pith. sign in

arxiv: 2408.04894 · v2 · submitted 2024-08-09 · 🧮 math.FA · math-ph· math.MP· math.SG

On generalization of Williamson's theorem to real symmetric matrices

Hemant K. Mishra This is my paper

Pith reviewed 2026-05-23 22:29 UTC · model grok-4.3

classification 🧮 math.FA math-phmath.MPmath.SG
keywords Williamson's theoremsymplectic eigenvaluesreal symmetric matricessymplectic matricesgeneralizationperturbation boundsEigSpSm(2n)
0
0 comments X

The pith

Williamson's theorem extends to every real symmetric matrix by allowing any real numbers as symplectic eigenvalues.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that for any 2n by 2n real symmetric matrix A, there always exists a real symplectic matrix M such that M transpose A M equals D direct sum D, with D an n by n diagonal matrix whose entries can be any real numbers. This removes the positivity restriction from the original theorem and from its known extension to certain semidefinite cases. The work supplies explicit formulas for the resulting symplectic eigenvalues, constructs the required M matrices, and derives perturbation bounds that apply to the subclass of positive semidefinite matrices whose kernels are symplectic subspaces. A reader would care because the result unifies the treatment of all quadratic forms under symplectic transformations rather than restricting to the positive definite setting.

Core claim

Williamson's theorem states that if A is a 2n×2n real symmetric positive definite matrix then there exists a 2n×2n real symplectic matrix M such that M^T A M = D ⊕ D, where D is an n×n diagonal matrix with positive diagonal entries known as the symplectic eigenvalues of A. The theorem is known to be generalized to 2n×2n real symmetric positive semidefinite matrices whose kernels are symplectic subspaces of R^{2n}, in which case some of the diagonal entries of D are allowed to be zero. In this paper we further generalize Williamson's theorem to 2n×2n real symmetric matrices by allowing the diagonal elements of D to be any real numbers, and thus extending the notion of symplectic eigenvalues.

What carries the argument

The real symplectic matrix M that satisfies M^T A M = D ⊕ D for arbitrary real symmetric A, carrying the decomposition that defines the symplectic eigenvalues.

If this is right

  • Symplectic eigenvalues are now defined for every real symmetric matrix and can take negative values.
  • Explicit constructions of the transforming symplectic matrices are available for any such A.
  • Perturbation bounds on the symplectic eigenvalues hold for the class of positive semidefinite matrices with symplectic kernels.
  • The decomposition unifies the positive definite, semidefinite, and indefinite cases under a single statement.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The result may simplify normal-form analysis for Hamiltonian systems whose quadratic part is indefinite.
  • It suggests that symplectic eigenvalues could serve as invariants for quadratic forms on symplectic vector spaces without a positivity assumption.
  • The perturbation bounds might be testable by direct computation on random symmetric matrices that include negative eigenvalues.

Load-bearing premise

A real symplectic matrix M always exists that achieves the stated block-diagonal form even when the original matrix has negative eigenvalues.

What would settle it

A concrete 2n by 2n real symmetric matrix A for which no real symplectic M satisfies M^T A M = D ⊕ D with D diagonal.

read the original abstract

Williamson's theorem states that if $A$ is a $2n \times 2n$ real symmetric positive definite matrix then there exists a $2n \times 2n$ real symplectic matrix $M$ such that $M^T A M=D \oplus D$, where $D$ is an $n \times n$ diagonal matrix with positive diagonal entries known as the symplectic eigenvalues of $A$. The theorem is known to be generalized to $2n \times 2n$ real symmetric positive semidefinite matrices whose kernels are symplectic subspaces of $\mathbb{R}^{2n}$, in which case, some of the diagonal entries of $D$ are allowed to be zero. In this paper, we further generalize Williamson's theorem to $2n \times 2n$ real symmetric matrices by allowing the diagonal elements of $D$ to be any real numbers, and thus extending the notion of symplectic eigenvalues to real symmetric matrices. Also, we provide an explicit description of symplectic eigenvalues, construct symplectic matrices achieving Williamson's theorem type decomposition, and establish perturbation bounds on symplectic eigenvalues for a class of $2n \times 2n$ real symmetric matrices denoted by $\operatorname{EigSpSm}(2n)$. The set $\operatorname{EigSpSm}(2n)$ contains $2n \times 2n$ real symmetric positive semidefinite whose kernels are symplectic subspaces of $\mathbb{R}^{2n}$. Our perturbation bounds on symplectic eigenvalues for $\operatorname{EigSpSm}(2n)$ generalize known perturbation bounds on symplectic eigenvalues of positive definite matrices given by Bhatia and Jain \textit{[J. Math. Phys. 56, 112201 (2015)]}.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper claims to generalize Williamson's theorem from positive (semi)definite 2n×2n real symmetric matrices to arbitrary real symmetric matrices by allowing the diagonal entries of D in the decomposition M^T A M = D ⊕ D (M symplectic) to be any real numbers, thereby extending the notion of symplectic eigenvalues. It provides explicit constructions of the symplectic matrices M and the diagonal D, and establishes perturbation bounds on these generalized symplectic eigenvalues for the subclass EigSpSm(2n) of positive semidefinite matrices whose kernels are symplectic subspaces, extending Bhatia-Jain bounds.

Significance. If the central existence claim held, the work would extend symplectic spectral theory to indefinite matrices with potential applications in Hamiltonian systems and optimization. The perturbation results for EigSpSm(2n) would be a modest incremental contribution generalizing known bounds. However, the existence statement for general symmetric matrices is false, rendering the claimed generalization invalid and limiting the paper's significance to at most a partial treatment of the positive semidefinite case already covered in the literature.

major comments (2)
  1. [Abstract] Abstract: The stated generalization asserts that for every 2n×2n real symmetric matrix A there exists a symplectic M such that M^T A M = D ⊕ D with D diagonal and arbitrary real entries. This is incompatible with Sylvester's law of inertia: any congruence (including symplectic) preserves inertia, but D ⊕ D has even positive and negative inertia indices (each d_i appears with multiplicity two), while arbitrary symmetric matrices need not (e.g., the 2×2 matrix diag(1,-1) has inertia (1,1)). The restriction of perturbation results to EigSpSm(2n) does not salvage the general existence claim.
  2. [Introduction / main theorem] Introduction / main theorem statement (presumed §2 or §3): The explicit constructions of M and the generalized symplectic eigenvalues for matrices with negative eigenvalues are presented as achieving the decomposition for all real symmetric A, but no argument is given that addresses or circumvents the inertia obstruction; the constructions therefore cannot be valid in general.
minor comments (1)
  1. [Abstract] Notation for EigSpSm(2n) is introduced without a precise set-theoretic definition in the abstract; a formal definition should appear early.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We appreciate the referee's detailed review and the identification of the fundamental obstruction from Sylvester's law of inertia. We acknowledge that our claimed generalization to arbitrary real symmetric matrices is not correct, as the decomposition M^T A M = D ⊕ D with symplectic M preserves the inertia, and D ⊕ D always has even inertia indices. We will revise the manuscript accordingly to limit the scope to matrices where this is possible, and focus on the perturbation results for EigSpSm(2n).

read point-by-point responses

  1. Referee: [Abstract] Abstract: The stated generalization asserts that for every 2n×2n real symmetric matrix A there exists a symplectic M such that M^T A M = D ⊕ D with D diagonal and arbitrary real entries. This is incompatible with Sylvester's law of inertia: any congruence (including symplectic) preserves inertia, but D ⊕ D has even positive and negative inertia indices (each d_i appears with multiplicity two), while arbitrary symmetric matrices need not (e.g., the 2×2 matrix diag(1,-1) has inertia (1,1)). The restriction of perturbation results to EigSpSm(2n) does not salvage the general existence claim.

    Authors: We fully agree with this observation. The general existence claim as stated in the abstract cannot hold due to the inertia preservation under symplectic congruence. The example provided is a valid counterexample. We will revise the abstract to remove the claim of generalization to arbitrary real symmetric matrices and instead specify that the decomposition holds when the inertia indices of A are even. This addresses the incompatibility. revision: yes

  2. Referee: [Introduction / main theorem] Introduction / main theorem statement (presumed §2 or §3): The explicit constructions of M and the generalized symplectic eigenvalues for matrices with negative eigenvalues are presented as achieving the decomposition for all real symmetric A, but no argument is given that addresses or circumvents the inertia obstruction; the constructions therefore cannot be valid in general.

    Authors: We concur that the constructions cannot be valid for all real symmetric matrices, as they do not address the inertia obstruction. The paper incorrectly presented the result as holding for arbitrary A. We will revise the introduction and main theorem statements to clarify the necessary condition on the inertia and restrict the explicit constructions to the appropriate class of matrices. The perturbation bounds for EigSpSm(2n) remain valid as they pertain to positive semidefinite matrices. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation extends prior results independently

full rationale

The paper states a direct generalization of Williamson's theorem to arbitrary real symmetric matrices by permitting any real values on the diagonal of D, while restricting its perturbation analysis to the already-known class EigSpSm(2n) of PSD matrices with symplectic kernels. It cites Bhatia and Jain only for the positive-definite perturbation bounds being extended, with no self-citation load-bearing on the existence claim, no fitted parameters renamed as predictions, and no ansatz or uniqueness result imported from the authors' own prior work. The derivation chain therefore remains self-contained against external benchmarks and does not reduce any claimed result to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper extends existing theorems using standard facts from linear algebra and symplectic geometry without introducing new free parameters or invented entities.

axioms (1)
  • standard math Standard properties of the real symplectic group Sp(2n,R) and its compatibility with symmetric bilinear forms
    The decomposition relies on well-established facts about symplectic matrices and real symmetric matrices.

pith-pipeline@v0.9.0 · 5844 in / 1171 out tokens · 29146 ms · 2026-05-23T22:29:50.739670+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages · 8 internal anchors

  1. [1]

    Gromov, Pseudo holomorphic curves in symplectic manifolds , Invent

    Mikhael Gromov. Pseudo holomorphic curves in symplecti c manifolds. Inventiones mathematicae, 82(2): 307–347, June 1985. http://dx.doi.org/10.1007/BF01388806

    work page doi:10.1007/bf01388806 1985

  2. [2]

    Symplectic invariants, entropic measures and correlations of Gaussian states

    Alessio Serafini, Fabrizio Illuminati, and Silvio De Si ena. Symplectic invariants, entropic measures and correlations of gaussian states. Journal of Physics B: Atomic, Molecular and Optical Physics , 37(2):L21, December 2003. arXiv:quant-ph/0307073

    work page internal anchor Pith review Pith/arXiv arXiv 2003

  3. [3]

    Pereira, Leonardo Banchi, and Stefano Pirandol a

    Jason L. Pereira, Leonardo Banchi, and Stefano Pirandol a. Symplectic decomposition from submatrix determinants. Proceedings of the Royal Society A, 477(2255):20210513, November 2021.arXiv:2108.05364

    work page arXiv 2021

  4. [4]

    F. Nicacio. Williamson theorem in classical, quantum, a nd statistical physics. American Journal of Physics, 89(12):1139–1151, December 2021. http://dx.doi.org/10.1119/10.0005944. arXiv:2106.11965

    work page doi:10.1119/10.0005944 2021

  5. [5]

    Quantum parameter estimation using multi-mode Gaussian states

    D. Šafránek, A. R. Lee, and I. Fuentes. Quantum parameter estimation using multi-mode Gaussian states. New Journal of Physics, 17(7):073016, 2015. arXiv:1502.07924

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  6. [6]

    On symplectic eigenvalues of positive definite matrices

    Rajendra Bhatia and Tanvi Jain. On symplectic eigenvalu es of positive definite matrices. Jour- nal of Mathematical Physics , 56(11):112201, November 2015. http://dx.doi.org/10.1063/1.4935852. arXiv:1803.04647

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1063/1.4935852 2015

  7. [7]

    Log-majorizations for the (symplectic) eigenvalues of the Cartan barycenter

    Fumio Hiai and Yongdo Lim. Log-majorizations for the (sy mplectic) eigenvalues of the Cartan barycenter. Linear Algebra and its Applications , 553:129–144, September 2018. http://dx.doi.org/10.1016/j.laa.2018.04.029. arXiv:1710.00494

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1016/j.laa.2018.04.029 2018

  8. [8]

    Hemant K. Mishra. First order sensitivity analysis of sy mplectic eigenvalues. Linear Algebra and its Applications , 604:324–345, November 2020. http://dx.doi.org/10.1016/j.laa.2020.07.003. arXiv:2007.10572

    work page doi:10.1016/j.laa.2020.07.003 2020

  9. [9]

    A Schur–Horn theorem for symplectic eigenvalues

    Rajendra Bhatia and Tanvi Jain. A Schur–Horn theorem for symplectic eigenvalues. Linear Alge- bra and its Applications , 599:133–139, August 2020. http://dx.doi.org/10.1016/j.laa.2020.04.005. arXiv:2004.03906

    work page doi:10.1016/j.laa.2020.04.005 2020

  10. [10]

    Variational principles for symplectic eigenvalues.Canadian Mathematical Bulletin, 64(3):553–559, September 2021

    Rajendra Bhatia and Tanvi Jain. Variational principles for symplectic eigenvalues.Canadian Mathematical Bulletin, 64(3):553–559, September 2021. http://dx.doi.org/10.4153/S0008439520000648

    work page doi:10.4153/s0008439520000648 2021

  11. [11]

    Sums and products of symplectic eigenvalue s

    Tanvi Jain. Sums and products of symplectic eigenvalue s. Linear Algebra and its Applications, 631:67–82, December 2021. http://dx.doi.org/10.1016/j.laa.2021.08.016. arXiv:2108.10741

    work page doi:10.1016/j.laa.2021.08.016 2021

  12. [12]

    Tanvi Jain and Hemant K. Mishra. Derivatives of symplec tic eigenvalues and a Lid- skii type theorem. Canadian Journal of Mathematics , 74(2):457–485, April 2022. http://dx.doi.org/10.4153/S0008414X2000084X. arXiv:2004.11024

    work page doi:10.4153/s0008414x2000084x 2022

  13. [13]

    The Horn cone associated with symp lectic eigenvalues

    Paul-Emile Paradan. The Horn cone associated with symp lectic eigenvalues. Comptes Rendus. Mathéma- tique, 360(G10):1163–1168, October 2022. http://dx.doi.org/10.5802/crmath.383. arXiv:2202.10260

    work page doi:10.5802/crmath.383 2022

  14. [14]

    Gajendra Babu and Hemant K. Mishra. Block perturbation of symplectic matrices in Williamson’s theorem. Canadian Mathematical Bulletin , 67(1):201–214, March 2024. http://dx.doi.org/10.4153/S0008439523000620. arXiv:2307.01078

    work page doi:10.4153/s0008439523000620 2024

  15. [15]

    Absil, Bin Gao, and Tatjana Styk el

    Nguyen Thanh Son, P.-A. Absil, Bin Gao, and Tatjana Styk el. Computing symplectic eigen- pairs of symmetric positive-definite matrices via trace mi nimization and Riemannian optimiza- tion. SIAM Journal on Matrix Analysis and Applications , 42(4):1732–1757, December 2021. http://dx.doi.org/10.1137/21M1390621. arXiv:2101.02618

    work page doi:10.1137/21m1390621 2021

  16. [16]

    A new version of Schur–Horn type theorem

    Shaowu Huang. A new version of Schur–Horn type theorem. Linear and Multilinear Algebra, 71(1):41–46, January 2023. http://dx.doi.org/10.1080/03081087.2021.2017834

    work page doi:10.1080/03081087.2021.2017834 2023

  17. [17]

    N. T. Son and T. Stykel. Symplectic eigenvalues of posit ive-semidefinite matrices and the trace minimization theorem. Electronic Journal of Linear Algebra , 38:607–616, September 2022. http://dx.doi.org/10.13001/ela.2022.7351. arXiv:2208.05291

    work page doi:10.13001/ela.2022.7351 2022

  18. [18]

    B. V. Rajarama Bhat and Tiju Cherian John. Real normal op erators and Williamson’s normal form. Acta Scientiarum Mathematicarum , 85:507–518, December 2019. http://dx.doi.org/10.14232/actasm-018-570-5 . arXiv:1804.03921

    work page doi:10.14232/actasm-018-570-5 2019

  19. [19]

    Tiju Cherian John, V. B. Kumar, and Anmary Tonny. An orde r relation between eigenvalues and symplectic eigenvalues of a class of infinite dimensional operators. June 2023. arXiv:2212.03900

    work page arXiv 2023

  20. [20]

    V. B. Kumar and Anmary Tonny. On approximating the sympl ectic spectrum of infinite-dimensional operators. Journal of Mathematical Physics , 65(4), April 2024. http://dx.doi.org/10.1063/5.0169600. arXiv:2303.09309. 2024/08/12 00:19 On generalization of Williamson’s theorem to real symmetri c matrices 21

    work page doi:10.1063/5.0169600 2024

  21. [21]

    Extremal entanglement and mixedness in continuous variable systems

    Gerardo Adesso, Alessio Serafini, and Fabrizio Illumi nati. Extremal entanglement and mixedness in continuous variable systems. Physical Review A , 70(2):022318, August 2004. http://dx.doi.org/10.1103/PhysRevA.70.022318. arXiv:quant-ph/0402124

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1103/physreva.70.022318 2004

  22. [22]

    Gaussian relative entropy of entanglement

    Xiao-yu Chen. Gaussian relative entropy of entangleme nt. Physical Review A, 71(6):062320, June 2005. http://dx.doi.org/10.1103/PhysRevA.71.062320. arXiv:quant-ph/0402109

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1103/physreva.71.062320 2005

  23. [23]

    Martin Idel, Sebastián Soto Gaona, and Michael M. Wolf. Perturbation bounds for Williamson’s symplectic normal form. Linear Algebra and its Applications , 525:45–58, July 2017. http://dx.doi.org/10.1016/j.laa.2017.03.013. arXiv:1609.01338

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1016/j.laa.2017.03.013 2017

  24. [24]

    Entangl ement dynamics of coupled quan- tum oscillators in independent nonMarkovian baths

    Jen-Tsung Hsiang, Onat Arısoy, and Bei-Lok Hu. Entangl ement dynamics of coupled quan- tum oscillators in independent nonMarkovian baths. Entropy, 24(12):1814, December 2022. http://dx.doi.org/10.3390/e24121814. arXiv:2211.07124

    work page doi:10.3390/e24121814 2022

  25. [25]

    On the algebraic problem concerning t he normal forms of linear dynamical systems

    John Williamson. On the algebraic problem concerning t he normal forms of linear dynamical systems. American Journal of Mathematics, 58(1):141–163, 1936. http://dx.doi.org/10.2307/2371062

    work page doi:10.2307/2371062 1936

  26. [26]

    Gerald B. Folland. Harmonic Analysis in Phase Space . Number 122. Princeton University Press, 1989. http://dx.doi.org/10.1515/9781400882427

    work page doi:10.1515/9781400882427 1989

  27. [27]

    Congruences and Canonical Forms for a Positive Matrix: Application to the Schweinler-Wigner Extremum Principle

    R. Simon, S. Chaturvedi, and V. Srinivasan. Congruence s and canonical forms for a positive matrix: Application to the Schweinler–Wigner extremum principle. Journal of Mathematical Physics, 40(7):3632– 3642, July 1999. arXiv:math-ph/9811003

    work page internal anchor Pith review Pith/arXiv arXiv 1999

  28. [28]

    Kh. D. Ikramov. On the symplectic eigenvalues of positi ve definite matrices. Moscow University Computational Mathematics and Cybernetics, 42:1–4, January 2018

    work page 2018

  29. [29]

    Hemant K. Mishra. Differential and subdifferential properties of symplecti c eigenvalues. PhD thesis, Indian Statistical Institute, New Delhi, 2021. URL http://hdl.handle.net/10263/7232

    work page 2021

  30. [30]

    R. J. de la Cruz and H. Faßbender. On the diagonalizabili ty of a matrix by a symplectic equivalence, similarity or congruence transformation. Linear Algebra Appl. , 496:288–306, 2016. http://dx.doi.org/10.1016/j.laa.2016.01.030

    work page doi:10.1016/j.laa.2016.01.030 2016

  31. [31]

    Matrix Analysis

    Rajendra Bhatia. Matrix Analysis. Springer, 1997. http://dx.doi.org/10.1007/978-1-4612-0653-8

    work page doi:10.1007/978-1-4612-0653-8 1997

  32. [33]

    and Johnson, Charles R

    Nathaniel Johnston. Advanced Linear and Matrix Algebra . Cambridge University Press, 2012. http://dx.doi.org/10.1017/CBO9781139020411

    work page doi:10.1017/cbo9781139020411 2012

  33. [34]

    de Gosson

    Maurice A. de Gosson. Symplectic Geometry and Quantum Mechanics , volume 166. Springer Science & Business Media, 2006. http://dx.doi.org/10.1007/3-7643-7575-2

    work page doi:10.1007/3-7643-7575-2 2006

  34. [35]

    Hemant K. Mishra. Equality in some symplectic eigenval ue inequalities. arXiv preprint, September 2023. arXiv:2309.04562. 2024/08/12 00:19

    work page arXiv 2023