Translations and extensions of the Nicomachean identity (Q6566541)
From MaRDI portal
| This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Translations and extensions of the Nicomachean identity |
scientific article; zbMATH DE number 7875570
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Translations and extensions of the Nicomachean identity |
scientific article; zbMATH DE number 7875570 |
Statements
Translations and extensions of the Nicomachean identity (English)
0 references
3 July 2024
0 references
For finite sequences of complex numbers \(\sigma=(a_1,a_2,\dots,a_n)\), a Nicomachean-type identity is defined by \N\[\N\nu(a_1,a_2,\dots,a_n):=\left(\sum_{i=1}^n a_i\right)^2-\sum_{i=1}^n a_i^3=0.\N\]\NThe ancient classical Nicomachean identity is \(\nu(1,2,\dots,n)=0\).\N\NThere are an infinity of sequences satisfying \(\nu=0\) that are generated by solutions to an infinite number of Pell's equations. Let \(R(u)[x]:=(\underbrace{x,\dots,x}_{u+1})\) be the operator that replaces any argument \(x\) by \(u+1\) copies of \(x\). Then\N\begin{multline*}\N\nu\bigl(\bigl(R(u)[1],R(u)[2],\dots,R(u)[n]\bigr)+t\bigr)\\\N=-\frac{n}{2}(u+1)(2 t+n+1)\left(t(t+1)-u\left(\frac{n(n+1)}{2}+n t\right)\right).\N\end{multline*}\NIts integral analogue is given by\N\begin{multline*}\N\left(\int_0^x(u+1)(z+t)d z\right)^2-(u+1)\int_0^x(z+t)^3 d z\\\N=-\frac{1}{2}(u+1)(2 t+x)\left(t^2-u\left(\frac{x^2}{2}+x t\right)\right).\N\end{multline*}\N\NThere are \(\sigma\) with \(\nu(\sigma)=0\) that have both arbitrarily small and arbitrarily large translations to other \(\nu=0\) sequences. A family of such sequences is given by \N\[\N\sigma=(1,2,2,3,4,5,7,8,8,10,10,12,12,\dots,2 s,2 s).\N\]\N\(\nu(\sigma+t)\) is translated into \(\nu(\sigma_m+t)\), where \N\[\N\sigma_m=(1,2,2,3,4,5,\dots,m-1,m,m+2). \N\]\NWhen \(t=0\), we have \(\nu(\sigma)=0\) and \(\nu(\sigma_m)=0\).\N\NConsider the right-most root \(t(r,m)\) of the quadratic polynomial\N\begin{multline*}\Nf_r(t)=t^2-\left(\prod_{i=1}^r(\alpha_1+1)-3\prod_{i=1}^r\left(\frac{\alpha_i}{2}+1\right)\right)t\\\N-2\prod_{i=1}^r(\alpha_i+1)\left(\frac{\alpha_i}{2}+1\right)+3\prod_{i=1}^r\frac{1}{6}(\alpha_i+2)(2\alpha_i+3).\N\end{multline*}\NThen, for \(r=2\) and \(\alpha_1=\alpha_2=m-1\), the sequence of second differences of \(t(2,m)\) converges to \((3+\sqrt{105})/12\) and the smaller root converges to the algebraic conjugate as \(m\to\infty\).\N\NBy translation of subsequences, we can find additional solutions to \(\nu(\sigma)=0\) with the elements of \(\sigma\) being positive rational numbers not all integers. Two examples are given involving arithmetic progressions modulo \(7\).\N\NBy adjoining nonzero real numbers \(x_1,x_2,x_3,\dots\) with \(x_j\ne -x_{j-1}\) to a finite sequence of real numbers \((a_1,a_2,\dots,a_n)\), each resulting sequence \(\nu=0\) is satisfied. A sequence with Fibonacci numbers satisfying \(\nu=0\) is also obtained.\N\NFor any positive integer \(N\), there are at least \(N\) distinct sequences \(\sigma\) with \(\nu(\sigma)=0\) such that each \(\nu(\sigma; a, b)\), where \((a,b)\) is adjoined to \(\sigma\), has the same quadratic polynomial factor. Then, connections with certain elliptic curves and their geometry are illustrated.
0 references
sum of cubes
0 references
variable parameter
0 references
translation parameter
0 references
Nicomachean identity
0 references
Pell's equation
0 references
sequential multiplication
0 references
finite difference
0 references
Fibonacci number
0 references
elliptic curve
0 references
0 references
0.7060778141021729
0 references
0.7054182291030884
0 references
0.6767232418060303
0 references