Papers

Engineering shape optimization faces challenges in both expert-dependent problem setup and surrogate-model reliability. In practical aerodynamic design, optimization settings such as editable regions, deformation ranges, and design-preservation constraints are typically specified manually by experienced engineers, while surrogate-based optimization may become unreliable for heterogeneous geometry databases and out-of-distribution designs. To address these challenges, we propose a knowledge-constrained shape-optimization framework that translates knowledge-based constraints and user intent into quantifiable parameters of DFFD-based deformation operators, enabling engineering-aware and controllable constrained optimization. We further develop a Mixture-of-Experts Neural Operator (MoE-NO) to improve drag prediction and trend consistency over heterogeneous aerodynamic datasets. Based on the MoE-NO encoder and Mahalanobis distance, an uncertainty-estimation strategy is introduced to detect out-of-distribution geometries and selectively trigger physics-solver feedback for local sample enrichment. Experiments on in-house MPV, SUV, and Sedan datasets show that MoE-NO achieves a test-set MAPE of $1.16\%$ and a trend-prediction accuracy of $94.34\%$, outperforming the best baseline results of $1.52\%$ and $90.34\%$, respectively. Vehicle shape-optimization experiments further yield CFD-validated drag coefficient reductions of approximately $4\%$ to $10\%$.

Motivated by the recent rebuttal of black hole–dark matter solutions based on velocity-matching, we reconstruct the spacetime by treating the dark matter halo density directly as an energy-density source in the Einstein field equations. We obtain an exact static, spherically symmetric black hole solution and investigate the effects of the halo on its geometry, optical properties, quasinormal modes, and thermodynamics. Using null geodesics, we show that the halo modifies photon trajectories, shifts the circular photon orbit, and enlarges the photon sphere and apparent shadow. The Lyapunov exponent of the unstable null orbit further determines the eikonal quasinormal modes, yielding explicit halo-induced corrections to their oscillation and damping rates. We also compute the mass, Hawking temperature, entropy, heat capacity, and Gibbs free energy to characterize the thermodynamic stability and phase structure. We find that the dark matter halo enlarges the domain of thermodynamic stability and introduces nontrivial phase behavior absent in the Schwarzschild case, demonstrating that astrophysical dark matter environments can leave potentially observable imprints on black hole observables.

Neutron stars and related compact objects are unique laboratories for probing matter at supranuclear densities and gravity in the strong-field regime. In this thesis, we investigate the hydrostatic equilibrium of compact stars in different geometric and gravitational scenarios, combining relativistic stellar structure, dense-matter microphysics, and numerical modeling. We derive the standard Tolman-Oppenheimer-Volkoff equation in General Relativity, introduce an effective one-parameter deformation scheme leading to the deformed TOV (D-TOV) formalism, and extend the hydrostatic equilibrium framework to f(R,T) gravity. Using these structure equations, we compute equilibrium sequences for neutron stars and strange stars described by the GM1 equation of state, the MIT Bag Model, and a polytropic equation of state. In General Relativity, the deformation parameter D significantly affects the global stellar properties: oblate configurations generally support larger masses and radii than the spherical case, while prolate configurations lead to less massive and more compact stars. In optimized and linear f(R,T) models, trace-dependent matter-geometry couplings further modify the mass-radius relation and the maximum supported mass. Overall, the results indicate that effective deformation and trace-dependent matter-geometry coupling can significantly affect the equilibrium structure of compact stars.

Variational Gaussian wave-packet dynamics is usually obtained by restricting quantum evolution to Gaussian wave packets. We instead construct finite-$N$ dynamics for a labeled bundle of classical trajectories under a constraint fixing the symplectic-area scale of its covariance. At finite $N$, the bundle itself is the dynamical state: it may be non-Gaussian, and the potential force on each trajectory is evaluated from the full potential without a local harmonic approximation. For sequences whose initial empirical distributions approach a smooth Gaussian density as $N \to \infty$, the limiting relative motion becomes linear and preserves Gaussianity. With the area scale set to $\hbar/2$, the limiting centroid and width equations coincide with those of the Gaussian time-dependent variational principle, and the limiting phase-space density coincides with the corresponding Gaussian Wigner density for matched initial data. Here $N$ counts trajectories, not orders in a Moyal or $\hbar$ expansion. This phase-space correspondence reveals two complementary constructions of the same Gaussian dynamics: a variational reduction of quantum evolution and the Gaussian large-$N$ limit of a constrained classical trajectory bundle.

A symmetric quartic potential is a physics model with expansive applications, ranging from broadband energy harvesters, quantum tunneling in molecules and the early universe, to torque-free spacecraft rotation. Its rich dynamics have been traditionally classified into regimes and expressed as disjointed time-domain solutions. Here we build a taxonomy for this broad class of motions and discover that their regimes exhibit a symmetrical spectral structure: they share a fundamental clock, obey parity selection, and dissolve into the separatrix through a discrete-to-continuum transition. Applied to the famous Dzhanibekov effect where a rotating body periodically undergoes rapid 180-degree flips in its attitude, the taxonomy reveals its spectral anatomy. The three principal-axis rotations share a common clock while occupying distinct parity channels, with stable-axis branches exchanging DC bias across the separatrix. We discuss a case study where the three spectral pillars: clock, parity, and continuum, survive the Wick rotation from real-time into imaginary-time kinematics.

We study three-dimensional Einstein gravity with negative cosmological constant on non-compact spatial boundaries within the Chern-Simons formulation. Using an exact fluid/gravity correspondence, we derive a closed radial flow equation for the quasi-local stress tensor and show that it realizes the holographic $\sqrt{T\bar T}$ deformation at finite cutoff. We further develop the inverse-scattering description of the boundary dynamics, identifying the gravitational interpretation of the associated spectral data and analyzing the finite-cutoff deformation of soliton solutions. Although the boundary evolution is governed by an integrable bi-Hamiltonian hierarchy, we show that the radial flow itself is not Hamiltonian with respect to the canonical Poisson structure. Our results establish a unified framework connecting integrability, quasi-local gravitational observables, inverse scattering, and finite-cutoff holography on non-compact boundaries.

Microscopic disorder in a coherent conductor is encoded in the magnitude of the transmitted current and in the energy- and channel-resolved structure of the scattering response. We develop a conductance-balanced learning framework for identifying microscopic disorder mechanisms from coherent quantum-transport spectra. Using armchair graphene nanoribbons as controlled multichannel tight-binding waveguides, we generate ensembles in which distinct disorder classes occupy the same intervals of integrated transmission, removing the dominant shortcut associated with average conductance. Within this constrained setting, we compare scalar transmission spectra with transmission-eigenvalue spectra, and use supervised classification, scale-normalizing controls, plateau-resolved tests, and principal-component analysis to identify the spectral structures that remain informative. The learned distinctions persist after normalization of the conductance envelope and are strongest in multichannel energy windows, where mode mixing and channel redistribution shape the scattering response. Principal components provide interpretable transport coordinates whose loadings identify the energy and eigenchannel sectors responsible for the dominant spectral deformations. The results establish a controlled AI-assisted protocol for learning disorder signatures from nonlinear quantum-scattering data.

The Dirac or Majorana nature of neutrinos remains one of the most fundamental open problems in particle physics. A natural intermediate scenario arises when small lepton-number-violating Majorana mass terms break the exact Dirac symmetry. These cause mass eigenstates to form nearly degenerate pairs, each separated by a small mass splitting. We study this quasi-Dirac scenario within a five-neutrino framework, extending the Standard Model by two right-handed neutrinos, which form a nearly degenerate sterile pair alongside the three active states. This extended mixing structure introduces additional phenomenological parameters, including new mixing angles, mass splittings, and CP phases. We use appearance and disappearance data from NO$\nu$A and T2K to constrain the active-sterile mixing angles $\theta_{sa}$ and the mass splitting within the sterile pair $\Delta m_{54}^2$ and forecast the sensitivity achievable at DUNE. We analyse how these parameters modify CP-violating observables relative to the standard three-neutrino case, comparing predicted event rates across both mass orderings and as a function of $\delta_{CP}$.

Chiral molecular structure can couple to electron spin, leading to unexpected spin polarization effects. This Chirality-Induced Spin Selectivity (CISS) was first reported for DNA molecules adsorbed on gold but its microscopic origin remains unclear. We demonstrated though simulation that the exchange arising from electron-electron Coulomb interactions within the self-consistent mean field (Hartree–Fock) approximation can yield significant ($\sim 2\%$) spin polarisation for $3$-methylcyclohexanone adsorbed on Cu(111) amplifying a much smaller ($\sim 0.0014\%$) initial bias, consistent with experiment. Symmetry considerations ensure the result is physically meaningful, while its ab-initio nature ensures all parameters are physically realistic. This amplification is connected to existing studies on spin-symmetry breaking in Hartree–Fock, providing a new pathway for understanding the magnitude of CISS as an emergent phenomenon of interacting electrons.

We present radiative-convective modeling of rocky exoplanets with $He$-dominated atmospheres and low envelope mass fractions. Helium has a steeper adiabatic temperature profile than $N_2$ and $H_2$ as it has fewer degrees of freedom. Line broadening differences are small, collision-induced absorption (CIA) is not important in $He$-dominated atmospheres, and Rayleigh scattering by $He$ is weak. The combined impact of these effects is that $He$-dominated atmospheres provide more warming than $N_2$-dominated atmospheres but less than $H_2$-dominated atmospheres, for the same surface pressure. For surface pressures in the range 1-20 bar, the habitable zone for rocky planets around M dwarfs with $He$-dominated atmospheres is narrower than for $H_2$-dominated atmospheres. Nonetheless, due to their large scale height compared to $N_2$ or $CO_2$-dominated atmospheres, temperate $He$-dominated atmospheres are favorable targets for characterization via transit spectroscopy with JWST.

Higgs Branches of 5d SCFTs are hyperkähler cones, symplectic varieties with interesting nested singularities governing the structure of possible Higgs branch RG flows. The main purpose of this note is to investigate 5d Higgs Branches via a geometric route: we study 5d SCFTs geometrically engineered from M-theory on non-compact Calabi-Yau threefolds, with non-isolated singularities. We match the (dynamical) complex structure deformations of the threefold with the symplectic leaves in the Higgs branch of the 5d SCFT, constructing its stratification purely relying on geometry. Consistency checks for our proposal are presented exploiting magnetic quivers.

The localization transition of the sub-Ohmic spin-boson model generates boson bath correlations that are, at low frequencies, analytically inaccessible to standard coherent-state-polaron variational ansatzes. In this paper, we introduce a “squeezed polaron” wavefunction incorporating generic Gaussian boson-boson correlations induced by the spin impurity. This gives the correct critical power-law scaling of bath observables near the localization transition together with a systematically improved ground state energy and a more accurate determination of the critical coupling. Our wavefunction captures the correct mean-field nature of the transition in the deep sub-Ohmic region. It also displays non-mean-field critical exponents in the shallow sub-Ohmic regime. Using the squeezed polaron as a starting point, we derive scattering phase shifts for bosons, and show how they encode the emergent energy scale that vanishes at the localization transition.

Purpose: To compare detuning performance and evaluate the power requirements of optical detuning methods, and to demonstrate the feasibility of an optically detuned four-channel receive array. Methods: Four optical detuning methods were compared in simulations, bench tests, and phantom measurements at 3T against conventional galvanic detuning. Passive detuning was also tested as an additional wireless detuning option. Optical power requirements for the detuning networks were investigated, and a flexible, optically detuned 4-channel shielded-loop resonator (SLR) array was constructed and tested in vivo. Results: A photodiode-PIN diode combination exhibited the highest unloaded Q (68.6) and Q ratio (1.9), with detuning performance and signal-to-noise ratio comparable to that of galvanic detuning at an optical power of 10 mW. Using this detuning strategy, in vivo images of the knee and brain were successfully acquired with a 4-channel flexible array. Conclusion: Optical detuning is a practical alternative to conventional galvanic detuning in flexible SLR arrays. With advances in optical signal and power transmission, optimizing optical detuning while meeting manageable power requirements is an important step toward fully optical receive-coil arrays. This study provides a baseline for the total optical power required for active detuning in such optical coil systems.

We present an efficient and scalable computational framework for pseudopotential Kohn-Sham density functional theory (KS-DFT) calculations using an enriched finite element (EFE) basis. The EFE basis is formed by augmenting the classical finite element (CFE) basis with compact atom-centered functions, which we term enrichment functions. The key idea is to combine the completeness of a finite element basis with the efficiency of an atom-centered basis. We orthogonalize the enrichment functions with respect to the underlying CFE basis to simultaneously improve the conditioning of the EFE basis and the efficiency of evaluating the inverse of the overlap matrix. To efficiently solve the Kohn-Sham eigenvalue problem, we employ a residual-based Chebyshev subspace iteration approach that is tolerant to approximations in the evaluation of the inverse of the overlap matrix. We demonstrate the accuracy of the framework as compared to the widely available DFT packages. For benchmark non-periodic calculations, ranging up to 39,083 electrons, the EFE basis offers a $5-7\times$ reduction in degrees of freedom over the CFE basis. As a result of this, EFE achieves a $5-9\times$ reduction in computational cost over the CFE basis. The EFE basis also provides a $4-5\times$ reduction in the required memory compared to the CFE basis, thus allowing for optimal utilization of computational resources. Finally, we demonstrate that the EFE basis affords good parallel scalability. Overall, the EFE basis offers a systematically convergent, fast, scalable, resource-efficient basis for pseudopotential DFT calculations.

Quantum many-body states with sufficiently low energy variance can serve as approximations to thermal states, and they may be prepared by energy filtering simple pure states. In this work, we examine how narrow the filter width must be to guarantee thermal behavior. To this end, we analyze the problem in the Floquet regime, where filtered states are found to be equivalent to time averages. This equivalence allows us to reproduce the distinct Rényi-$\alpha$ entropy scalings as reported in [Morettini et al., Physical Review Letters 133, 240401 (2024)]. Crucially, we show that under the Floquet eigenstate thermalization hypothesis, the trace distance between Floquet-filtered states and thermal states is bounded by the square root of filter width. We further demonstrate that these results extend naturally to the conventional Hamiltonian setting by mapping Hamiltonian-filtered states to their Floquet counterparts.

Electrons trapped above the surface of superfluid helium are a disorder-free platform for investigating the formation and dynamics of low-dimensional Wigner crystals. A characteristic nonlinear transport feature of this electronic solid suspended above the helium surface is the Bragg-Cherenkov effect, in which the mobility of the smoothly moving crystal is limited by the coherent emission of helium surface waves (ripplons). The effect has been understood in the conventional Cherenkov setting in which the crystal moves at a constant speed. Here we report on transport measurements of electrons on helium confined in a microchannel geometry to investigate the non-equilibrium response of the Wigner solid when it is subjected to a high-frequency driving field. Surprisingly, the experiments reveal a strongly nonlinear transport response of the confined Wigner solid at frequencies nearly an order of magnitude larger than the ripplon frequencies contributing to the conventional Bragg-Cherenkov effect. We relate this observation to the coupling of the Wigner solid to ripplons with higher-order Bragg vectors, which gives rise to a dynamical friction that provides a mechanism for the observed high-frequency pinning.

Outflows in active galactic nuclei (AGN) are recognized as a fundamental mechanism driving the co-evolution of supermassive black holes (SMBHs) and their host galaxies. Although powerful outflows are frequently detected in gas-rich, active star-forming galaxies, their existence and potential impact within gas-poor, quiescent galaxies remain poorly understood. We report the first detection of a powerful ultra-fast outflow (UFO) in a nearby quiescent galaxy KUG 1208+386, providing a multiscale analysis of AGN winds from nuclear to galactic scales. We detect a nuclear X-ray UFO with a velocity of $v_{out} \approx -0.07c$ and a kinetic power of $\dot{E}_{\rm UFO}=(0.8\mbox{--}6.5) \times 10^{43}$erg/s, sufficient to drive effective AGN feedback ($\dot{E}_{\rm UFO}/L_\mathrm{Edd}=(1\mbox{--}8)\%$) and far exceeding the galactic [OIII] outflow power $\sim 10^{40}$ erg/s. Host galaxy analysis reveals a massive quiescent system (specific star formation rate $\sim3\times10^{-12}\mathrm{yr}^{-1}$) that has been quenched $\sim$9 Gyr ago. The central AGN is obscured by a line-of-sight (LOS) column density of $\log (N_\mathrm{H}^\mathrm{LOS}/\mathrm{cm}^{-2})\sim23$ and the circumnuclear scattering material is Compton-thick $\log(N_\mathrm{H}^\mathrm{scatter}/\mathrm{cm}^{-2})=24.7^{+0.8}_{-0.5}$. The discovery of a nuclear UFO in a long-quenched massive galaxy challenges the paradigm that UFOs are exclusive to gas-rich, star-forming systems, suggesting instead that they are governed by the local circumnuclear environments, rather than the global gas reservoir. Our results indicate that episodic, powerful winds can maintain the quiescent state of KUG 1208+386 over Gyr timescales, supporting a wind-driven `maintenance' mode of AGN feedback that is distinct from the classical jet mode.

The production of large-scale entangled states is one of the main goals of next-generation quantum technologies, with an immediate potential for applications in the context of entanglement-assisted quantum sensing. A very promising platform to achieve this goal is offered by ultracold spinor gases, made of atoms with a large internal spin sensitive to magnetic fields. Here we show that the native spin-changing collisions in a spinor Bose-Einstein condensate, combined with an arbitrary quadratic Zeeman shift, can generate scalable spin squeezing in the collective spin of the ensemble, following the universal paradigm of the celebrated one-axis-twisting model. Squeezing dynamics is driven by the quadratic Zeeman shift when this shift is small; and by the spin-changing collisions for large shifts, in the form of stroboscopic squeezing. Turning off the Zeeman shift freezes out the collective-spin dynamics, so that the ensuing collective spin dynamics can be uniquely governed by an external field to be sensed. Our theoretical results pave the way for the use of spinor Bose gases with a large spin in fundamental studies of entanglement, as well as in advanced metrological applications.

Plasma-based mineral-processing routes, such as hydrogen plasma smelting reduction (HPSR), which converts iron-ore fines directly to liquid metal in a single scalable step are commonly modeled by treating the arc as a spatially uniform heat source. Yet the reduction chemistry is governed by the strongly non-uniform conditions at the plasma-melt interface, which spatially averaged diagnostics cannot resolve. Here we spatially and temporally resolve the arc of a transferred arc HPSR reactor using multi-species optical emission spectroscopy (OES), in which neutral and ionic argon (Ar I, Ar II), hydrogen Balmer, and neutral iron (Fe I) emissions serve as intrinsic spatial filters set by their differing ionization thresholds. Combined with infrared thermography of the melt surface and an LTE thermal-plasma model validated against the benchmark free-burning argon arc, the measurements reveal a strongly stratified, non-isothermal discharge: an argon-defined core (>10,000 K), a partially recombined Balmer envelope (7,000-10,000 K), and an Fe I-traced interfacial boundary layer (3,000-4,000 K) directly above a melt surface at ~1,900-2,300 K. Across this steep thermal drop, positive hydrogen ions recombine before reaching the surface, so the reductant flux delivered to the oxide is overwhelmingly neutral; atomic hydrogen (H) and vibrationally excited molecular hydrogen H2(v), rather than the energetic ions often assumed. The measured electron density and excitation temperature bound the interfacial ionization. These findings redefine the boundary conditions for kinetic modeling of plasma-based ore reduction and establish a spatially resolved multi-species diagnostic framework transferable across plasma mineral-processing systems.

Currently, remote energy transfer and immunity to dissipation are hot topics in quantum batteries (QBs). In this work, we propose a protocol to realize energy transfer between two remote atoms (a quantum charger and a quantum battery) each coupled to a separate optical cavity with the cavities connected by a fiber. The cavities and fiber are coupled to their individual baths. After optimizing inter-system couplings to achieve an efficient transfer, we uncover the effect of suppressing dissipation by introducing parity deformation of the cavities fields. We also prove that the charger-battery entanglement is a consumable resource for energy storage: it is initially stored until the charger and battery reach energy balance, and then subsequently consumed to maintain the increase in energy stored in the battery. The present scheme is the first execution of energy transfer to a distant battery assisted by entanglement, which may help better understand quantum thermodynamics and open new possibilities toward harnessing decoherence as a resource to improve the charging performance of QBs.

We study a class of 3-wave kinetic equations arising in wave turbulence theory, with regularized kernels. For radial, nonnegative initial data, we construct an exact global-in-time strong solution which remains nonnegative and is analytic with respect to time. The proof combines a careful analysis of the resonant interaction surfaces with a time power-series construction and a continuation argument based on the conservation of the energy moment.

Pulsar timing array data analysis is computationally expensive, limiting the complexity of models which can be studied. As pulsar timing datasets and their respective models grow in size and sophistication, faster and scalable inference methods are essential. In this paper, we accelerate pulsar timing analyses by sampling in the space of Fourier coefficients instead of analytically marginalizing over them. Previous studies have shown the Fourier space induces a complex, high-dimensional posterior geometry, from which it is generally difficult to sample. We show that under an appropriate coordinate transformation the Fourier coefficients approximately follow a standard normal distribution, and may be efficiently sampled using a Hamiltonian Monte Carlo scheme. Under this coordinate transformation, for datasets of size and complexity comparable to the NANOGrav 15-year release, the new method produces converged posterior distributions for a range of models which include inter-pulsar correlations, stochastic, and deterministic signals in approximately 15 minutes on an NVIDIA GeForce RTX 3090 GPU. By comparison, the legacy pulsar timing analysis software ENTERPRISE would require months of computation on a CPU cluster to analyze comparable datasets under the same joint stochastic and deterministic models.

Recent observations of young galaxies in the high-redshift Universe reveal signs of early enrichment of nitrogen. Extremely massive stars ($M \gtrsim 10^{2}-10^3\,M_\odot$) with strong stellar winds have been proposed as a potential driver of this phenomenon. Here, we show that the observed fraction of nitrogen rich stars with masses $\gtrsim 100M_\odot$ cannot be explained solely by mass loss, requiring significantly more efficient mixing beyond their convective cores than accounted in present evolutionary models. We compile a representative sample of 122 stars in the Tarantula Nebula of the Large Magellanic Cloud (LMC) with masses $M \gtrsim 30\,M_\odot$. Nearly all stars with masses $M \gtrsim 100\,M_\odot$ exhibit strong nitrogen enrichment, by factors $\gtrsim 5-10$. We demonstrate that this trend cannot be reproduced by varying assumptions on binary fraction, star formation history, or mass-loss rates within ranges predicted by current empirical and theoretical models. In contrast, enhanced core overshooting of $\alpha_\text{ov}\gtrsim 1$ can account for the observed enrichment, but leads to quasi-chemically homogeneous evolution that is inconsistent with the observed Hertzsprung-Russell diagram. While the origin of this discrepancy remains unclear, our results, in combination with observational and theoretical constraints on mass-loss rates, suggest the presence of efficient early mixing operating during or shortly after the formation of very massive stars. Such mixing models are currently not included in stellar evolution models. These findings have immediate implications for the formation, radial expansion, evolution, and final fates of stars at the upper mass end, and provide a potential pathway to explaining the rapid nitrogen enrichment observed in the high-redshift Universe.

We review existing and construct new two-dimensional field theories that exhibit BMS$_3$ symmetry, with and without central extensions. These include interacting electric, magnetic and canonical BMS$_3$ scalar theories, as well as couplings between the electric and magnetic sectors. We provide a careful analysis of boundary contributions at the corner points $u\rightarrow\pm\infty$, determine the counterterms and boundary conditions required by BMS$_3$ invariance, and study the corresponding variational principles. Furthermore, we introduce external sources and derive the associated flux-balance laws. Within the framework of flat-space holography, we show that a simple free electric BMS$_3$ model with the appropriate central charge reproduces the monodromy classification of three-dimensional Einstein gravity. Finally, we investigate the flat-space limits of both AdS$_3$ and dS$_3$, whose boundary dynamics are described by Carrollian limits of Liouville and Euclidean Liouville theory, respectively. Although the resulting BMS$_3$ theories possess isomorphic spectra, they differ in the signs of the supertranslation charges and the central charge. This suggests that the flat-space limits of AdS$_3$ and dS$_3$ provide complementary realizations of flat-space holography.

We train a neural-network Maxwell's demon to extract work from a model of an underdamped micromechanical cantilever subject to thermal noise. The demon, which periodically adjusts the position of a harmonic trap, is trained to maximize the power extracted under steady-state operation. When the demon is given the cantilever position and trap position as inputs it learns a refined version of an existing hand-designed protocol, yielding a substantial improvement in performance. When the demon receives the oscillator velocity as input it discovers a qualitatively different strategy that extracts substantially more work, close to the theoretical power bound. Analysis of the protocol shows that it implements cold damping: the trap position is displaced approximately linearly with velocity, producing an effective increase of the oscillator's damping coefficient and a reduction of its effective temperature. Thus a neural-network Maxwell's demon rediscovers a well-known cooling strategy from optomechanics, revealing a simple physical mechanism underlying near-optimal work extraction from thermal fluctuations in an underdamped system.

We study Collins azimuthal asymmetries for pion-in-jet production in polarized proton-proton and lepton-proton collisions. We adopt a hybrid transverse momentum dependent approach, with a collinear configuration for the initial state, and employ the transversity and Collins fragmentation functions extracted from semi-inclusive deep inelastic scattering and $e^+e^-$ annihilation data. After recalling the good description of the STAR data in $pp$ collisions, which supports the universality of the Collins function, we present predictions for Electron-Ion Collider kinematics, both at leading order and by including the quasireal photon exchange in the Weizsäcker-Williams approximation. This contribution is sizable but does not spoil the dominance of quark-initiated channels. This implies that $\ell p$ processes allow for a clearer access to the transversity distribution, including its sea-quark component.

We have synthesized layered superconducting LiNbO$_2$ crystals through a bulk phase transformation from LiNbO$_3$ single crystals via CaH$_2$ reduction. As the Nb valence is reduced from 5+ to 3+, the material undergoes a structural transformation to the resulting product, LiNbO$_2$, which is accompanied by metallic behavior and a superconducting transition, Tc onset, as high as 14.4 K. Secondary ion mass spectroscopy (SIMS) and X-ray photoelectron spectroscopy (XPS) show that the resulting phase is hole-doped through de-lithiation during the reduction. Magnetization and AC susceptibility measurements from a tunnel diode resonator confirm the bulk nature of superconductivity with a superconducting volume fraction of approximately 77% and an upper critical field approaching 26 T. Our study demonstrates extreme hydride reduction as an effective method to induce phase transformations with non-topotactic pathways and can be used to synthesize bulk materials with exotic properties.

The Microscopic Dynamical Entropy (MDE) introduced in Ref. [1] describes irreversible relaxation of selected variables x within a finite closed Hamiltonian system of fixed total energy E. Here we extend the framework to arbitrary total-energy distributions and time-dependent Hamiltonians. This allows work interactions with external agents or protocols, crucial in stochastic and classical thermodynamics, and driven systems more generally, to be addressed directly. The central step is to reduce the full microscopic description in terms of selected variables x and unmonitored variables y to one in terms of the selected variables and the instantaneous total energy (x,E). The unmonitored degrees of freedom y in Y enter through their conditional phase-space volume (CPV) $\Omega_Y(x,E,t)$. This construction allows work and heat to be defined directly for finite Hamiltonian composites, including cases where Y is so small that its temperature is not well defined. We demonstrate monotonic MDE growth, corresponding to the macroscopic second law, under mixing and time-scale separation even for driven systems with arbitrary energy distributions. At trajectory level, we derive detailed and integral fluctuation relations for the MDE, generalising standard large-bath results. We verify these numerically in a driven few-particle Hamiltonian system, showing the emergence of stochastic thermodynamics for collective coordinates, such as the centre-of-mass position, even in closed systems with only ten or twenty degrees of freedom. Physically, these fluctuation relations connect irreversibility to the change in the number of unmonitored microstates compatible with initial and final data for the observed coordinates. Together, our results provide a finite-system Hamiltonian foundation for the emergence of classical and stochastic thermodynamics and extend their applicability to surprisingly small heat baths.

We show fast mixing of Glauber dynamics for certain quadratic Gibbs measures with large external fields. The main ingredient is an overlap condition that allows us to control correlation matrices uniformly over all pinnings, by controlling norms of small submatrices of the interaction matrix. Using stochastic localization, we then obtain a lower bound on the spectral gap and, consequently, polynomial-time mixing of Glauber dynamics. As a direct application, we consider the Sherrington-Kirkpatrick model, whose interaction matrix is a scaled GOE matrix. For this model, we show that for any fixed finite inverse temperature $\beta$, there exists a strength of external field $\theta$, not depending on the size of the system, for which Glauber dynamics mixes in polynomial time (with high probability on the draw of the interaction matrix).

We present a comprehensive computational study of driven quantum dot arrays in a square lattice configuration, subject to an external magnetic field and coupled to a cylindrical far-infrared photon cavity. The driving is introduced through a harmonic modulation of the full electron-photon interaction, therefore including both paramagnetic and diamagnetic contributions. The electron-electron Coulomb interactions are treated within density functional theory, while the electron-photon coupling is modeled using a many-body configuration interaction approach at each iteration of the density functional. By exploiting the unique properties of the cylindrical TE$_{011}$ cavity mode, we demonstrate selective enhancement of diamagnetic two-photon transitions. Our results reveal that the effectiveness of harmonic modulation of the electron-photon interaction is strongly dependent on both the driving frequency and the electron occupation number per dot. When the driving frequency approaches twice the cavity photon frequency, the system exhibits resonant behavior characterized by efficient photon pumping, occupation of higher-order photon replicas, and activation of collective radial Coulomb breathing modes. These findings establish a controllable mechanism for manipulating photon states in coupled quantum dot-cavity systems and provide insights into the interplay among harmonic modulation, photonic excitations, magnetic confinement, and many-body electron correlations in dimensionally reduced nanostructures.