8
$\begingroup$

For fixed integer parameters $(P,Q)$, Lucas sequences represent a pair of complimentary integer sequences satisfying the same recurrence with the characteristic polynomial $f(x):=x^2 - Px + Q$. The two sequences can be defined via roots $x_{1,2}$ of $f(x)$ as $U_n:=\frac{x_1^n - x_2^n}{x_1-x_2}$ and $V_n:=x_1^n + x_2^n$.

Lucas sequences form a linear basis for the other sequences satisfying the same recurrence, and are tightly interconnected with numerous identities connecting arithmetic operations on terms to those on their indices.

Q: What would be a generalization of Lucas sequences to integer recurrences of higher order, in particular of order 3 with a characteristic polynomial $g(x):=x^3 - Px^2 + Qx - R$ with integer coefficients?

Generalized Lucas sequences should be integer, form a basis for the other sequences with the same characteristic polynomial, and have nice interconnecting relations (generalizing those for order 2).

A natural extension of the order 2 case seems to be $U_n := h_{n-1}$ and $V_n := p_n$ (ie. complete homogeneous and power sum symmetric polynomials) evaluated at the roots of $g(x)$. What other sequence(s) should we include into consideration and what would the corresponding relations?

$\endgroup$
2
  • 3
    $\begingroup$ Shannon and Horadam, fq.math.ca/Scanned/10-2/shannon.pdf suggest, for $S_n=PS_{n-1}+QS_{n-2}+RS_{n-3}$, the sequences that begin $0,1,P$; $1,0,Q$; and $0,0,R$. $\endgroup$ Commented May 7, 2023 at 0:10
  • 4
    $\begingroup$ @GerryMyerson: Thanks for the pointer, although I'm not that impressed by relations they develop (eg. they connect $J_n$ to $u_n$, which in general is not integer). I have a gut feeling that a proper generalization should be based on symmetric polynomials. $\endgroup$ Commented May 7, 2023 at 1:50

0

You must log in to answer this question.