arXiv:2610.13477
Harmonic analysis and circuit approximation of Sobolev functions on quantum input spaces
Philipp Schmocker
Abstract
We derive approximation rates for Sobolev functions defined on the complex unit sphere and complex projective space, representing quantum states, and on the special and projective unitary groups, representing quantum unitary operations. By using harmonic analysis on compact groups and compact homogeneous spaces, we characterize the corresponding Peter--Weyl decompositions and Sobolev spaces via irreducible representations and the Laplace--Beltrami operator. Combining spectral truncation with finite-depth quantum-circuit approximation, invoking the Solovay--Kitaev theorem, we establish quantitative rates that separate the spectral truncation error from the circuit approximation error.
AI Summary
The paper studies approximation of scalar-valued Sobolev functions whose inputs are quantum states or unitary operations, rather than classically encoded Euclidean data. It treats the complex sphere $\mathbb S^{2d-1}$, projective state space $\mathbb{CP}^{d-1}$, and groups $\mathrm{SU}(d)$ and $\mathrm{PU}(d)$, with $d=2^n$. The approximation first truncates the function’s Laplace–Beltrami expansion, then evaluates that truncation at a circuit-generated approximation to the input. For $f\in H^s(G)$, $s\geq0$, the common bound is $\|f-\mathcal Q_{K,L}f\|_{L^2(G)}\lesssim_{d,s}[(1+K)^{-s}+(1+K)^{m/2+1}\varepsilon_n(L)]\|f\|_{H^s(G)}$, where $m=\dim G$ and $N_K\asymp_d(1+K)^m$ modes are retained. Thus the two contributions scale as $N_K^{-s/m}$ and $N_K^{1/2+1/m}\varepsilon_n(L)$. Solovay–Kitaev synthesis supplies $\varepsilon_n(L)\leq\exp[-a_nL^{1/c}]$ for sufficiently large $L$, with admissible $c>1.44042$ using a cited refinement. The derivation combines Peter–Weyl decompositions, spectral Sobolev tail estimates, and Bernstein–Nikolskii gradient bounds controlling sensitivity to input perturbations. Representation matrix elements and state harmonics are expressed through tensor-copy circuit matrix elements or expectation values. Two single-qubit numerical examples compare empirical errors with fitted versions of the bounds. The synthesis constants depend on dimension and gate set. Circuit evaluation requires additional implementation assumptions: suitable representation states must be available, or the harmonic matrices must admit efficient implementations. The stated depth controls input synthesis, not a complete evaluation-resource budget. For the analytic numerical target, the fitted spectral exponent is explicitly interpreted as a finite-range effective exponent.
Error found: The approximation bounds control exact evaluation at circuit-synthesized inputs, not the additional state-preparation, operator-implementation, and measurement resources needed to evaluate the function on quantum hardware. Section 4 identifies generator word length with circuit depth although its arbitrary generating sets are not required to consist of single-layer circuits; this identification needs an additional assumption or a generator-dependent depth factor. The second numerical experiment incorrectly gives log_phi(2)+10^(-6) as approximately 1.44052 rather than approximately 1.44042109.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 27 pages, 2 figures