Boston University Algebra Seminar

Boston University Algebra Seminar -- Spring 2005


Hyperelliptic curves, continued fractions, and Somos sequences


Alf van der Poorten
Centre for Number Theory Research and Brown University


Monday, March 21st at 4:15pm
111 Cummington Street, MCS B33


Abstract

In the early nineties Michael Somos asked for the inner meaning of the fact that the quadratic recursions

Ch-2Ch+2 =Ch-1Ch+1+ Ch2

and

Bh-2Bh+3 =Bh-1Bh+2+ BhBh+1

are respectively satisfied by the bi-directional sequences

...2,1,1,1,1,2,3,6,23,59,...
...1,1,1,1,2,3,5,11,37,83,...

surprisingly consisting entirely of integers (which incidentally, increase at a hefty clip: the log of their h-th term is O(h2) as h tends to infinity.

In the meantime I had found quite independent reason for studying the continued fraction expansion of square roots of polynomials. Nonetheless, I found that those expansions inter alia produce sequences of integers satisfying quadratic recursions of Somos type. For instance, the sequence (Ch) [called 4-Somos by some of its friends] reports the denominators of a sequence of points on the quartic curve Y2=(X2-3)2+4(X-2), equivalently of the points M+hS, with M=(-1,1), S=(0,0) on the cubic curve V2-V=U3+3U2+2U. Indeed all Somos 4 and Somos 5 sequences are `elliptic sequences' in such a sense.

I will tell the story from very first principles and will mention generalisations. Thus, again for instance, the sequence

...2,1,1,1,1,1,1,2,3,4,8,17,50...
given by the recursion
Bh-3Bh+3 =Bh-2Bh+2+ Bh2

arises from adding multiples of the class of the divisor at infinity on the Jacobian of the curve Y2=(X3-4X+1)2+4(X-2) of genus 2 to the class of the divisor given by [(φ,0),( φ ,0)]; here, no doubt to the joy of adherents to the cult of Fibonacci, φ denotes the golden ratio.