Sign up ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Consider a large, fixed $M>2$. For each $n$, let $\alpha_n$ denote the smallest algebraic integer of degree at most $n$, all of whose Galois conjugates lie in the real interval $(0,M)$.

Is there anything known on the rate at which $\alpha_n$ decays to $0$ as $n\to\infty$ ?

The exponential lower-bound $\alpha_n>\frac1 {M^{n}}$ is not hard to prove, but is it reasonable to expect that $\alpha_n$ will actually decay much more slowly, i.e. like an inverse power of $n$ ?

share|cite|improve this question
I don't think $\zeta_n$ is the best (or even second-best) notation in the world for a real algebraic integer... –  KConrad Apr 7 '14 at 15:24
Ah, $\alpha_n$ is definitely an improvement. –  KConrad Apr 7 '14 at 19:01

1 Answer 1

up vote 12 down vote accepted

$\alpha_n$ can be exponentially small once $M$ is large enough, say $M \geq 6$.

For $m > 0$ let $\tau_m$ be the (monic, degree-$m$) polynomial such that $\tau_m(z+z^{-1}) = z^m + z^{-m}$; in other words, $\tau_m(x) = 2 T_m(x/2)$ where $T_m$ is the $m$-th Čebyšev polynomial. Then $\tau_m$ has all its roots real and contained in the interval $(-2,2)$, while there are $m+1$ points $x_0,x_1,\ldots,x_m$ with $$ 2 = x_0 > x_1 > x_2 > \ldots > x_m = -2 $$ at which $\tau_m(x_k) = (-1)^k \cdot 2$.

Now fix an integer $M>5$, for $n$ odd let $m$ be the even number $n-1$ and consider $P(x) = x \, \tau_m (x-(M-2)) - 1$. This is a monic polynomial with $P(1)=-1$, $P(M-4)>0$, and $P$ changing sign between $x_{k-1}$ and $x_k$ for each $k=1,2,\ldots,m$. Therefore it has $m+1=n$ real roots, all in $(0,M)$, and all but one greater than $M-4$. Moreover the product of these roots is $-P(0) = 1$. Hence the remaining root is smaller than $(M-4)^{1-n}$, which decays exponentially once $M-4>1$. (In fact $M=5$ is good enough because about half the roots exceed $M-2$.)

For example, if $M=10$ and $n=7$ this recipe yields the septic polynomial $$ x^7 - 48x^6 + 954x^5 - 10048x^4 + 59145x^3 - 184464x^2 + 238142x - 1, $$ with one root just below $4.2 \cdot 10^{-6}$ and the other six ranging from about $6.061$ to $9.936$.

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.