Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

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 have a question related to Lemma 4.2 from this pdf (which is, btw quite a nice exposition of Hoffman Singleton work on the classifications of Moore graphs of diameter 2 and 3.)

We are given a $n \times n$ real symmetric matrix $A$ with eigenvalues $d,\lambda_1,\lambda_2$ where $d$ has multiplicity 1. The key part of Lemma 4.2 computes the multiplicity of $\lambda_1$ and $\lambda_2$ when these two values are irrational.

The part I don't understand is when it tries to deduce that if $\lambda_1$ and $\lambda_2$ are irrational then they have both equal multiplicity $\frac{n-1}{2}.$ Somehow I don't understand the part when it says that if $(x-\lambda_1)$ divides $p(x)$ then $(x-\lambda_1)(x-\lambda_2)$ also divisides $p(x).$

How does that follows? Is anyone able to clarify the proof of the Lemma to me?

share|cite|improve this question
up vote 1 down vote accepted

As far as I can see, $A$ isn't an arbitrary real matrix, but rather the adjacency matrix of some graph, and thus contains only values $0$ and $1$. It follows that the characteristic polynomial $p$ is in $\mathbb{Z}[x]$.

The trick is then to realize that if the multiplicities of $\lambda_1,\lambda_2$ do not agree, then $p(x)$ cannot be a polynomial over $\mathbb{Z}[x]$. Their argument seems a bit strange to me, though - I don't believe that it's generally true that $(x-\lambda_1)(x-\lambda_2)$ is a polynomial over $\mathbb{Z}$, but rather over $\mathbb{Q}$. You'll thus need to multiply with some constant to get rid of all the fractions, but at that point, why would the result necessarily be a factor of $p(x) \in \mathbb{Z}[x]$?

I believe a more elementary argument should suffice, though. Simply watch what happens if you expand $(x-\lambda_1)^{n_1}(x-\lambda_2)^{n_2}$ where $\lambda_1=a+\sqrt{b}$,$\lambda_2=a-\sqrt{b}$. If $n_1 \neq n_2$, those squares won't cancel, and thus you won't get a polynomial over $\mathbb{Z}$.

Or you could argue (less elementary) that $\lambda_1,\lambda_2$ are algebraically indistinguishable over $\mathbb{Z}$, so if one of them is a zero of some polynomial, the other must be too.

share|cite|improve this answer
That's exactly what I needed. Thank you. – Jernej Oct 13 '12 at 19:44

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.