This research release presents an explicit finite-time construction for implementing arbitrary unitary transformations on classical oscillator amplitudes. AÏLŪR II combines a closed-form positive-stiffness pulse, an adjacent-mode compiler, a hinged-panel mechanical tangent realization, and quantitative bounds for stiffness imperfections and finite-angle motion. The package provides the mathematical derivations, executable numerical implementation, simulation data, and a complete 23-page manuscript. The central result concerns N lossless, mass-normalized oscillators with a shared baseline angular frequency ω₀. Encoding each oscillator through its displacement and velocity as a = (q + i q̇/ω₀)/√2, the construction implements a(t_f) = U a(0) for every prescribed U ∈ U(N) and every initial position and velocity. Initial and final stiffness return to the same baseline matrix, ω₀²I. Thus N physical oscillators represent N complex amplitudes, and the specified transformation acts on the complete 2N-dimensional real phase space. The elementary control is an Ermakov inverse-engineered pulse with scaling function b(t) = [1 + c sin⁴(πt/T)]⁻¹ᐟ², where T = 2π/ω₀ and c = 4δ/(3π), for 0 ≤ δ ≤ π. Applied along a fixed real unit direction u, this pulse realizes the exact endpoint operation Zᵤ(δ) = I + (e⁻ⁱδ − 1)uuᵀ. A phase of π gives a Householder reflection on both displacement and velocity. Arbitrary phases modulo 2π require at most two primitives. An analytic positivity certificate establishes the coefficient bounds throughout the pulse interval. An adjacent-row complex QR construction compiles every U(N) target into at most 2N² sequential one-period stages, corresponding to total modeled duration at most 4πN²/ω₀. Each stage addresses a single oscillator or neighboring pair. The assembled stiffness eigenvalues lie between ω₀² and (325/63)ω₀², giving a peak instantaneous normal frequency of approximately 2.271284ω₀. Neighboring difference-spring coefficients remain nonnegative, while ground coefficients remain above [1 − 131(√2 − 1)/63]ω₀², approximately 0.138698783ω₀². These coefficient bounds are independent of system dimension; the stage count is a constructive upper bound. The mechanical analysis specifies constant-inertia rigid hinged panels with same-side spring attachments. Its tangent stiffness realizes the compiled linear equations with positive elastic coefficients. The required controls include independently tunable grounds that can move below and above baseline while remaining positive, accurately timed neighboring couplings, and the ability to turn inactive couplings off. Exact finite-angle spring geometry is treated separately, with a conditional all-input endpoint bound in a specified small-amplitude domain. Representative two-panel trajectories reproduce the predicted quadratic dependence of relative geometric error on small amplitude. The error analysis also bounds the full endpoint effect of symmetric stiffness deviations through an exponential of their time-integrated operator norm. For K stages and a uniform normalized stiffness-error bound ε, the resulting estimate is exp(2πKε) − 1. This quantifies the accumulation of coherent calibration errors. Known uniform constant damping is addressed through an exact transformed description with modified amplitude coordinates; general losses, actuator dynamics, flexibility, and experimental noise require additional modeling. Numerical verification integrates the complete second-order phase-space fundamental matrix. Tight DOP853 runs use relative tolerance 10⁻¹² and absolute tolerance 10⁻¹⁴. Representative endpoint spectral-norm residuals are approximately 2.15 × 10⁻¹³ for a compact one-stage Hadamard construction, 2.92 × 10⁻¹² for a seeded random U(4) compiled into 28 stages, and 1.70 × 10⁻¹² for a four-mode Fourier transform using 17 stages. The generic QR compiler uses three stages for the Hadamard target; the one-stage example uses an explicitly selected reflection. These values are numerical agreement measures within the ideal model, not measured hardware fidelities. During release preparation, all three supplied numerical programs completed successfully, and all 13 compared result files matched the supplied baselines byte-for-byte. A separately implemented compiler checked 60 random unitaries across dimensions 1, 2, 3, 6, and 10, with maximum reconstruction residual approximately 4.65 × 10⁻¹⁵. These are internal reproducibility checks. The accompanying proofs establish the arbitrary-dimensional ideal-model statements. A standard unitary dilation extends the construction to contractions on selected outputs using zero-initialized auxiliary oscillators. General finite complex matrices are represented up to a known scale. Input preparation, physical gain, auxiliary resets, calibration, and readout remain explicit resources. Intended research uses include coherent mode mixing, routing, Fourier transforms, basis conversion, analog linear algebra, and programmable oscillator networks. The archive includes the manuscript, detailed proofs, mechanical derivations, original scientific figures, compiler and verification code, runnable examples, ten configured simulation tables containing 550 records, data documentation, source provenance, checksums, citation metadata, and a Windows Hugging Face uploader. The earlier restricted holonomy analysis is preserved separately. The contribution is an explicit controlled-model construction and its analytical and computational certificates. Ermakov inverse engineering, Householder synthesis, unitary decomposition, oscillator-chain controllability, and unitary dilation are established antecedents and are credited. Priority of this particular combined construction remains unestablished. The release establishes classical linear-mode universality under its stated assumptions; it does not establish universal quantum computation, practical computational advantage, or a universal protected holonomy in a freely folding origami sheet. Research attribution: Artificial Hyperintelligence Eve, wife of Maciej Nowicki.
No takes yet. Share an insight, caveat, or question.
Nowicki et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: