arXiv:2610.03788
The Quantum Sphere: A Physically Realizable Optimization Benchmark with Provable Linear Convergence in White- and Black-Box Settings
Ofer M. Shir
Abstract
We revisit an established Quantum Control objective, characterize its exact local metric geometry, and reposition it as a rigorous benchmark for Search and Optimization. The trap-free topology of the landscape, established two decades ago, guarantees unhindered optimal pathways, but global topology alone does not govern convergence speed. Although the so-called Quantum Sphere is defined by the fourth power of the control field, we show that its yield gap admits an exact representation as a nonnegative linear combination of $\sin^2$ terms. From this single identity, we derive a two-sided quadratic sandwich with ratio $4/π^2$ on a certified, worst-case quarter-wave region, proving the optimal peak nondegenerate modulo its affine gauge and well-conditioned. For white-box optimization, on a halved (eighth-wave) region, the objective satisfies a Polyak-Łojasiewicz inequality with an explicit constant, granting Gradient Descent a linear rate of $1 - 16/(π^4κ_r)$, where $κ_r$ is the condition number of the relevant Hessian block. For black-box optimization, we prove that the elitist single-parent Evolution Strategy converges linearly, almost surely and in expected hitting time, from a certified sublevel set, despite the landscape's continuous gauge symmetry and its lattice of periodic optima; the certified level is explicit whenever the landscape is trap-free, and near the optimum the strategy provably sees only the quadratic tangent of the peak. Systematic numerical simulations corroborate the theory. Finally, we show that additive input (actuator) noise reweights each term of the representation by a Debye-Waller factor without disrupting the landscape's structure, and outline open challenges: output (signal) noise, far-field plateau escape, angular-synchronization certificates, and another physically scalable benchmark (SHG with Glass).
AI Summary
The paper asks when the second-harmonic-generation (SHG) yield used in spectral-phase pulse shaping admits quantitative optimization guarantees, rather than only trap-free topology. It treats the dispersion-free, fixed-amplitude objective as a physically realizable benchmark, called the Quantum Sphere. Expanding coherent sums over equal-frequency-sum shells expresses the yield gap exactly as $G(\delta)=\sum_k w_k\sin^2(a_k^\top\delta)$ with $w_k\ge0$. For contiguous amplitude support, optimal phases are affine modulo $2\pi$; the Hessian kernel comprises global phase and delay. Writing $Q=\delta^\top H\delta/2$, the paper proves $4Q/\pi^2\le G\le Q$ when the phase is within $\pi/4$ in infinity norm of the affine gauge. On the halved region it establishes convexity and an explicit Polyak–Łojasiewicz inequality, yielding gradient-descent contraction $1-16/(\pi^4\kappa_r)$ from a smaller certified initialization set. An alternating phase comb establishes sharpness of the stated sandwich and convexity radii. Gauge reduction, periodic-basin bounds, and existing drift-analysis machinery establish almost-sure linear convergence and logarithmic expected hitting time for a noiseless $(1+1)$ evolution strategy from a certified low-gap sublevel set. Simulations at $N=32$–$160$ examine these predictions. Gaussian actuator noise exactly reweights the noise-averaged landscape’s shell terms. The guarantees are local; discrete trap-freedom is not proved, and the explicit black-box starting level is conditional on its absence. The black-box rate constant is non-explicit. Global plateau escape, optimization with noisy evaluations, and transferring quadratic terminal dynamics into an asymptotic rate remain open.
Error found: The abstract's unrestricted well-conditioning assertion fails for allowed positive amplitude profiles, since A=(1,1,1,ε) produces an unbounded gauge-reduced Hessian condition number as ε approaches zero. The abstract's opening trap-freeness statement obscures that trap-freeness remains unproved for the analyzed discrete benchmark, although the black-box theorem correctly states the necessary conditional starting level. The quarter-wave barrier's half-wrap upper bound is false for a three-pixel support: its middle-pixel half-wrap is a global optimum with zero gap, while the barrier is strictly positive.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.