Preprint points to a logarithmic-depth threshold for learning quantum circuits
A model of random all-to-all two-qubit circuits links the transition to lightcone growth, but the result depends on concentration assumptions.
2241–2260
A model of random all-to-all two-qubit circuits links the transition to lightcone growth, but the result depends on concentration assumptions.
A structure-aware compiler produced shorter modeled execution plans than existing methods, while larger jobs took many hours to compile.
A theoretical study says polarization, frequency and dephasing could reveal whether fluctuating Cooper pairs show thermodynamic or kinetic nonreciprocity.
Using one selected basis cut measurement settings 81-fold in numerical tests, but hardware accuracy fell to the five-class random-guess baseline at 16 qubits.
In five modeled inspirals, no-inversion waveforms matched above 0.999865, but the result comes from a restricted calculation rather than a full LISA signal model.
A Self–Exchange Relation system recorded lower negative log-likelihood than matched multi-head attention models at three tested scales, while speed results and secondary benchmarks were less consistent.
A mathematical study identifies the unique pairing that maximizes Spearman’s rho at each fixed footrule, while leaving a related ξ–η region partly unresolved.
Finite model calculations map a boundary with chiral superfluids and suggest momentum patterns for telling the phases apart.
For every ellipsoid with ordered positive semiaxes, its unit disk bundle’s Hofer–Zehnder capacity is the smaller of twice the systole and a shortest closed geodesic with Morse index 3.
A theoretical qubit study finds that independently tuned damping and anti-damping can alter divisibility and sometimes reduce deviation from ideal evolution.
An adaptive radial-basis-function controller was tested on a modeled master–slave pair; the baseline setting had the best reported tracking and settling measures among three settings.
A theoretical study gives algorithms with quadratic or linear graph-size terms, while a related model-checking problem is hard across every fixed level of the polynomial hierarchy—even on trees of depth 4.
The symbolic method returns exact fractions for a repeated-weight polynomial neural-network benchmark, while deeper configurations sometimes ran beyond the reported one-hour limit.
An analytical model links internal composition to compactness limits, but does not test realistic neutron stars or dynamical stability.
In four mathematical market settings, a binding terminal CVaR rule was associated with lower risky exposure after adverse outcomes while preserving participation in favorable states.
The mathematical model shows an infiltration front can follow square-root, linear or exponential paths depending on hydration and porosity.
The local method starts with a restricted Ricci-flat geometry, reshapes it with a Buchdahl parameter and completes the result with a path through scalar target space.
A decoder found a gain over a permuted baseline in one participant’s recordings, but the study does not show that spontaneous inner speech can be decoded.
In computer-based tests, the method put known links near the top, but it cannot show that changing a driver would change an outcome.
Calculations on a W(110) surface connect SARPES with its inverse, but only for the corresponding time-reversed setup.