MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Consider the class of bounded sequences to which every Banach limit (non-negative shift-invariant continuous functional on $l^\infty$ taking convergent sequences in the usual sense to their limits) assigns the same limit value.

Does a sequence belong to this class if its Cesaro means have a limit?

Also, is the converse true?

share|cite|improve this question
I like the question, but I would like to make sure I understand correctly the notion of Banach limit. From your formulation I understand that a Banach limit is a linear map $L:\ell^infty(\mathbb{R})\to\mathbb{R}$ with the property that $L(\;(x_n)_{n\geq 0}\;)=L(\;(y_n)_{n\geq 0}\;) $ if $x_n=y_n$ for all $n$ sufficiently large. Is the functional $L$ continuous in the $\ell^\infty$ topology? – Liviu Nicolaescu Apr 8 '12 at 20:29
If $L$ is a Banach limit, then $|Lx|\le\|x\|_\infty$ for any sequence $x$. Does this answer your question? – kap44 Apr 8 '12 at 20:34
Yes, thanks. $L$ is therefore an element of the topological dual of $\ell^\infty$. – Liviu Nicolaescu Apr 8 '12 at 21:21
Sequences with this property are called almost convergent and there exists a well-known characterization of such sequences due to Lorenz; which is described in the Wikipedia article. One possibility how to show this is using Hahn-Banach theorem, see this answer at Math.SE. Also the references from the Wikipedia article might be useful. – Martin Sleziak Apr 9 '12 at 13:47
up vote 12 down vote accepted

We can characterize Banach limits as continuous functionals on $\ell^\infty$ which vanish on $$ X := \{(x_n - x_{n+1}): (x_n) \in \ell^\infty\} $$ and which send the constant sequence $(1,1,\dots)$ to $1$.

Note that $X$ is a subspace. The Hahn-Banach Theorem tells us that we are asking: if $(y_n) \in \ell^\infty$ has Cesaro mean $0$, is it in the closure of $X$? (And the converse question is: does every element of $X$ have Cesaro mean $0$? Yes; since the $n^\text{th}$ Cesaro mean of $(x_n-x_{n+1})$ is $(x_1-x_{n+1})/n$, which converges to $0$ since $(x_n)$ is uniformly bounded.)

The answer is no. Consider the sequence $(y_n)$ that has $1$ once, followed by $-1$ three times, then $1$ five times, and so on. One can compute the Cesaro mean, and see that it approaches $0$ in the limit. But $(y_n)$ is not in the closure of $X$.

Surely, if it were, then let $(x_n) \in \ell^\infty$ be such that $$ \|(y_n) - (x_n-x_{n+1})\|_\infty < 1/2. $$ Let $M$ be a natural number, $M \geq \|(x_n)\|$. Let $n$ be an index such that $$ y_n = \cdots = y_{n+4M} = 1. $$ Then for $i=1,\dots,4M$, $$ x_{n+i} < x_{n + i-1} - y_{n + i - 1} + 1/2 = x_{n + i - 1} - 1/2, $$ and summing these up, we find $$ x_{n+4M} < x_n - 4M/2. $$ This contradicts the assumption that $\|(x_n)\| \leq M$.

share|cite|improve this answer
Is there a typo in the second equation from the bottom? In the middle expression you write $x+y$, where must be $x-y$. (I believe the idea does work anyway!) Also, it is not quite obvious why every element of $X$ has Cesaro mean 0. – kap44 Apr 9 '12 at 7:03
Thank you, I think that it is all correct now. I have added a bit of explanation about why the elements of $X$ have Cesaro mean zero. – Aaron Tikuisis Apr 9 '12 at 10:40
Many thanks to you! – kap44 Apr 9 '12 at 10:45
Nice answer! By the way? Is there a constructive example of Banach limit functional? – Liviu Nicolaescu Apr 9 '12 at 13:43
@Liviu: I believe there isn't. See‌​banach-theorem/5366 – Mark Meckes Apr 9 '12 at 16:23

If $(x_n) \in \ell^\infty$. According to Lorenz the Banch limit is unique (also known as almost convergent) iff $$\lim_{p\mapsto\infty} \frac{ x_n + x_{n+1} + \cdots + x_{n+p}}{p} = L \quad (*) $$ uniformly in $n$. Setting $n=0$ yields Cesaro summability.

As Aaron says, the converse is false. If each $x_n$ is chosen uniformly at random from $\{0,1\}$ then this sequence almost never has property $(*)$ (see Connor's appropriately named article Almost none of the sequences of 0's and 1's are almost convergent)

However the Cesaro limit of this random sequence $(x_n)$ is almost always $1/2$ by the law of large numbers.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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