Covariants and simultaneous diagonalization of pairs of ternary quadratic forms, and binary quartic forms
Stanley Yao Xiao · 2025
Read Covariants and simultaneous diagonalization of pairs of ternary quadratic forms, and binary quartic forms by Stanley Yao Xiao (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-03848.txt.
Introduction
Stanley Yao Xiao proves a correspondence between two families of polynomial forms in a 2025 paper. One side is a quartic form in two variables, carrying a companion polynomial of degree six. The other side is a pair of quadratic forms in three variables, with a companion of degree three. The companions match under a precise dictionary between the settings. The paper also gives a canonical way to diagonalize pairs of quadratic forms. A corollary tests when rational forms can be diagonalized rationally.
Catalog record
| CREATOR | Stanley Yao Xiao |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-03848
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Covariants and simultaneous diagonalization of pairs of ternary quadratic forms, and binary quartic forms Thanks: The author is supported by NSERC Discovery Grant RGPIN-2024-06810. Stanley Yao Xiao Abstract . In this paper we prove a correspondence between a canonical degree six covariant of binary quartic forms F F and a cubic covariant of a pair of ternary quadratic forms ( f A , f B ) (f_{A},f_{B}) . In the process we obtain a canonical way to diagonalize a pair of n n -ary quadratic forms over any field K K of characteristic zero. As a corollary, we give a precise criterion to decide whether a pair of n n -ary quadratic forms over ℚ {\mathbb{Q}} is diagonalizable over ℚ {\mathbb{Q}} . 1. Introduction Perhaps one of the most well-known stories in number theory in the past two decades is the one told primarily by Manjul Bhargava. Among Bhargava’s achievements is the parametrization [ 4 ] and enumeration [ 5 ] of quartic rings and fields (ordered by absolute discriminant). Here Bhargava’s key insight is that integral orbits of pairs of integral ternary quadratic forms correspond to quartic rings, as well as an identified cubic resolvent ring. In subsequent work with Arul Shankar [ 6 ] , they used a related parametrization of 2 -Selmer elements of elliptic curves by \operatorname{GL}_{2}({\mathbb{Z}}) -equivalence classes of integral binary quartic forms to show that the average rank of elliptic curves in short Weierstrass model ordered by naive height is at most 3/2 , a stunning achievement that marked an epochal shift. One of the key ingredients of Bhargava and Shankar’s paper [ 6 ] is the following embedding result. Let (1.1) V_{4}({\mathbb{Z}})=\{a_{4}x^{4}+a_{3}x^{3}y+a_{2}x^{2}y^{2}+a_{1}xy^{3}+a_{0}y^{4}:a_{i}\in{\mathbb{Z}},i=0,1,2,3,4\} be the lattice of integral binary quartic forms and (1.2) W_{4}({\mathbb{Z}})=\left\{\left(\begin{bmatrix}a_{11}&a_{12}/2&a_{13}/2\\ a_{12}/2&a_{22}&a_{23}/2\\ a_{13}/2&a_{23}/2&a_{33}\end{bmatrix},\begin{bmatrix}b_{11}&b_{12}/2&b_{13}/2\\ b_{12}/2&b_{22}&b_{23}/2\\ b_{13}/2&b_{23}/2&b_{33}\end{bmatrix}\right):a_{ij},b_{ij}\in{\mathbb{Z}},1\leq i,j\leq 3\right\} be the lattice corresponding to Gram matrices of pairs of ternary quadratic forms with integer coefficients. We also note the notation W_{4}({\mathbb{Z}})=({\mathbb{Z}}^{2}\otimes\operatorname{Sym}^{2}{\mathbb{Z}}^{3})^{\ast} in [ 5 ] . Then M. Matchett Wood proved in [ 16 ] that there is a canonical embedding \phi:V_{4}({\mathbb{Z}})\hookrightarrow W_{4}({\mathbb{Z}}) given by (1.3) \phi:F(x,y)=a_{4}x^{4}+a_{3}x^{3}y+a_{2}x^{2}y^{2}+a_{1}xy^{3}+a_{0}y^{4} \mapsto\left(\begin{bmatrix}0&0&1/2\\ 0&-1&0\\ 1/2&0&0\end{bmatrix},\begin{bmatrix}a_{4}&a_{3}/2&0\\ a_{3}/2&a_{2}&a_{1}/2\\ 0&a_{1}/2&a_{0}\end{bmatrix}\right)=(A_{0},B_{F}). This map is canonical in the sense that the action of \operatorname{PGL}_{2}({\mathbb{Z}}) on V_{4}({\mathbb{Z}}) is realized in the group \operatorname{GL}_{2}({\mathbb{Z}})\times\operatorname{SL}_{3}({\mathbb{Z}}) by the mapping (1.4) \rho:\operatorname{PGL}_{2}({\mathbb{Z}})\rightarrow\operatorname{GL}_{2}({\mathbb{Z}})\times\operatorname{SL}_{3}({\mathbb{Z}}) \begin{bmatrix}a_{11}&a_{12}\\ a_{21}&a_{22}\end{bmatrix}\mapsto\frac{1}{a_{11}a_{22}-a_{12}a_{21}}\begin{bmatrix}a_{22}^{2}&a_{21}a_{22}&a_{22}^{2}\\ 2a_{12}a_{22}&a_{11}a_{22}+a_{12}a_{21}&2a_{11}a_{21}\\ a_{12}^{2}&a_{11}a_{12}&a_{11}^{2}\end{bmatrix}. (see Equation (29) in [ 6 ] ). Importantly, the map \phi in ( 1.3 ) is discriminant preserving . This is crucial, because the prehomogeneous vector space F^{2}\otimes\operatorname{Sym}^{2}F^{3} with the action of \operatorname{GL}_{2}\times\operatorname{GL}_{3} has just one polynomial discriminant, namely the discriminant. This crucial property was needed by Bhargava and Shankar to carry out their proof in [ 6 ] . To relate the rational space of binary quartic forms V_{4}({\mathbb{Q}}) to 2 -Selmer elements of elliptic curves over {\mathbb{Q}} , Bhargava and Shankar had consulted the important work of Cremona in [ 9 ] . There Cremona studied the theory of invariants and covariants of binary cubic and quartic forms extensively, with the motivation of carrying out reduction for such forms. It turns out that reduction theory of binary forms is important for solving Thue equations (see [ 13 ] for an authoritative modern reference). Motivated by this relation, I carried out an extensive study of the relation of invariants and covariants of binary cubic and quartic forms with their \operatorname{GL}_{2} -automorphism groups in [ 18 ] . This study eventually led to the joint work [ 14 ] with C. L. Stewart. Indeed, the problem we studied in [ 14 ] was partially motivated by extending the work of C. Skinner and T. D. Wooley on sums of two k -th powers [ 12 ] and subsequent work of Bennett, Dummigan, and Wooley [ 1 ] . In [ 6 ] there does not seem to be a necessity to rely on the structure of covariants of binary quartic forms nor pairs of ternary quadratics for their argument. Indeed, the theory of \operatorname{GL}_{3} -covariants of pairs of quadratic forms seemed rather mysterious. A common occurrence in the subject of arithmetic statistics is that a brilliant parametrization or identity is thought to be recently discovered, only for the community to realize that it had been known to some old masters in days of yore. This is such an occurrence, as we note that the structure of covariants of pairs of n -ary quadratic forms had been discussed nearly a century ago by J. Williamson in [ 15 ] . Our goal in this paper is to show that the embedding ( 1.3 ) not only gives a discriminant preserving map, which establishes a connection between the rings of polynomial invariants of V_{4}({\mathbb{Z}}) and W_{4}({\mathbb{Z}}) respectively, but also gives a connection to covariants . The ring of covariants of V_{4} under \operatorname{GL}_{2} -action is generated by F , the original form, H_{F} , the Hessian covariant, and F_{6}(x,y)=\frac{1}{36}\begin{vmatrix}\dfrac{\partial F}{\partial x}&&\dfrac{\partial F}{\partial y}\\ \\ \dfrac{\partial H_{F}}{\partial x}&&\dfrac{\partial H_{F}}{\partial y}\end{vmatrix}, which as far as I know is not given any particular name. We will simply refer to it as the “ F_{6} -covariant". The triple (F,H_{F},F_{6}) are not independent but satisfies a syzygy; see [ 9 ] . On the other hand, as noted in [ 15 ] the ring of polynomial covariants of W_{n} , or pairs of n -ary quadratic forms, under \operatorname{GL}_{n} -action is generated by n quadratic covariants (including the two forms in the pair) as well as the Jacobian determinant of the n -quadratic forms. For n=3 , the three quadratic covariants can be taken to be (1.5) f_{A}({\mathbf{x}})={\mathbf{x}}^{t}A{\mathbf{x}},\quad f_{B}({\mathbf{x}})={\mathbf{x}}^{t}B{\mathbf{x}},\quad\text{and}\quad g_{B}({\mathbf{x}})={\mathbf{x}}^{t}(AB^{\dagger}A){\mathbf{x}} where M^{\dagger}=\operatorname{Adj}(M) is the adjugate matrix of M . The Jacobian determinant of the three forms is denoted by C_{3}({\mathbf{x}}) , or the cubicovariant of (A,B) . 1 1 1 “While normally deprecating the use of portmanteau or blended words, we have the authority of Cayley, Salmon, and others for this nomenclature” - C. Hooley, [ 10 ] Our main result is the following observation: Theorem 1.1 . Let (1.6) F(x,y)=a_{4}x^{4}+a_{3}x^{3}y+a_{2}x^{2}y^{2}+a_{1}xy^{3}+a_{0}y^{4} be a binary quartic form defined over a field K and let (A_{0},B_{F})=\phi(F) , with \phi given in ( 1.3 ). Let F_{6} be the F_{6} -covariant of F and let C_{3} be the cubicovariant of the pair (A_{0},B_{F})\in W_{4}(F) . Then (1.7) F_{6}(x,y)=C_{3}(x^{2},xy,y^{2}) holds identically. A related result, which relates the Hessian covariant F_{4} of F with a linear combination of quadratic covariants of (A_{0},B_{F}) , is the following. For (A,B)\in W_{4} put g_{A}({\mathbf{x}})={\mathbf{x}}^{t}(BA^{\dagger}B){\mathbf{x}}. We then have: Theorem 1.2 . Let F(x,y)=a_{4}x^{4}+a_{3}x^{3}y+a_{2}x^{2}y^{2}+a_{1}xy^{3}+a_{0}y^{4} be a binary quartic form defined over a field K and let (A_{0},B_{F})=\phi(F) , with \phi given in ( 1.3 ). Let F_{4} be the Hessian covariant of F . We then have (1.8) F_{4}(x,y)=6g_{B_{F}}(x^{2},xy,y^{2})+3g_{A_{0}}(x^{2},xy,y^{2}). The cubicovariant C_{3} of a pair (A,B)\in W_{4} is known to be a decomposable form . That is, C_{3} splits into a product of linear forms in an algebraic closure of its field of definition. Despite being a “folklore" result, it does not appear to be prominently featured in the existing feature (to the best of my knowledge). A particular cubic decomposable form (actually, a norm form ) appeared in my paper [ 18 ] , namely the form (1.9) {\mathcal{G}}_{c_{1},c_{2}}(u,s,t)=u^{3}-3c(s^{2}-st+t^{2})u+c((2c_{1}-c_{2})s^{3}-3(c_{1}+c_{2})s^{2}t+3(2c_{2}-c_{1})st^{2}+(2c_{1}-c_{2})t^{3}) with c=c_{1}^{2}-c_{1}c_{2}+c_{2}^{2} and c_{1},c_{2}\in{\mathbb{Z}} . We gave a