Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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

Prove or disprove: A power series that converges on $\left[0,1\right]$ is uniquely determined by any sequence of points converging to $1$.

More precisely, let $f, g:\left[0,1\right]\rightarrow\left[0,\infty\right)$ by expandable into power series $$\begin{align}\forall x\in\left[0,c\right] &\bullet f\left(x\right)=\sum_{i=0}^\infty a_ix^i\\ \forall x\in\left[0,c\right] &\bullet g\left(x\right)=\sum_{i=0}^\infty b_i x^j\end{align}$$ with non-negative coefficients that sum to $1$: $$\left\{\left.a_i, b_i\space:\right|i\in\mathbb{N}_0\right\}\subseteq\left[0,\infty\right)$$ $$\sum_{i=0}^\infty a_i = 1 = \sum_{i=0}^\infty b_i$$

Let $\left(s_j\right)_{j=0}^\infty$ be a strictly ascending sequence of points in $\left(0,1\right)$ that converges to $1$.

Is it true that $$\left(\forall j\in\mathbb{N}_0\bullet f\left(s_j\right)=g\left(s_j\right)\right)\implies\left(\forall i\in\mathbb{N}_0\bullet a_i=b_i\right)$$


  1. If the radii of convergence $R_f, R_g$ were both $>1$, or if $\left( s_j\right)_{j=0}^\infty$ converged to any number $\in\left[0,1\right)$, the claim would be an immediate consequence of Baby Rudin's Theorem 8.5:

    Suppose the series $\sum a_n x^n$ and $\sum b_n x^n$ converge in the segment $S=\left(-R, R\right)$. Let $E$ be the set of all $x\in S$ at which $$\sum_{n=0}^\infty a_n x^n = \sum_{n=0}^\infty b_n x^n$$ If $E$ has a limit point in $S$, then $a_n=b_n$ for $n=0,1,2,\dots$

  2. The claim derives from Klenke, where it is stated as true without proof. (Theorem 3.2(iii))

share|cite|improve this question
up vote 3 down vote accepted

It's not true. Consider the following series: $$ \eqalign{f(x) &= \sum_{j=1}^\infty \frac{3}{2^{2j}} x^{n_{2j-1}}\cr g(x) &= \frac{1}{2} + \sum_{j=1}^\infty \frac{3}{2^{2j+1}} x^{n_{2j}}\cr}$$ where $n_j$ is an increasing sequence of positive integers to be chosen. Thus $$h(x) = g(x) - f(x) = \frac{1}{2} + \frac{3}{2}\sum_{k=1}^\infty (-1)^k \frac{x^{n_k}}{2^k}$$ Let $$h_N(x) = \frac{1}{2} + \frac{3}{2}\sum_{k=1}^N (-1)^k \frac{x^{n_k}}{2^k}$$ Note that $(-1)^N h_N(1) = (1/2)^{N+1} > 0$. I will choose the $n_j$ and an increasing sequence $t_j \in [0,1)$ inductively so that $(-1)^j h(t_j) > 0$. This can be done as follows. We start with $t_0 = 0$, so $h(t_0) = 1/2 > 0$. Given $N$, $t_{N-1}$ and $n_1, \ldots, n_N$, take $t_N \in (t_{N-1},1)$ so close to $1$ that $(-1)^N h_N(t_N) > 0$. Since $|h(t) - h_N(t)| < 3 t^{n_{N+1}}/2^{N+1}$, by taking
$n_{N+1}$ large enough we can ensure that $(-1)^N h(t_N) > 0$.

Now note that by the Intermediate Value Theorem, for each $N$ there is $s_N$ with $t_N < s_N < t_{N+1}$ such that $h(s_N) = 0$.

share|cite|improve this answer
Thank you! It's a beautiful counter-example. I should have been more careful and posed my question as: "Prove or disprove..." – Evan Aad Sep 5 '12 at 7:15

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.