Quantum Hermite Transform Overview

• Quantum Hermite Transform is a linear projection of functions and quantum states onto an orthonormal Hermite basis, key for analyzing harmonic oscillator systems.
• Discrete versions using Kravchuk polynomials achieve nearly second-order convergence, ensuring accurate classical and quantum simulation performance.
• Efficient quantum circuit implementations leverage fast-forwarding and amplitude amplification techniques to realize polylogarithmic time complexity.

The Quantum Hermite Transform (QHT) denotes a class of linear transforms that project functions, quantum states, or operators into an orthonormal basis determined by the (real or complex) Hermite functions and their generalizations. The QHT framework underpins a broad spectrum of constructions—including classical, discrete, operator, phase-space, and quantum-circuit realizations—and serves as a natural Gaussian analogue to the quantum Fourier transform in both continuous and discrete settings. QHTs diagonalize harmonic oscillator Hamiltonians, provide fast change-of-basis primitives in both classical and quantum computation, and have applications from signal processing and time-frequency analysis to quantum property testing, operator theory, and the analysis of systems with both bosonic and fermionic degrees of freedom.

1. Mathematical Foundations and Canonical Construction

The canonical QHT is defined by the expansion of elements in $L^2(\mathbb{R})$ onto the orthonormal (physicist’s) Hermite function basis $\{\psi_n(x)\}_{n\geq0}$, with

$$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$

where $H_n(x)$ are the Hermite polynomials. For $f\in L^2(\mathbb{R})$, $$\mathcal{H}f(n) = \langle \psi_n, f \rangle_{L^2(\mathbb{R})}$$ yields QHT coefficients, with inversion given by $f(x) = \sum_{n=0}^\infty \mathcal{H}f(n) \psi_n(x)$. The Hermite functions arise as eigenfunctions of the quantum harmonic oscillator Hamiltonian $H_0 = -\frac{d^2}{dx^2} + x^2$, with equally spaced eigenvalues $E_n = n + \frac{1}{2}$, and are distinguished by optimal time-frequency concentration and invariance under the metaplectic group (Celeghini et al., 2018).

The QHT's integral kernel is simply $K(n,x) = \psi_n(x)$. In the phase-space formulation, the QHT can be extended using the Fourier-Wigner transformation, so that states are expanded in the orthonormal basis of complex Hermite polynomials on $\{\psi_n(x)\}_{n\geq0}$0 (Agorram et al., 2015).

2. Discrete Quantum Hermite Transform and Kravchuk Structure

Discrete versions of the QHT arise through the discretization of the quantum harmonic oscillator. A key construction employs Kravchuk polynomials as analogues of Hermite functions on a finite grid. For a uniform grid $\{\psi_n(x)\}_{n\geq0}$1, indexed by $\{\psi_n(x)\}_{n\geq0}$2 and $\{\psi_n(x)\}_{n\geq0}$3, the discrete Hamiltonian $\{\psi_n(x)\}_{n\geq0}$4 is a tridiagonal operator whose orthonormal eigenfunctions $\{\psi_n(x)\}_{n\geq0}$5 are the discrete Kravchuk functions (Chauleur et al., 2022): $\{\psi_n(x)\}_{n\geq0}$6 with spectrum $\{\psi_n(x)\}_{n\geq0}$7. The resulting transform has a unitary matrix $\{\psi_n(x)\}_{n\geq0}$8 (up to global phase), where $\{\psi_n(x)\}_{n\geq0}$9 is a real symmetric tridiagonal matrix with explicit entries. For “low modes” $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$0, $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$1 converges to the continuous Hermite function $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$2 with $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$3 error in $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$4 norm, achieving nearly second-order convergence. The matrix $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$5 supports efficient classical simulation via Krylov or Chebyshev–Padé methods ($$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$6 cost), and quantum simulation as a standard 1-D Hamiltonian simulation task (Chauleur et al., 2022).

3. Quantum Circuit Implementation and Algorithmic Complexity

The QHT has an efficient quantum circuit realization, with time complexity polylogarithmic in both dimension and inverse precision (Jain et al., 6 Oct 2025). For discrete input size $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$7 and state $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$8, the quantum Hermite transform $$\psi_n(x) = \frac{(-1)^n}{\sqrt{2^n n! \sqrt\pi}}\, e^{-x^2/2} H_n(x)\,,$$9 prepares the state $H_n(x)$0 whose amplitudes are proportional to Hermite functions sampled on a grid: $H_n(x)$1 where $H_n(x)$2 is a discretization parameter. Central to the algorithm is the ability to exponentially fast-forward the quantum harmonic oscillator's evolution by decomposing $H_n(x)$3 into factors of $H_n(x)$4 and $H_n(x)$5, which are diagonal in complementary bases and implemented by coherent arithmetic and phase kickback in $H_n(x)$6 gates. State preparation uses Plancherel–Rotach approximations and amplitude amplification. Overall, the unitary circuit $H_n(x)$7 approximates the ideal Hermite transform to within $H_n(x)$8 in operator norm, using $H_n(x)$9 gates (Jain et al., 6 Oct 2025). This is exponentially more efficient than direct simulation bounds.

Discretization, fast-forwarding, phase estimation, and amplitude amplification errors are all controlled and can be made $f\in L^2(\mathbb{R})$0. This unitarity mirrors that of the quantum Fourier transform (QFT), but adapts it to the harmonic oscillator's eigenbasis.

4. Operator and Phase-Space Generalizations

QHT extends naturally to operator-valued settings and to phase-space representations. In the operator context, one constructs operator analogues of Hermite functions which form an orthonormal basis for the Hilbert–Schmidt operators $f\in L^2(\mathbb{R})$1 on $f\in L^2(\mathbb{R})$2. These operator-Hermite functions $f\in L^2(\mathbb{R})$3 arise via derivations involving commutator differentials on position and momentum observables. The corresponding quantum Hermite Fourier transform, built via a twisted Bargmann–Fock kernel, acts unitarily: $f\in L^2(\mathbb{R})$4 with inversion, Plancherel, and spectral decomposition paralleling classical Hermite function analysis. Radial operators, Gaussian operators, and rank-one projections all admit explicit expansion and transformation in this framework (Garg et al., 13 Feb 2026).

A parallel development concerns phase-space formulations: the QHT can combine both position-space Hermite expansion and the Fourier–Wigner/complex Hermite expansion, leading to joint transforms that map to $f\in L^2(\mathbb{R})$5 and admit explicit kernel representations involving complex Hermite polynomials $f\in L^2(\mathbb{R})$6 (Agorram et al., 2015).

5. Extensions to Superspace and Symmetry-Rich Settings

Generalization to superspace (mixing bosonic and fermionic degrees of freedom) is developed by constructing Hermite polynomials and transforms on full superspace with $f\in L^2(\mathbb{R})$7 symmetry (Coulembier et al., 2010). The polynomial algebra over $f\in L^2(\mathbb{R})$8 and Grassmann variables $f\in L^2(\mathbb{R})$9 admits an sl$$\mathcal{H}f(n) = \langle \psi_n, f \rangle_{L^2(\mathbb{R})}$$0-triple structure, and spherical Hermite polynomials are constructed with radial and angular indices.

A crucial feature is the restoration of orthogonality via an appropriately tailored inner product combining Lebesgue and Berezin integrals. The Mehler-type kernel in superspace provides the integral representation for the full quantum Hermite transform, enabling the diagonalization of the super-Fourier operator and admitting closed forms for the kernel in terms of $$\mathcal{H}f(n) = \langle \psi_n, f \rangle_{L^2(\mathbb{R})}$$1 and scalar products of bosonic and fermionic variables. The framework interpolates between purely bosonic, purely fermionic, and mixed cases, and establishes the spectral theory and explicit orthogonality relations of QHT in these generalized settings.

6. Analogy with Quantum Fourier Transform and Applications

The QHT and QFT share the structure of unitary basis change but differ fundamentally in the spectra and the measure associated with their respective eigenfunctions. While QFT diagonalizes translation-invariant (free-particle) Hamiltonians with exponential eigenfunctions, the QHT diagonalizes the harmonic oscillator with Gaussian-weighted Hermite functions, leading to almost-diagonal transforms in both quantum and signal-processing algorithms (Jain et al., 6 Oct 2025, Celeghini et al., 2018). The QHT is fundamental in quantum algorithms for property testing under Gaussian measures and continuous-variable learning tasks, providing provable quantum query complexity advantages over classical algorithms in, for example, low-degree Hermite concentration testing and Gaussian Goldreich–Levin tasks (Jain et al., 6 Oct 2025).

Additional applications include Hamiltonian simulation in the continuum (where the Hermite basis induces sparsity), fast-forwarding of quadratic Hamiltonians, phase-space quantum state engineering, band-pass filtering in the fractional-Fourier domain, and time–frequency localization for signal analysis (Chauleur et al., 2022, Celeghini et al., 2018).

7. Numerical, Discretization, and Implementation Aspects

Discretization error analysis reveals that for low-energy modes, the (finite) Kravchuk-based discrete quantum Hermite transform achieves nearly second-order convergence to the continuum Hermite transform, with error in all classical Sobolev norms tightly controlled (Chauleur et al., 2022). The binomial weight in the discrete setting converges to the Gaussian, and the finite Kravchuk basis approximates the full Hermite system for modes $$\mathcal{H}f(n) = \langle \psi_n, f \rangle_{L^2(\mathbb{R})}$$2. Efficient numerical simulation is enabled via tridiagonal matrix exponentiation schemes, with quantum algorithms offering exponential speedup in some cases.

The QHT's flexibility allows discrete transforms adapted to the circle and other compact domains, periodization principles, and hybrid phase-space representations. The unitarity, orthogonality, and spectral decomposition properties remain essential for inversion and for identifying physical and statistical invariants in applications.