Symmetric matrices defined by plane vector sequences

Mikiya Masuda · 2025

Read Symmetric matrices defined by plane vector sequences by Mikiya Masuda (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-2503-06836.txt.

Introduction

Mikiya Masuda studies a family of square tables of numbers in his 2025 math paper. Each table is built from a chain of arrows drawn in a plane, and the table is the same on both sides of its diagonal. He gives a formula that reads off a key trait of the table, called its signature, straight from the arrow chain.

When the table can be flipped, in the math sense of having an inverse, he links the table and its inverse through ideas from topology, the study of shape. He then builds a four dimensional shape from the same arrow chain. The inverse table turns out to count how the key parts of that shape cross each other. The paper ties algebra, geometry, and shape study into one neat package for a set of objects that show up in all three fields.

Catalog record

CREATORMikiya Masuda
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-06836

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Symmetric matrices defined by plane vector sequences Mikiya Masuda Abstract . Motivated by a work of Fu-So-Song, we associate a symmetric matrix A A to a plane vector sequence v v and give a formula to find the signature of A A in terms of the sequence v v . When A A is nonsingular, we interpret the relation between A A and A − 1 A^{-1} from a topological viewpoint. Finally, we associate an omnioriented quasitoric orbifold X X of real dimension four to the sequence v v and show that A − 1 A^{-1} is the intersection matrix of the characteristic suborbifolds of X X . 2000 Mathematics Subject Classification Primary 57S12, 14M25; Secondary 15B99 1. Introduction Let v=(v_{0},v_{1},\dots,v_{n},v_{n+1}) (n\geq 1) be a sequence of plane vectors where (1.1) \text{$v_{0}=\begin{pmatrix}0\\ 1\end{pmatrix}$, $v_{i}=\begin{pmatrix}a_{i}\\ b_{i}\end{pmatrix}\in\mathbb{R}^{2}$ for $1\leq i\leq n$, and $v_{n+1}=\begin{pmatrix}1\\ 0\end{pmatrix}$}. Motivated by Fu-So-Song [ 2 ] , we form a symmetric matrix A of order n with a_{i}b_{j} as (i,j) entry for i\leq j from the sequence v , that is (1.2) A=\begin{pmatrix}a_{1}b_{1}&a_{1}b_{2}&\dots&a_{1}b_{n-1}&a_{1}b_{n}\\ a_{1}b_{2}&a_{2}b_{2}&\dots&a_{2}b_{n-1}&a_{2}b_{n}\\ \vdots&\vdots&\ddots&\vdots&\vdots\\ a_{1}b_{n-1}&a_{2}b_{n-1}&\dots&a_{n-1}b_{n-1}&a_{n-1}b_{n}\\ a_{1}b_{n}&a_{2}b_{n}&\dots&a_{n-1}b_{n}&a_{n}b_{n}\end{pmatrix} One can easily find that (1.3) \det A=(-1)^{n+1}\prod_{i=0}^{n}\det(v_{i},v_{i+1})=a_{1}b_{n}\prod_{i=1}^{n-1}(a_{i+1}b_{i}-a_{i}b_{i+1}). Since A is symmetric, its eigenvalues are all real numbers and the signature of A , denoted by \Sign(A) , is defined to be the number of positive eigenvalues of A minus the number of negative eigenvalues of A . If \det(v_{i},v_{i+1})=0 for some i , then one can see that the symmetric matrix A^{\prime} defined by a vector sequence v^{\prime} with v_{i} or v_{i+1} removed from v has the same signature as A . Therefore, we assume (1.4) \text{$\det(v_{i},v_{i+1})\not=0$ for every $i=0,\dots,n$, i.e. $\det A\not=0$}. We understand that v_{i} rotates toward v_{i+1} with angle less than \pi for each i . The vectors v_{0},v_{1},\dots,v_{n},v_{n+1},v_{0} rotate around the origin of \mathbb{R}^{2} possibly many times and may go back and forth. We introduce two invariants of the sequence v=(v_{0},v_{1},\dots,v_{n},v_{n+1}) : (1.5) \begin{split}R(v)&=\text{ the rotation number of $v$ around the origin of $\mathbb{R}^{2}$},\\ S(v)&=1+\sum_{i=0}^{n}{\rm sgn}(\det(v_{i},v_{i+1}))\end{split} where {\rm sgn}(a) denotes 1 if a>0 and -1 if a<0 , and 1 at the right hand side above comes from {\rm sgn}(\det(v_{n+1},v_{0})) . Our first main result is the following. Theorem 1.1 . \Sign(A)=4R(v)-S(v) . Remark . For an almost complex compact manifold X of complex dimension two, it is well-known that T(X)=\frac{1}{4}(\Sign(X)+\chi(X)) where T(X) is the Todd genus of X , \Sign(X) is the signature of X , and \chi(X) is the Euler characteristic of X . Hence we have \Sign(X)=4T(X)-\chi(X) . Theorem 1.1 is a combinatorial analogue of this identity. Example 1.2 . (1) We take n=2 and v_{1}=\begin{pmatrix}-1\\ -1\end{pmatrix} , v_{2}=\begin{pmatrix}0\\ -1\end{pmatrix} . Then A=\begin{pmatrix}1&1\\ 1&0\end{pmatrix} (so \Sign(A)=0 ) while R(v)=1 and S(v)=4 . (2) We take n=2 and v_{1}=\begin{pmatrix}-1\\ -1\end{pmatrix} , v_{2}=\begin{pmatrix}-2\\ -1\end{pmatrix} . Then A=\begin{pmatrix}1&1\\ 1&2\end{pmatrix} (so \Sign(A)=2 ) while R(v)=1 and S(v)=2 . (3) We take n=4 and v_{1}=\begin{pmatrix}-1\\ -1\end{pmatrix},\ v_{2}=\begin{pmatrix}1\\ 0\end{pmatrix},\ v_{3}=\begin{pmatrix}0\\ 1\end{pmatrix},\ v_{4}=\begin{pmatrix}-1\\ -1\end{pmatrix} Then A=\begin{pmatrix}1&0&-1&1\\ 0&0&1&-1\\ -1&1&0&0\\ 1&-1&0&1\end{pmatrix} One can find \Sign(A)=2 while R(v)=2 and S(v)=6 . It follows from the definition of the signature, \Sign(A) must be n-2k for some 0\leq k\leq n . The following corollary shows that any such value can be realized as \Sign(A) . Corollary 1.3 . Suppose that all v_{i} ’s lie on the upper half plane of \mathbb{R}^{2} . If the sequence v=(v_{0},v_{1},\dots,v_{n+1}) rotates counterclockwise up to v_{k} for some 0\leq k\leq n and clockwise from v_{k} around the origin, then \Sign(A)=n-2k . In particular, A is positive definite ( i.e. \Sign(A)=n ) if v=(v_{0},v_{1},\dots,v_{n},v_{n+1}) rotates around the origin clockwise. Proof. In this case R(v)=0 and S(v)=-n+2k since \det(v_{i},v_{i+1})<0 if and only if k\leq i\leq n . ∎ A simple calculation shows that the inverse A^{-1} of our matrix A is a nonsingular tridiagonal symmetric matrix with entries in the superdiagonal (i.e. the diagonal just above the main diagonal) of A^{-1} being all nonzero (Proposition 3.1 ). Conversely, any such nonsingular tridiagonal symmetric matrix can be obtained as A^{-1} of our matrix A (Proposition 3.3 ). We observe the relation between A and A^{-1} from a topological viewpoint in Section 4 . This observation is made in [ 3 ] from a different perspective when each v_{i} is a primitive integer vector and the sequence v rotates around the origin once counterclockwise. To any integer vector sequence v satisfying the condition ( 1.4 ), one can associate an omnioriented quasitoric orbifold X of real dimension four. When \{v_{i},v_{i+1}\} forms a basis of \mathbb{Z}^{2} for any 0\leq i\leq n , X is smooth. For instance, the X associated to the integer vector sequence v in Example 1.2 is respectively \mathbb{C}P^{2}\#\overline{\mathbb{C}P^{2}} , \mathbb{C}P^{2}\#{\mathbb{C}P^{2}} , 3\mathbb{C}P^{2}\#\overline{\mathbb{C}P^{2}} , where \overline{\mathbb{C}P^{2}} denotes \mathbb{C}P^{2} with the opposite orientation. Note that \Sign(X)=\Sign(A) . In general, the characteristic suborbifolds X_{1},\dots,X_{n} of X corresponding to v_{1},\dots,v_{n} define a basis \alpha_{1},\dots,\alpha_{n} of H_{2}(X;\mathbb{R}) . We prove that the intersection matrix of X_{1},\dots,X_{n} is the inverse of A (Theorem 5.1 ). This result together with the observation made in Sections 3 and 4 implies the following. Theorem 1.4 . Let X be an omnioriented quasitoric orbifold associated to an integer vector sequence v satisfying ( 1.4 ) and let \alpha_{1}^{*},\dots,\alpha_{n}^{*} be the Kronecker dual to \alpha_{1},\dots,\alpha_{n} , i.e. \alpha_{i}^{*}\in H^{2}(X;\mathbb{R}) and \langle\alpha_{i}^{*},\alpha_{j}\rangle is the Kronecker delta \delta_{ij} . Then a_{i}b_{j}=\langle\alpha_{i}^{*}\cup\alpha_{j}^{*},[X]\rangle for i\leq j , i.e. A=(\langle\alpha_{i}^{*}\cup\alpha_{j}^{*},[X]\rangle) , where [X] denotes the fundamental class of X . This result is proved in [ 2 ] (by a different method) also when each v_{i} is a primitive integer vector and the sequence v rotates around the origin once counterclockwise (in this case X is a genuine toric orbifold). Acknowledgment. The author thanks Xin Fu, Tseleung So, and Jongbaek Song for helpful discussions and Hiraku Abe for informing him of the reference [ 5 ] . He is supported in part by the HSE University Basic Research Program, JSPS Grant-in-Aid for Scientific Research 22K03292, and Osaka Central Advanced Mathematical Institute (MEXT Joint Usage/Research Center on Mathematics and Theoretical Physics JPMXP0723833165). 2. Proof of Theorem 1.1 For a while, we allow v_{i} ’s (1\leq i\leq n) to be arbitrary plane (even zero) vectors. The following is ( 1.3 ) in the introduction. Lemma 2.1 . \det A=a_{1}b_{n}\prod_{i=1}^{n-1}(a_{i+1}b_{i}-a_{i}b_{i+1}) . Proof. We regard a_{i} ’s and b_{i} ’s as indeterminates and prove the lemma by induction on n . When n=1 , the lemma is obvious. Suppose that n\geq 2 and the lemma holds for n-1 . Adding the (n-1) st column multiplied by -b_{n}/b_{n-1} to the n -th column in ( 1.2 ), we obtain (2.1) \det A=\begin{vmatrix}a_{1}b_{1}&a_{1}b_{2}&\dots&a_{1}b_{n-1}&0\\ a_{1}b_{2}&a_{2}b_{2}&\dots&a_{2}b_{n-1}&0\\ \vdots&\vdots&\ddots&\vdots&\vdots\\ a_{1}b_{n-1}&a_{2}b_{n-1}&\dots&a_{n-1}b_{n-1}&0\\ a_{1}b_{n}&a_{2}b_{n}&\dots&a_{n-1}b_{n}&a_{n}b_{n}-a_{n-1}b_{n}\cdot\frac{b_{n}}{b_{n-1}}\end{vmatrix} Since the (n,n) entry above is (a_{n}b_{n-1}-a_{n-1}b_{n})b_{n}/b_{n-1} , it follows from the induction assumption that ( 2.1 ) turns into \begin{split}\det A&=a_{1}b_{n-1}\left(\prod_{i=1}^{n-2}(a_{i+1}b_{i}-a_{i}b_{i+1})\right)(a_{n}b_{n-1}-a_{n-1}b_{n})\frac{b_{n}}{b_{n-1}}\\ &=a_{1}b_{n}\prod_{i=1}^{n-1}(a_{i+1}b_{i}-a_{i}b_{i+1}),\end{split} proving the lemma. ∎ If v_{i} is zero for some 1\leq i\leq n , then entries in the i -th row and i -th column of A are all zero. Therefore, if A^{\prime} is A with i -th row and i -th column removed, then the set of eigenvalues of A^{\prime} agrees with the set of eigenvalues of A with one 0 removed. In particular, \Sign(A^{\prime})=\Sign(A) . The A^{\prime} is associated to the sequence v with v_{i} removed, so we may remove v_{i}

Read the complete text