Unitriangular factorization of holomorphic symplectic matrices
Gaofeng Huang, Frank Kutzschebauch, Phan Quoc Bao Tran · 2025
Read Unitriangular factorization of holomorphic symplectic matrices by Gaofeng Huang, Frank Kutzschebauch, Phan Quoc Bao Tran (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-2507-18963.txt.
Introduction
Gaofeng Huang, Frank Kutzschebauch, and Phan Quoc Bao Tran factor special matrices into simple ones in a 2025 paper. Their matrices are holomorphic and symplectic, meaning they vary smoothly with complex inputs while preserving a geometric pairing. The team proves every such matrix splits into a product of unitriangular factors, the matrix world's simplest building blocks, under a skew-diagonal version of the pairing.
Those factors then split further, needing at most seven factors under the standard pairing. That bound answers an open problem. Combining both steps estimates the count needed for the standard factorization, whose existence was known without numbers. A cleaner proof of the older result falls out as a bonus. The work sits in complex geometry with ties to Hamiltonian systems.
Catalog record
| CREATOR | Gaofeng Huang, Frank Kutzschebauch, Phan Quoc Bao Tran |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-18963
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Unitriangular factorization of holomorphic symplectic matrices Gaofeng Huang, Frank Kutzschebauch, Phan Quoc Bao Tran Abstract . We prove that every holomorphic symplectic matrix can be factorized as a product of holomorphic unitriangular matrices with respect to the symplectic form [ 0 L n − L n 0 ] \left[\begin{array}[]{ccc}0&L_{n}\\ -L_{n}&0\end{array}\right] where L L is the n × n n\times n matrix with 1 1 along the skew-diagonal. Also we prove that holomorphic unitriangular matrices with respect to this symplectic form are products of not more than 7 7 holomorphic unitriangular matrices with respect to the standard symplectic form [ 0 I n − I n 0 ] \left[\begin{array}[]{ccc}0&I_{n}\\ -I_{n}&0\end{array}\right] , thus solving an open problem posed in [ HKS22 ] . Combining these two results allows for estimates of the optimal number of factors in the factorization by holomorphic unitriangular matrices with respect to the standard symplectic form. The existence of that factorization was obtained earlier by Ivarsson-Kutzschebauch and Schott, however without any estimates. Another byproduct of our results is a new, much less technical and more elegant proof of this factorization. 1991 Mathematics Subject Classification 32Q56 (primary); 32Q28, 15A54, 32A17, 20H25 (secondary) 1. Introduction Factorization of a given matrix with elements in a field has been studied widely with many applications to different fields of sciences. The most well-known one is the factorization of a matrix A\in\mathrm{SL}_{n}(\mathbb{C}) into product of some elementary matrices of the form E+\alpha e_{ij} , i\neq j , called elementary matrices. This factorization is equivalent to factorization into so-called unitriangular matrices (triangular matrices with 1s on the diagonal). Another interesting class of matrices is the class of symplectic matrices. Generalizing the case of matrices with entries from a field, it is much more challenging to consider matrices with entries from a ring, the subject of K -theory. In this paper we always work with very specific rings, namely the ring of holomorphic functions on a Stein space or the ring of continuous functions on a normal finite dimensional topological space. Recall that the unitriangular symplectic matrices with respect to the standard symplectic form \displaystyle\Omega=\left[\begin{array}[]{ccc}0&I_{n}\\ -I_{n}&0\end{array}\right] are of the following form M^{-}(G)=\left[\begin{array}[]{ccc}I_{n}&0\\ G&I_{n}\end{array}\right] and M^{+}(G)=\left[\begin{array}[]{ccc}I_{n}&G\\ 0&I_{n}\end{array}\right] , where G is symmetric. Since the entries of G are holomorphic functions on a reduced Stein space, it can be understood as a holomorphic map: G:X\to\mathbb{C}^{\frac{n(n+1)}{2}} . In [ IKL22 ] (for n=2 ) and [ Scho24 ] (for general n ), the following theorem about factorizing matrices into product of unitriangular symplectic matrices has been proved: Theorem 1.1 . There exists a natural number K=K(n,d) such that given any finite dimensional reduced Stein space X of dimension d and any null-homotopic holomorphic map f:X\to\mathrm{Sp}_{2n}(\mathbb{C}) there exists a holomorphic map: G=(G_{1},\cdots,G_{K}):X\to(\mathbb{C}^{\frac{n(n+1)}{2}})^{K} such that f(x)=M^{-}(G_{1}(x))M^{+}(G_{2}(x))\cdots M^{\pm}(G_{K}(x)), where x\in X . It is immediate that any product of unitriangular symplectic matrices is connected by a path to the constant identity matrix I_{2n} , by multiplying the off-diagonal entries by t\in[0,1] , i.e. \left[\begin{array}[]{ccc}I_{n}&tG\\ 0&I_{n}\end{array}\right] and \left[\begin{array}[]{ccc}I_{n}&0\\ tG&I_{n}\end{array}\right] . Therefore the requirement of null-homotopy of the map f is necessary. In our paper, we study factorization of a holomorphic symplectic matrix into unitriangular symplectic factors with respect to the symplectic form \tilde{\Omega}=\left[\begin{array}[]{ccc}0&L_{n}\\ -L_{n}&0\end{array}\right] , where L is the n\times n matrix with 1 along the skew-diagonal. The lower unitriangular \tilde{\Omega} -symplectic factors are the \Omega -symplectic matrices of the form \left[\begin{array}[]{ccc}A^{-T}&0\\ B&A\end{array}\right], BA^{T} symmetric and A upper unitriangular, In order to introduce convenient coordinates (A,Z)\in\mathbb{C}^{\frac{n(n-1)}{2}}\times\mathbb{C}^{\frac{n(n+1)}{2}} on this set of matrices we will write them as M^{-}(A,Z)=\left[\begin{array}[]{ccc}I_{n}&0\\ Z&I_{n}\end{array}\right]\left[\begin{array}[]{ccc}A^{-T}&0\\ 0&A\end{array}\right]=\left[\begin{array}[]{ccc}A^{-T}&0\\ B&A\end{array}\right], Z=BA^{T} symmetric and A upper unitriangular. The same for upper unitriangular M^{+}(A,Z)=\left[\begin{array}[]{ccc}I_{n}&Z\\ 0&I_{n}\end{array}\right]\left[\begin{array}[]{ccc}A^{-1}&0\\ 0&A^{T}\end{array}\right]=\left[\begin{array}[]{ccc}A^{-1}&C\\ 0&A^{T}\end{array}\right], Z=CA^{-T} symmetric and A upper unitriangular. Throughout the paper, we call these matrices the elementary matrices . Our main theorem is a solution to the Gromov-Vaserstein problem for the symplectic form \tilde{\Omega} . Theorem 1.2 . There exists a natural number K=K(n,d) such that given any finite dimensional reduced Stein space X of dimension d and any null-homotopic holomorphic map f:X\to\mathrm{Sp}_{2n}(\mathbb{C}) there exist holomorphic maps: A=(A_{1},\cdots,A_{K}):X\to(\mathbb{C}^{\frac{n(n-1)}{2}})^{K} and Z=(Z_{1},\cdots,Z_{K}):X\to(\mathbb{C}^{\frac{n(n+1)}{2}})^{K} such that f(x)=M^{-}(A_{1}(x),Z_{1}(x))M^{+}(A_{2}(x),Z_{2}(x))\cdots M^{\pm}(A_{K}(x),Z_{K}(x)), where x\in X . This theorem together with our second main result, saying that the unitriangular \tilde{\Omega} -symplectic matrices from Theorem 1.2 are products of not more than 8 unitriangular \Omega -symplectic matrices, gives a new proof of Schott’s result. Moreover our proof is much more elegant and much easier to follow. In fact we use a novel strategy to prove the stratified ellipticity of a certain map, which is the core of the proof. More precisely, we can reduce the number of equations for the fibers of this map, given from the beginning by 2n polynomial equations, to one single equation. From this equation the stratified ellipticity is as easily obtained as in the work of Ivarsson and Kutzschebauch [ IK12 ] . In Schott’s proof (and in the preceding work of Ivarsson-Kutzschebauch-Løw) the reduction is only done from 2n to n equations. Studying fibers given by n polynomial equations is the hard and unpleasant technical part of the papers [ IKL22 ] , [ Scho24 ] . We denote by K_{symp}(n,d) the minimal number of unitriangular factors for 2n\times 2n \Omega -symplectic matrices with holomorphic functions as entries in any d -dimensional Stein space and \tilde{K}_{symp}(n,d) the same for \tilde{\Omega} -symplectic factors. Observe that the optimal number of factors over the field of complex numbers is classically known in the \tilde{\Omega} -case over fields (and rings of Bass stable rank 1) to be 4 , however, the optimal number for the \Omega -case has been only very recently found to be 5 by Jin, Tang and Zhu in [ JTZ20 ] . As such proofs over the complex numbers usually do, their proof depends on Jordan normal form. It is the fact that Jordan normal form is discontinuous making the problem for continuous and holomorphic matrices so delicate. However, we can reduce the problem to a simpler case by multiplying the unitriangular matrix A with a diagonal matrix \Lambda with pairwise different elements on the diagonal. This new matrix \Lambda A is diagonalizable and the diagonalization process is algebraic. From that point on we can apply the factorization technique in [ JTZ20 ] to get the optimal results as follows: Theorem 1.3 . For n\geq 4 we have 6\leq K_{symp}(n,d)\leq 7\tilde{K}_{symp}(n,d). For matrices of smaller sizes, there is a better upper bound (1) \displaystyle K_{symp}(n,d)\leq 4\tilde{K}_{symp}(n,d),\quad\text{ for }n=2\text{ and }n=3. This theorem solves the problem posed in [ HKS22 , p. 6] . Moreover, combining Theorem 1.3 with K -theoretic results from [ VSS11 ] , [ HKS22 ] we are in certain cases able to give explicit bounds for the number of factors in Schott’s theorem as follows: Proposition 1.4 . The number of factors for N\geq n+1 is bounded by the number for n : (2) \displaystyle K_{symp}(N,d)\leq 7\tilde{K}_{symp}(n,d). When the Stein space X is one-dimensional, (3) \displaystyle 5 \displaystyle\leq K_{symp}(n,1)\leq 16\ \ \text{for}\ \ n=2,3, (4) \displaystyle 6 \displaystyle\leq K_{symp}(n,1)\leq 28\ \ \text{for }\ n\geq 4. When the Stein space X is two-dimensional, (5) \displaystyle 5 \displaystyle\leq K_{symp}(n,2)\leq 20\ \ \text{for}\ \ n=2,3, (6) \displaystyle 6 \displaystyle\leq K_{symp}(n,2)\leq 35\ \ \text{for }\ n\geq 4. 1.1. Strategy of the proof We consider