Search

This paper presents the successful application of a phase-lock feedback system to maintain the resonance condition of a radio frequency (rf) spin rotator (specifically, an rf Wien filter) with respect to a 120 kHz spin precession in the Cooler Synchrotron (COSY) storage ring. Real-time monitoring of the spin precession and the rf Wien filter signal allows the relative phase between the two to be stabilized at an arbitrary setpoint. The feedback system compensates for deviations in the relative phase by adjusting the frequency and/or phase as needed. With this method, a variation in phase relative to the demand phase with a standard deviation of $\sigma_{\Delta\varphi}\approx 0.2\mathrm{rad}$ could be achieved. The system was implemented in two runs aiming at a first direct measurement of the deuteron electric dipole moment in 2018 and 2021. In addition, the difference between a single-bunch beam affected by the spin rotator and a two-bunch system in which only one bunch is exposed to the spin rotator fields is discussed. Both methods have been used during these beam times. The ability to keep the spin precession and the rf fields synchronized is also crucial for future investigations of electric dipole moments of charged particles using storage rings., Comment: 13 pages, 10 figures

In this work, we address the challenge of approximating unknown system dynamics and costs by representing them as a bilinear system using Koopman-based Inverse Optimal Control (IOC). Using optimal trajectories, we construct a bilinear control system in transformed state variables through a modified Extended Dynamic Mode Decomposition with control (EDMDc) that maintains exact dynamical equivalence with the original nonlinear system. We derive Pontryagin's Maximum Principle (PMP) optimality conditions for this system, which closely resemble those of the inverse Linear Quadratic Regulator (LQR) problem due to the consistent control input and state independence from the control. This similarity allows us to apply modified inverse LQR theory, offering a more tractable and robust alternative to nonlinear Inverse Optimal Control methods, especially when dealing with unknown dynamics. Our approach also benefits from the extensive analytical properties of bilinear control systems, providing a solid foundation for further analysis and application. We demonstrate the effectiveness of the proposed method through theoretical analysis, simulation studies and a robotic experiment, highlighting its potential for broader applications in the approximation and design of control systems., Comment: This work has been submitted to the IEEE for possible publication

The assembly of supermassive black hole (SMBH) mass ($M_{\bullet}$) and stellar mass ($M_{*}$) in galaxies can be studied via the redshift evolution of the $M_{\bullet}-M_{*}$ relation, but the ways in which selection bias and physical assembly channels affect this evolution are uncertain. To address this, we compare the $M_{\bullet}-M_{*}$ relation for local massive ($M_{*}>10^{10.5}$M$_{\odot}$) quiescent early-type galaxies (ETGs) to that for massive ETGs hosting active galactic nuclei (AGN) at $z\sim0.8$. The restrictions on stellar mass and galaxy type limit the assembly channels that may connect the two relations. For the local sample we find $\log(M_{\bullet}) = 8.80 + 1.10(\log{M_{*}-11})$, in line with prior work. For the $z\sim0.8$ sample we find a bias-corrected relation: $\log(M_{\bullet}) = 7.80 + 1.25(\log{M_{*}-11})$. We show, however, that this relation depends on the stellar and SMBH mass functions used to compute the selection bias, the virial relation, the virial factor, and the active fraction, which together introduce uncertainty of up to $\sim0.6$\,dex in the $z\sim0.8$ relation. Adopting reasonable choices of these parameters then our $z\sim0.8$ relation lies above that for $z\sim0$ AGN by $\sim0.5$\,dex, but below our $z\sim0$ ETG relation by $0.4-1$\,dex in SMBH mass. We discuss possible sources of this offset, including further bias corrections, `downsizing" in SMBH mass assembly, and preferential SMBH growth. Our results highlight the need to reduce uncertainties from selection and measurement bias in SMBH and stellar masses at all redshifts., Comment: ApJ, accepted

The Euclid mission is generating a vast amount of imaging data in four broadband filters at high angular resolution. This will allow the detailed study of mass, metallicity, and stellar populations across galaxies, which will constrain their formation and evolutionary pathways. Transforming the Euclid imaging for large samples of galaxies into maps of physical parameters in an efficient and reliable manner is an outstanding challenge. We investigate the power and reliability of machine learning techniques to extract the distribution of physical parameters within well-resolved galaxies. We focus on estimating stellar mass surface density, mass-averaged stellar metallicity and age. We generate noise-free, synthetic high-resolution imaging data in the Euclid photometric bands for a set of 1154 galaxies from the TNG50 cosmological simulation. The images are generated with the SKIRT radiative transfer code, taking into account the complex 3D distribution of stellar populations and interstellar dust attenuation. We use a machine learning framework to map the idealised mock observational data to the physical parameters on a pixel-by-pixel basis. We find that stellar mass surface density can be accurately recovered with a $\leq 0.130 {\rm \,dex}$ scatter. Conversely, stellar metallicity and age estimates are, as expected, less robust, but still contain significant information which originates from underlying correlations at a sub-kpc scale between stellar mass surface density and stellar population properties.

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 3,000 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai., Comment: 26 pages, 6 figures

This paper explores the profound impact of User Experience (UX) design on user retention and conversion rates in mobile applications. As the mobile app market becomes increasingly competitive, understanding how UX design can enhance user satisfaction, engagement, and loyalty is crucial for developers and businesses. Through a comprehensive review of existing literature and statistical insights, this study identifies key UX design principles that contribute to improved user retention and conversion rates. Intuitive navigation, appealing visuals, performance optimization, and integration of user feedback emerge as essential components of effective UX design that drive app success. Applications that prioritize these elements foster a positive user experience, leading to higher engagement and greater retention. Additionally, UX design strategies, such as personalization and customization, have been shown to significantly increase conversion rates, demonstrating the critical the role that tailored experiences play in app success. By analyzing these principles and their impact, this paper provides valuable insights for developers aiming to enhance user satisfaction, optimize app performance, and ultimately improve business outcomes.

We investigate the thermalization dynamics of 1D systems with local constraints coupled to an infinite temperature bath at one boundary. The coupling to the bath eventually erases the effects of the constraints, causing the system to tend towards a maximally mixed state at long times. We show that for a large class of local constraints, the time at which thermalization occurs can be extremely long. In particular, we present evidence for the following conjecture: when the constrained dynamics displays strong Hilbert space fragmentation, the thermalization time diverges exponentially with system size. We show that this conjecture holds for a wide range of dynamical constraints, including dipole-conserving dynamics, the $tJ_z$ model, and a large class of group-based dynamics, and relate a general proof of our conjecture to a different conjecture about the existence of certain expander graphs., Comment: 42 pages, 13 figures

Large Language Models (LLMs) excel at rewriting tasks such as text style transfer and grammatical error correction. While there is considerable overlap between the inputs and outputs in these tasks, the decoding cost still increases with output length, regardless of the amount of overlap. By leveraging the overlap between the input and the output, Kaneko and Okazaki (2023) proposed model-agnostic edit span representations to compress the rewrites to save computation. They reported an output length reduction rate of nearly 80% with minimal accuracy impact in four rewriting tasks. In this paper, we propose alternative edit phrase representations inspired by phrase-based statistical machine translation. We systematically compare our phrasal representations with their span representations. We apply the LLM rewriting model to the task of Automatic Speech Recognition (ASR) post editing and show that our target-phrase-only edit representation has the best efficiency-accuracy trade-off. On the LibriSpeech test set, our method closes 50-60% of the WER gap between the edit span model and the full rewrite model while losing only 10-20% of the length reduction rate of the edit span model., Comment: accepted by ICASSP 2025

For commercial grade electronic devices, stable structures are required to ensure a long device life span. When such devices contain nanomaterials like graphene, it is crucial that these materials resist industrial processes and harsh environments. For environments that contain water, graphene delamination is a notorious drawback, as water intercalation and eventually liftoff readily occur in aqueous and especially in alkaline solutions. This limitation renders graphene incompatible with key wafer-processing steps in the semiconductor industry. In this work, a covalent pyrene-based adhesion layer is synthesized in a facile, two-step procedure. Through {\pi}-{\pi} interactions, the adhesion of graphene to silicon wafers was maintained under conditions that resemble harsh processes, i.e. acidic and alkaline solutions, several organic solvents, and sonication. Moreover, they could be produced with a device measurement yield up to 99.7% and reproducible device-to-device electronic performance on 4-inch silicon wafers. Our results show that a straightforward functionalization of silicon wafers with an adhesive layer can be directly applicable in industrial-scale fabrication processes, giving access to robust graphene field effect devices that are built to last long.

This work considers simultaneous homogenization dimension reduction of a poroelastic model for thin fiber-reinforced hydrogels. The analysed medium is defined as a two-component system consisting of a continuous fiber framework with hydrogel inclusions arranged periodically throughout. The fibers are assumed to operate under quasi-stationary linear elasticity, whereas the hydrogel's hydromechanical behavior is represented using Biot's linear poroelasticity model. The asymptotic limit of the coupled system is established when the periodicity and thickness parameters are of the same order and tend to zero simultaneously, utilizing the re-scaling unfolding operator. It is demonstrated that the limit displacement exhibits Kirchhoff-Love-type behavior through Griso's decomposition of plate displacements. Towards the end, a unique solution for the macroscopic problem has been demonstrated., Comment: 20 pages, 1 figure

Two-dimensional (2D) Van der Waals compounds exhibit interesting electronic and magnetic properties due to complex intra-layer and inter-layer interactions, which are of immense importance in realizing exotic physics as well as advanced technology. Various experimental and theoretical studies led to significantly different ground state properties often contrasting each other. Here, we studied a novel 2D material, CrSiTe3 employing magnetic, specific heat and Raman measurements. Experimental results reveal evidence of incipient antiferromagnetism below 1 kOe concomitant to ferromagnetic order at 33 K. Antiferromagnetic and ferromagnetic interactions coexists at low field in the temperature regime, 15 - 33 K. Low field data reveal an additional magnetic order below 15 K, which melts on application of external magnetic field and remain dark in the heat capacity data. Raman spectra exhibit anomalies at the magnetic transitions; an evidence of strong spin-lattice coupling. Below 15 K, Eg modes exhibit hardening while Ag modes become significantly softer suggesting weakening of the inter-layer coupling at low temperatures which might be a reason for the unusual magnetic ground state and field induced melting of the magnetic order. These results reveal evidence of exceptional ground state properties linked to spin-lattice coupling and also suggest a pathway to study complex magnetism in such technologically important materials., Comment: 6 figures

We obtain the equations of fluctuating hydrodynamics for many-particle systems whose microscopic units have both translational and rotational motion. The orientational dynamics of each element are studied in terms of the rotational Brownian motion of a corresponding fixed-length director ${\bf u}$. The time evolution of a set of collective densities $\{\hat{\psi}\}$ is obtained as an exact representation of the corresponding microscopic dynamics. For the Smoluchowski dynamics, noise in the Langevin equation for the director ${\bf u}$ is multiplicative. We obtain that the equation of motion for the collective number-density has two different forms, respectively, for the I\"{t}o and Stratonvich interpretation of the multiplicative noise in the ${\bf u}$-equation. Without the ${\bf u}$ variable, both reduce to the Standard Dean-Kawasaki form. Next, we average the microscopic equations for the collective densities $\{\hat{\psi}\}$ (which are, at this stage, a collection of Dirac delta functions) over phase space variables and obtain a corresponding set of stochastic partial differential equations for the coarse-grained densities $\{\psi\}$ with smooth spatial and temporal dependence. The coarse-grained equations of motion for the collective densities $\{\psi\}$ constitute the fluctuating non-linear hydrodynamics for the fluid with both rotational and translational dynamics. From the stationary solution of the corresponding Fokker-Planck equation, we obtain a free energy functional ${\cal F}[\psi]$ and demonstrate the relation between the ${\cal F}[\psi]$s for different levels of the FNH descriptions with its corresponding set of $\{\psi\}$., Comment: 45 pages

Energy transfer in the dark sector of the universe gives rise to new phenomena of special interest in modern cosmology. When dark energy is modeled as a phantom scalar field, interactions become crucial to avoid Big Rip singularities. In this work, we revisit the phase-space analysis of the field equations by introducing a new set of dimensionless variables distinct from the traditional Hubble normalization approach. These new variables define a compactified phase space for the evolution of physical parameters. We demonstrate that these compactified variables offer fresh insights into the phase-space analysis in gravitational theories, particularly when the dark energy fluid is allowed to possess a negative kinetic energy density., Comment: 18 pages, 2 tables, 8 compound figures

While Speech Foundation Models (SFMs) excel in various speech tasks, their performance for low-resource tasks such as child Automatic Speech Recognition (ASR) is hampered by limited pretraining data. To address this, we explore different model merging techniques to leverage knowledge from models trained on larger, more diverse speech corpora. This paper also introduces Selective Attention (SA) Merge, a novel method that selectively merges task vectors from attention matrices to enhance SFM performance on low-resource tasks. Experiments on the MyST database show significant reductions in relative word error rate of up to 14%, outperforming existing model merging and data augmentation techniques. By combining data augmentation techniques with SA Merge, we achieve a new state-of-the-art WER of 8.69 on the MyST database for the Whisper-small model, highlighting the potential of SA Merge for improving low-resource ASR., Comment: To appear in ICASSP 2025

Intriguing phenomenon of dimensional crossover is comprehensively studied by experimental and theoretical investigation of electronic structure in quasi-one-dimensional hollandite K$_{2}$Ru$_{8}$O$_{16}$. Valence band photoemission spectra in conjunction with density functional theory within local density approximation combined with dynamical mean field theory (LDA+DMFT) reveal moderately correlated electronic structure. Anomalous temperature dependence of high-resolution spectra in the vicinity of Fermi level suggests Tomonaga-Luttinger liquid state down to 150 K, below which it undergoes a dimensional crossover from one-dimensional to three-dimensional electronic behaviour. Monotonously decreasing spectral intensity at the Fermi level along with Fermi cut-off at low temperature suggests non-Fermi liquid like behaviour. Many body effects captured within LDA+DMFT reveal increased warping of the Fermi surface with lowering temperature. A simple analysis suggests that the warping dominates the thermal energy induced momentum broadening at low temperature, leading to the 3D electronic behaviour. Our results offer valuable insight in understanding the interplay of dimensionality, electron correlation and thermal energy governing various exotic phenomena in quasi-one-dimensional systems., Comment: to appear in Phys. Rev. B

In recent years, AI red teaming has emerged as a practice for probing the safety and security of generative AI systems. Due to the nascency of the field, there are many open questions about how red teaming operations should be conducted. Based on our experience red teaming over 100 generative AI products at Microsoft, we present our internal threat model ontology and eight main lessons we have learned: 1. Understand what the system can do and where it is applied 2. You don't have to compute gradients to break an AI system 3. AI red teaming is not safety benchmarking 4. Automation can help cover more of the risk landscape 5. The human element of AI red teaming is crucial 6. Responsible AI harms are pervasive but difficult to measure 7. LLMs amplify existing security risks and introduce new ones 8. The work of securing AI systems will never be complete By sharing these insights alongside case studies from our operations, we offer practical recommendations aimed at aligning red teaming efforts with real world risks. We also highlight aspects of AI red teaming that we believe are often misunderstood and discuss open questions for the field to consider.

This work presents an approach to synthesize a Lyapunov-like function to ensure incrementally input-to-state stability ($\delta$-ISS) property for an unknown discrete-time system. To deal with challenges posed by unknown system dynamics, we parameterize the Lyapunov-like function as a neural network, which we train using the data samples collected from the unknown system along with appropriately designed loss functions. We propose a validity condition to test the obtained function and incorporate it into the training framework to ensure provable correctness at the end of the training. Finally, the usefulness of the proposed technique is proved using two case studies: a scalar non-linear dynamical system and a permanent magnet DC motor.