sequences? I find this topic really confusing. $\endgroup$ – Stefanie Jan 4 at 21:19

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.