Bi-Orthogonal Symmetry in Quantum & Math

• Bi-Orthogonal symmetry is defined by dual bases meeting strict reciprocal orthonormality, ensuring well-behaved spectral properties even in non-Hermitian contexts.
• It is applied in fields like quantum mechanics, signal processing, and coding theory to maintain real spectra and consistent operator dynamics.
• It underpins methods in functional analysis, multiwavelets, and tight-binding models, enabling effective transformations between continuous and discrete systems.

Bi-orthogonal symmetry refers to the structure, invariance, and transformation properties emerging from systems equipped with a pair of mutually dual bases—right and left eigenvectors, or more generally, dual function spaces—that satisfy strict bi-orthonormality relations. This symmetry generalizes conventional orthogonal (unitary or orthonormal) symmetry by replacing single self-dual bases with paired, mutually conjugate systems, fundamentally influencing spectral theory, quantum mechanics (especially in non-Hermitian and pseudo-Hermitian regimes), signal processing, and mathematical physics. Bi-orthogonal symmetry naturally arises in contexts where Hermiticity is relaxed but real spectra, completeness, and normalization are preserved, ensuring a well-defined, probability-consistent operator or polynomial calculus.

1. Core Definition and Mathematical Foundations

The defining feature of a bi-orthogonal system is the existence of two sets of vectors (or functions), $\{|\phi_j\rangle\}$ and $\{|\psi_i\rangle\}$, satisfying

$$\langle \psi_i | \phi_j \rangle = \delta_{ij}, \qquad i,j = 1, ..., N$$

with both sets spanning the relevant Hilbert (or vector) space. Unlike orthonormal systems (where typically $|\phi_i\rangle = |\psi_i\rangle$), bi-orthogonality decouples the right and left eigenvectors. For a non-Hermitian operator $T$ with real, non-degenerate spectrum,

$$T|\psi_n\rangle = \lambda_n|\psi_n\rangle,\qquad T^\dagger|\phi_n\rangle = \lambda_n|\phi_n\rangle,$$

and the bi-orthogonal resolution of the identity reads

$\sum_n |\psi_n\rangle\langle\phi_n| = I.$

This structure guarantees completeness and enables unique expansions and probabilistic interpretations even in the absence of Hermiticity (Brody, 2013).

In Fock space, or reproducing kernel Hilbert spaces (RKHS), the bi-orthogonal symmetry of basis functions allows kernel decompositions: $K(z, w) = \sum_n \psi_n(z)\, \overline{\phi_n(w)},$ with explicit orthonormality enforced via $\langle \psi_m | \phi_n \rangle = \delta_{mn}$ (Chakraborty, 2022).

2. Bi-Orthogonal Symmetry in Non-Hermitian Quantum Systems

Bi-orthogonal symmetry becomes essential in non-Hermitian, particularly pseudo-Hermitian or $\mathcal{PT}$-symmetric, quantum mechanics. Notably, the bi-orthogonal formalism enables preservation of reality of spectra, normalization of dynamics, and interpretation of observables through generalized Born rules (Brody, 2013, Chakraborty, 2022, Chakraborty, 2022). The time evolution under a non-Hermitian Hamiltonian $H$ with a complete bi-orthonormal spectrum is mediated by bi-unitary evolution: $$U(t) = e^{-iHt}, \quad \langle \tilde\psi(t) |\psi(t) \rangle = \text{const},$$ where $|\tilde\psi\rangle = \sum_n C_n |\phi_n\rangle$ relates the physical state to its dual.

Bi-orthogonal symmetry also underpins the theory of exceptional points, where eigenvalue coalescence signals the breakdown of diagonalization and the onset of Jordan block structure—accompanied by spontaneous symmetry breaking (e.g., in partial $\mathcal{PT}$-symmetric systems) (Chakraborty, 2022).

In models such as the partial $\mathcal{PT}$-symmetric quadratic boson Hamiltonian,

$$H(\lambda) = \omega a^\dagger a + i\lambda(a^2 + (a^\dagger)^2),$$

the spectrum remains real and the algebraic structure bi-orthogonally invariant as long as $|\lambda| < \lambda_\text{crit} = \omega/2$. At the exceptional point $|\lambda| = \lambda_\text{crit}$, the Bogoliubov frequency vanishes and the level spacings collapse (Chakraborty, 2022).

3. Operator Theory, Functional Analogs, and Sequences

The structure extends to functional analysis through biisometric pairs of operators $(V,W)$ on Hilbert space $H$ such that $V^* W = I_H$, yielding bi-orthogonal sequences $\{f_n = V^n w\}$ and $\{g_n = W^n v\}$ with $\langle f_n, g_m \rangle = \delta_{nm}$ (Kubrusly et al., 2019). This framework generalizes shift symmetry beyond orthonormal bases:

• Both $V$ and $W$ act as “half-shifts,”
• The friendlier mutual adjoint-inverse symmetry, $V^* W = W^* V = I_H$,
• Explicit realization via Laguerre polynomial-weighted functions in $L^2(0,\infty)$.

In mathematical physics, the Asymmetric Simple Exclusion Process (ASEP) and other integrable models benefit from bi-orthogonal polynomials, exhibiting a bi-orthogonal matrix structure in the moment algebra and boundary basis. Two interlocking families, $\{P_n\}$ and $\{Q_n\}$, are bi-orthogonal under the model’s linear functional, with associated operators yielding Askey–Wilson-type recurrences (Brak et al., 2019).

4. Bi-Orthogonal Symmetry in Signal Processing and Wavelets

In signal processing, bi-orthogonal symmetry is central to the design of bi-orthogonal multiwavelets. Given a pair of $r \times s$ Laurent-polynomial matrices $(P(z), \tilde{P}(z))$ with compatible symmetry and bi-orthogonality ($P(z)\tilde{P}^*(z)=I_r$), the extension problem seeks square matrices $(P_e(z), \tilde{P}_e(z))$ with the same symmetry and bi-orthogonality whose first $r$ rows recover the original matrices. The convolutional and symmetry-preserving algorithm furnishes explicit multiwavelet bases with prescribed reflection or antisymmetry properties (Zhuang, 2010). These symmetries control the construction of high-pass filters and ensure the desired even/odd structure of scaling functions and wavelets.

A summary of properties for such matrix systems:

Property	Requirement or Structure	Reference
Bi-orthogonality	$P(z)\tilde{P}^*(z) = I_r$	(Zhuang, 2010)
Compatible symmetry	Symmetry pattern $\epsilon,z^c$ matches	(Zhuang, 2010)
Extension algorithm	Cascade of $s \times s$ biorthogonal factors	(Zhuang, 2010)
Application	Symmetric/antisymmetric multiwavelets	(Zhuang, 2010)

5. Bi-Orthogonal Symmetry in Band Theory and Tight-Binding Models

For non-Hermitian periodic Hamiltonians, bi-orthogonal symmetry manifests in band structure analysis via bi-orthogonal Bloch and Wannier functions. Right and left Bloch waves, $\psi_{R/L,n,k}(x)$, are bi-orthonormal on the unit cell,

$$\int\,dx\, [\psi_{L,m,k}(x)]^*\,\psi_{R,n,k}(x) = \delta_{mn},$$

yielding bi-orthogonal Wannier functions for localized projections. The tight-binding Hamiltonian constructed from these bi-orthogonal bases reflects the underlying non-Hermitian physics and preserves the spectral properties found in the original continuum operator, provided the potential is sufficiently deep. Bi-orthogonal symmetry is thus critical for accurately mapping dispersion relations and observables between continuum and lattice descriptions of non-Hermitian systems (Mochizuki et al., 2022).

6. Bi-Orthogonal (Hyperoctahedral) Symmetry in Coding and Combinatorics

In coding theory, bi-orthogonal symmetry is exemplified by bi-orthogonal signal constellations, such as those formed by $\pm e_i$ vectors in $\mathbb{R}^M$: $$\mathcal{S}_{\rm bi} = \{ \pm e_1, \pm e_2, \dots, \pm e_M \},$$ invariant under the hyperoctahedral group of signed permutations. The Gram matrix displays a block structure linking each axis to its negative. This algebraic symmetry ensures equidistant signal pairs (except sign pairs), facilitating analytical computation of distance spectra and coding capacity. Crucially, careful labeling (e.g., Gray-like) and group invariance optimize bit-interleaved coded modulation (BICM) performance in power-limited regimes (Schenk et al., 2011).

7. Implications for Symmetry Principles, Transition Probabilities, and Classification

Generalized probabilistic theories relate bi-orthogonal symmetry to “bit symmetry”—the transitivity property on orthogonal pairs (2-frames) of pure states. When bit symmetry is postulated, it enforces a constant off-diagonal inner product for all orthogonal pairs, forcing a symmetric transition probability. Stronger symmetry postulates (frame-transitivity) further restrict admissible models to classical or simple Euclidean Jordan algebraic state spaces (Niestegge, 2024). This illustrates the profound link between structural symmetry in the dual basis—bi-orthogonality—and the deepest foundations of quantum theory, such as transition probability symmetry and spectral decomposability.

---

Bi-orthogonal symmetry, in its various guises, thus provides a unifying language for non-Hermitian quantum mechanics, operator theory, integrable systems, signal processing, coding, and mathematical physics. Its study clarifies the algebraic and analytic mechanisms that preserve spectral and probabilistic structures even when conventional orthogonality or Hermiticity is absent, and it underpins broad classes of physical and mathematical phenomena across disciplines.