Intereting Posts

Bubble sorting question
Solving base e equation $e^x – e^{-x} = 0$
Refining Rudin's proof of $\lim \left (1+\frac 1 n\right)^n =\lim \sum_{k=1}^n \frac {1}{k!}$.
Showing that the sequence $x_n=\frac{1}{3}x_{n-1}(4+x_{n-1}^3)$ where $x_0=-0.5$ quadratically converges
Let n>30. Then prove there exists a natural number $ 1< m\leq n$ such that $(n,m)=1$ and $m$ is not prime.
Find an example of a sequence not in $l^1$ satisfying certain boundedness conditions.
Great books on all different types of integration techniques
Does a $\Pi_2^0$ sentence becomes equivalent to a $\Pi_1^0$ sentence after it has been proven?
How to solve $a_n = 2a_{n-1} + 1, a_0 = 0, a_1 = 1$?
Prove some properties of determinant by the universal property of the tensor product of vector spaces
Calculating Dini derivatives for $f(x)=\begin{cases}x\,\sin{\left(\frac{1}{x}\right)} & x\neq 0\\ 0 & x=0\end{cases}$
Expression for normal product similar to semidirect product
How to find $\lim_{x\to 0} \frac{1-\cos x \sqrt{\cos 2x}}{x^2}$
Find a positive integer $n$ such that $ϕ(n) = ϕ(n + 1) = ϕ(n + 2)$
Evaluating $\int_0^{\infty} \text{sinc}^m(x) dx$

I am looking at the proof of the following theorem from Apostol’s Mathematical Analysis. I am having trouble showing the last part that the author left to the reader.

I’m trying to show that $y$ is the limit superior of this rearrangement. To do so, I need to show two things based on the definition of limit superior from the text. First, for every $\epsilon \gt 0$ there is an integer $N$ such that $n \gt N$ implies $h_n \lt y + \epsilon$.

- differential inverse matrix
- If $f^3=\rm id$ then it is identity function
- Calculate price score
- Contraction Mapping question
- Prove that $C^1()$ with the $C^1$- norm is a Banach Space
- Compactness in the weak* topology

Second, given $\epsilon \gt 0$ and $m \gt 0$, there is an integer $n \gt m$ such that $h_n \gt y-\epsilon$. Where I put $h_n$ as the rearrangement of the original series.

The second condition is immediately satisfied since for every $y_n$ there is a rearrangement greater than it by construction. However, I’m having trouble showing the first part. How can I guarantee that for any $\epsilon$, all but finitely many $h_n$ is less than $y+ \epsilon$. I don’t know how to show this part since our construction only guarantees that we have some rearrangement greater than every $y_n$.

Finally, how does this theorem lead to the conclusion that any conditionally convergent series of real terms can be rearranged to yield a series which converges to any prescribed sum?

I’d greatly appreciate it if anyone could rigorously establish the above facts for me.

- Every closed subspace of ${\scr C}^0$ of continuously differentiable funcions must have finite dimension.
- Lebesgue measure as $\sup$ of measures of contained compact sets
- Limit of solution of linear system of ODEs as $t\to \infty$
- A question about Hölder spaces.
- Why the $\nabla f(x)$ in the direction orthogonal to $f(x)$?
- Show that the union of two sets of measure zero, has a measure zero
- A minimization problem for a function involving maximum
- Can a norm take infinite value? For example, $\|\cdot \|_1$?
- Proving if $\limsup x_n = \liminf x_n = c$, then $x_n \rightarrow c, n \rightarrow \infty$ using $\epsilon$
- A question on second symmetric derivatives

Since the original series $\sum a_{n}$ is convergent it follows that $p_{n}, q_{n}$ both tend to $0$.

Next note that $y_{n} \to y$ so that for any $\epsilon > 0$ you have a value $n_{0}$, such that $y_{n} < y + \epsilon/2$ for $n > n_{0}$. If you note carefully the selection of terms from sequences $p_{n}, q_{n}$ to make it just greater than $y_{n}$, you will notice that removing the last term of type $p_{n}$ will make the sum less than $y_{n}$. Since $p_{n} \to 0$, it follows that we will reach a point when this last term in the sum corresponding to $y_{n}$ is less than $\epsilon / 2$ and therefore the difference of this sum from $y_{n}$ is less than $\epsilon / 2$. It follows that the sum itself will be less than $y_{n} + \epsilon / 2$. Thus we can find the finite sums of type $\sum p – \sum q$ such that $$y_{n} < \sum p_{k_{i}} – \sum q_{k_{j}} < y_{n} + \epsilon / 2 < y + \epsilon $$ This is the way we find a value $n_{1}$ such that $t_{n} < y + \epsilon$ for all $n > n_{1}$.

If you want to make the rearranged series $\sum b_{n}$ to converge to a specific value $x$ simply choose $x = y$ in the construction mentioned in your post so that $\liminf\, t_{n} = \limsup\, t_{n} = \lim\, t_{n} = x$.

- Divergence of $\sum\limits_{k=1}^n \frac{a_n}{S_n}$ where $S_n = \sum\limits_{k=1}^n a_k$, when $S_n$ diverges
- Evaluating $\int_{0}^{\pi}\ln (1+\cos x)\, dx$
- 9 hidden in every number regardless of number of digits?
- A neater solution to an interesting question part 2
- Determinant of a large block matrix
- relation between integral and summation
- Prove the Countable additivity of Lebesgue Integral.
- Uniqueness of Duals in a Monoidal Category
- $\sum_{k=1}^{\infty} \frac{a_1 a_2 \cdots a_{k-1}}{(x+a_1) \cdots (x+a_k)}$
- Simple Math Problem
- $\chi(G)+\chi(G')\leq n+1$
- Morphism from a line bundle to a vector bundle
- Pre-requisites needed for algebraic number theory
- How do I prove that $2^n=O(n!)$?
- How do you integrate $\int \frac{1}{a + \cos x} dx$?