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

I don't understand an equation I am reading in my notes:

Suppose, $|\cdot |$ is a nonarchimedean absolute value on a field $K$ complete wrt this absolute value. Suppose, $|a_0|>|a_i|$ for all $i>0$. Then, $|a_0+a_1+a_2+...|=|a_0|$. I understand that $|a_0+a_1+a_2+...|\leq max\{|a_i|:i\geq 0\}=|a_0|$ by ultrametric inequality. I am not sure how to get the reverse inequality.

share|cite|improve this question
up vote 6 down vote accepted

Let $a = \sum a_i$. You can apply the argument to $a_0 = a- (a_1 + \dots + a_n)$. We have then that $$ |a_0| \leq \max(|a|, |a_1 + \dots a_n|) = |a|$$ because $|a_0|$ is bigger than $|a_1 + \dots + a_n|$, and consequently the second term cannot contribute to the maximum. This implies that $|a| \geq |a_0|$, which is the reverse inequality you wanted.

share|cite|improve this answer
That's what I was attempting. However I am not sure why, $|a_0|>|a_1+a_2+...|$. – Derek Scavo Dec 5 '10 at 7:16
@Derek: This follows from $|a_1 + \dots + a_n| \leq \max_{i>0} |a_i| < |a_0|$, which is in the hypotheses. – Akhil Mathew Dec 5 '10 at 7:17
Damn, I should have seen that. Thanks a lot. – Derek Scavo Dec 5 '10 at 7:18
I think this argument has circular reasoning in it (the part where you say "consequently the second term cannot contribute") so I did the proof thoroughly in an own answer. – born Sep 2 '12 at 20:36

$\newcommand{\set}[1]{\left\{#1\right\}} \newcommand{\jleq}[1]{\justifyed{#1}{\leq}} \newcommand{\jeqtxt}[1]{\jeq{\text{#1}}} \newcommand{\jl}[1]{\justifyed{#1}{<}} \newcommand{\justifyed}[2]{\stackrel{#1}{#2}} \newcommand{\jeq}[1]{\justifyed{#1}{=}} \newcommand{\jleqref}[1]{\jleqtxt{\ref{#1}}} \newcommand{\jeqref}[1]{\jeqtxt{\ref{#1}}} \newcommand{\jleqtxt}[1]{\jleq{\text{#1}}} \newcommand{\df}{\mathrel{\mathop:}=} \newcommand{\norm}[1]{\left|#1\right|}$ Isosceles triangle principle: Let $\norm{\cdot}$ be a non-Archimedean norm on a field $\mathbb{F}$. Then $\forall\ x,y\in\mathbb{F}$ with $\norm{x}<\norm{y}$ it holds that $\norm{x+y}=\norm{y}$.

Proof: Let $x,y\in\mathbb{F}$ with $\norm{x}<\norm{y}$. We now calculate \begin{align} \norm{x-y}\jleqtxt{non-Arch.}\max\set{\norm{x},\norm{y}}=\norm{y}=\norm{x-(x-y)}\jleqtxt{non-Arch.}\max\set{\norm{x},\norm{x-y}}. \end{align} From this equation we get that $\norm{y}\leq\max\set{\norm{x},\norm{x-y}}$ but since we assumed $\norm{x}<\norm{y}$, we have that $\norm{y}\nleq\norm{x}$ so $\max\set{\norm{x},\norm{x-y}}=\norm{x-y}$ actually. Together with the equation above, we get that $\norm{x-y}=\norm{y}$.

Remark: The name "Isosceles Triangle Principle" comes from the geometric interpretation: If $x$ and $y$ are two points of a triangle which has the origin $0$ as it's third point, the sides have the length $x$, $y$ and $x-y$. The principle then says that the third side is as long as the longer one of the others, so every "triangle" is isosceles with respect to a non-Archimedean norm.

Lemma: Let $\norm{\cdot}: \mathbb{F}\rightarrow\mathbb{R}_{\geq 0}$ be a non-Archimedean norm and let $\forall\ i\in \set{0, \ldots, n}: a_i\in\mathbb{F}$. If $\forall\ 0<i\leq n: \norm{a_0}>\norm{a_i}$ it holds that $\norm{\sum_{i=0}^n a_i}=\norm{a_0}$.

Proof: We prove this statement by induction. For $n=0$, the statement is trivial. Now assume that the statement is proven for $n$ and let $a_{n+1}\in\mathbb{F}$ with $\norm{a_0}>\norm{a_{n+1}}$. Now set $x\df a_{n+1}$ and $y=\sum_{i=0}^n a_i$ and note that: $$ \norm{y}\jeqtxt{I.H.}\norm{a_0}>\norm{a_{n+1}}=\norm{x} $$ and we can apply the Isosceles Triangle Principle (ITP): $$ \norm{\sum_{i=0}^{n+1} a_i}=\norm{\sum_{i=0}^{n} a_i + a_{n+1}} = \norm{x+y}\jeqtxt{ITP}\norm{y}=\norm{\sum_{i=0}^n a_i}\jeqtxt{I.H.}\norm{a_0}.$$

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.