Search

The development of extremum seeking (ES) has progressed, over the past hundred years, from static maps, to finite-dimensional dynamic systems, to networks of static and dynamic agents. Extensions from ODE dynamics to maps and agents that incorporate delays or even partial differential equations (PDEs) is the next natural step in that progression through ascending research challenges. This paper reviews results on algorithm design and theory of ES for such infinite-dimensional systems. Both hyperbolic and parabolic dynamics are presented: delays or transport equations, heat-dominated equation, wave equations, and reaction-advection-diffusion equations. Nash equilibrium seeking (NES) methods are introduced for noncooperative game scenarios of the model-free kind and then specialized to single-agent optimization. Even heterogeneous PDE games, such as a duopoly with one parabolic and one hyperbolic agent, are considered. Several engineering applications are touched upon for illustration, including flow-traffic control for urban mobility, oil-drilling systems, deep-sea cable-actuated source seeking, additive manufacturing modeled by the Stefan PDE, biological reactors, light-source seeking with flexible-beam structures, and neuromuscular electrical stimulation., Comment: Preprint submitted to IEEE Control Systems Magazine (Special Issue: Into the Second Century of Extremum Seeking Control, 38 pages and 34 figures)

Motivated by a case study of vegetation patterns, we introduce a mixture model with concomitant variables to examine the association between the orientation of vegetation stripes and wind direction. The proposal relies on a novel copula-based bivariate distribution for mixed axial and circular observations and provides a parsimonious and computationally tractable approach to examine the dependence of two environmental variables observed in a complex manifold. The findings suggest that dominant winds shape the orientation of vegetation stripes through a mechanism of neighbouring plants providing wind shelter to downwind individuals.

Natural policy gradient (NPG) is a common policy optimization algorithm and can be viewed as mirror ascent in the space of probabilities. Recently, Vaswani et al. [2021] introduced a policy gradient method that corresponds to mirror ascent in the dual space of logits. We refine this algorithm, removing its need for a normalization across actions and analyze the resulting method (referred to as SPMA). For tabular MDPs, we prove that SPMA with a constant step-size matches the linear convergence of NPG and achieves a faster convergence than constant step-size (accelerated) softmax policy gradient. To handle large state-action spaces, we extend SPMA to use a log-linear policy parameterization. Unlike that for NPG, generalizing SPMA to the linear function approximation (FA) setting does not require compatible function approximation. Unlike MDPO, a practical generalization of NPG, SPMA with linear FA only requires solving convex softmax classification problems. We prove that SPMA achieves linear convergence to the neighbourhood of the optimal value function. We extend SPMA to handle non-linear FA and evaluate its empirical performance on the MuJoCo and Atari benchmarks. Our results demonstrate that SPMA consistently achieves similar or better performance compared to MDPO, PPO and TRPO.

This paper deals with sliding mode control for multivariable polytopic uncertain systems. We provide systematic procedures to design variable structure controllers (VSCs) and unit-vector controllers (UVCs). Based on suitable representations for the closed-loop system, we derive sufficient conditions in the form of linear matrix inequalities (LMIs) to design the robust sliding mode controllers such that the origin of the closed-loop system is globally stable in finite time. Moreover, by noticing that the reaching time depends on the initial condition and the decay rate, we provide convex optimization problems to design robust controllers by considering the minimization of the reaching time associated with a given set of initial conditions. Two examples illustrate the effectiveness of the proposed approaches., Comment: 8 pages, 4 figures

This paper addresses the design and analysis of a multivariable gradient-based stochastic extremum-seeking control method for multi-input systems with arbitrary input delays. The approach accommodates systems with distinct time delays across input channels and achieves local exponential stability of the closed-loop system, guaranteeing convergence to a small neighborhood around the extremum point. By incorporating phase compensation for dither signals and a novel predictor-feedback mechanism with averaging-based estimates of the unknown gradient and Hessian, the proposed method overcomes traditional challenges associated with arbitrary, distinct input delays. Unlike previous work on deterministic multiparameter extremum-seeking with distinct input delays, this stability analysis is achieved without using backstepping transformations, simplifying the predictor design and enabling a more straightforward implementation. Specifically, the direct application of Artstein's reduction approach results in delay- and system-dimension-independent convergence rates, enhancing practical applicability. A numerical example illustrates the robust performance and advantages of the proposed delay-compensated stochastic extremum-seeking method., Comment: 8 pages, 8 figures

As one of the main pillars of quantum technologies, quantum metrology aims to improve measurement precision using techniques from quantum information. The two main strategies to achieve this are the preparation of nonclassical states and the design of optimized measurement observables. We discuss precision limits and optimal strategies in quantum metrology and sensing with a single mode of quantum continuous variables. We focus on the practically most relevant cases of estimating displacements and rotations and provide the sensitivities of the most important classes of states that includes Gaussian states and superpositions of Fock states or coherent states. Fundamental precision limits that are obtained from the quantum Fisher information are compared to the precision of a simple moment-based estimation strategy based on the data obtained from possibly sub-optimal measurement observables, including homodyne, photon number, parity and higher moments. Finally, we summarize some of the main experimental achievements and present emerging platforms for continuous-variable sensing. These results are of particular interest for experiments with quantum light, trapped ions, mechanical oscillators, and microwave resonators., Comment: Review article on quantum metrology with a single continuous-variable system, with a focus on theoretical results. Comments are extremely welcomed!

Several attempts have been made to handle multiple source separation tasks such as speech enhancement, speech separation, sound event separation, music source separation (MSS), or cinematic audio source separation (CASS) with a single model. These models are trained on large-scale data including speech, instruments, or sound events and can often successfully separate a wide range of sources. However, it is still challenging for such models to cover all separation tasks because some of them are contradictory (e.g., musical instruments are separated in MSS while they have to be grouped in CASS). To overcome this issue and support all the major separation tasks, we propose a task-aware unified source separation (TUSS) model. The model uses a variable number of learnable prompts to specify which source to separate, and changes its behavior depending on the given prompts, enabling it to handle all the major separation tasks including contradictory ones. Experimental results demonstrate that the proposed TUSS model successfully handles the five major separation tasks mentioned earlier. We also provide some audio examples, including both synthetic mixtures and real recordings, to demonstrate how flexibly the TUSS model changes its behavior at inference depending on the prompts., Comment: Submitted to ICASSP 2025

Carcinogenesis is a proteiform phenomenon, with tumors emerging in various locations and displaying complex, diverse shapes. At the crucial intersection of research and clinical practice, it demands precise and flexible assessment. However, current biomarkers, such as RECIST 1.1's long and short axis measurements, fall short of capturing this complexity, offering an approximate estimate of tumor burden and a simplistic representation of a more intricate process. Additionally, existing supervised AI models face challenges in addressing the variability in tumor presentations, limiting their clinical utility. These limitations arise from the scarcity of annotations and the models' focus on narrowly defined tasks. To address these challenges, we developed ONCOPILOT, an interactive radiological foundation model trained on approximately 7,500 CT scans covering the whole body, from both normal anatomy and a wide range of oncological cases. ONCOPILOT performs 3D tumor segmentation using visual prompts like point-click and bounding boxes, outperforming state-of-the-art models (e.g., nnUnet) and achieving radiologist-level accuracy in RECIST 1.1 measurements. The key advantage of this foundation model is its ability to surpass state-of-the-art performance while keeping the radiologist in the loop, a capability that previous models could not achieve. When radiologists interactively refine the segmentations, accuracy improves further. ONCOPILOT also accelerates measurement processes and reduces inter-reader variability, facilitating volumetric analysis and unlocking new biomarkers for deeper insights. This AI assistant is expected to enhance the precision of RECIST 1.1 measurements, unlock the potential of volumetric biomarkers, and improve patient stratification and clinical care, while seamlessly integrating into the radiological workflow.

Humans have the ability to learn new tasks by inferring high-level concepts from existing solution, then manipulating these concepts in lieu of the raw data. Can we automate this process by deriving latent semantic structures in a document collection using foundation models? We introduce fPLSA, a foundation-model-based Probabilistic Latent Semantic Analysis (PLSA) method that iteratively clusters and tags document segments based on document-level contexts. These tags can be used to model the structure of given documents and for hierarchical sampling of new texts. Our experiments on story writing, math, and multi-step reasoning datasets demonstrate that fPLSA tags help reconstruct the original texts better than existing tagging methods. Moreover, when used for hierarchical sampling, fPLSA produces more diverse outputs with a higher likelihood of hitting the correct answer than direct sampling and hierarchical sampling with existing tagging methods.

Large language models (LLMs) are increasingly applied to complex reasoning tasks that require executing several complex steps before receiving any reward. Properly assigning credit to these steps is essential for enhancing model performance. Proximal Policy Optimization (PPO), a state-of-the-art reinforcement learning (RL) algorithm used for LLM finetuning, employs value networks to tackle credit assignment. However, value networks face challenges in predicting the expected cumulative rewards accurately in complex reasoning tasks, often leading to high-variance updates and suboptimal performance. In this work, we systematically evaluate the efficacy of value networks and reveal their significant shortcomings in reasoning-heavy LLM tasks, showing that they barely outperform a random baseline when comparing alternative steps. To address this, we propose VinePPO, a straightforward approach that leverages the flexibility of language environments to compute unbiased Monte Carlo-based estimates, bypassing the need for large value networks. Our method consistently outperforms PPO and other RL-free baselines across MATH and GSM8K datasets with fewer gradient updates (up to 9x), less wall-clock time (up to 3.0x). These results emphasize the importance of accurate credit assignment in RL finetuning of LLM and demonstrate VinePPO's potential as a superior alternative.

The goal of text-queried target sound extraction (TSE) is to extract from a mixture a sound source specified with a natural-language caption. While it is preferable to have access to large-scale text-audio pairs to address a variety of text prompts, the limited number of available high-quality text-audio pairs hinders the data scaling. To this end, this work explores how to leverage audio-only data without any captions for the text-queried TSE task to potentially scale up the data amount. A straightforward way to do so is to use a joint audio-text embedding model, such as the contrastive language-audio pre-training (CLAP) model, as a query encoder and train a TSE model using audio embeddings obtained from the ground-truth audio. The TSE model can then accept text queries at inference time by switching to the text encoder. While this approach should work if the audio and text embedding spaces in CLAP were well aligned, in practice, the embeddings have domain-specific information that causes the TSE model to overfit to audio queries. We investigate several methods to avoid overfitting and show that simple embedding-manipulation methods such as dropout can effectively alleviate this issue. Extensive experiments demonstrate that using audio-only data with embedding dropout is as effective as using text captions during training, and audio-only data can be effectively leveraged to improve text-queried TSE models., Comment: Submitted to ICASSP 2025

Single photon emitters (SPEs) are building blocks of quantum technologies. Defect engineering of 2D materials is ideal to fabricate SPEs, wherein spatially deterministic and quality-preserving fabrication methods are critical for integration into quantum devices and cavities. Existing methods use combination of strain and electron irradiation, or ion irradiation, which make fabrication complex, and limited by surrounding lattice damage. Here, we utilise only ultra-low energy electron beam irradiation (5 keV) to create dilute defect density in hBN-encapsulated monolayer MoS2, with ultra-high spatial resolution (< 50 nm, extendable to 10 nm). Cryogenic photoluminescence spectra exhibit sharp defect peaks, following power-law for finite density of single defects, and characteristic Zeeman splitting for MoS2 defect complexes. The sharp peaks have low spectral jitter (< 200 {\mu}eV), and are tuneable with gate-voltage and electron beam energy. Use of low-momentum electron irradiation, ease of processing, and high spatial resolution, will disrupt deterministic creation of high-quality SPEs., Comment: 31 pages, 16 figures, 1 table. Supplementary Material included

This study evaluates the performance of analog-based methodologies to predict the longitudinal velocity in a turbulent flow. The data used comes from hot wire experimental measurements from the Modane wind tunnel. We compared different methods and explored the impact of varying the number of analogs and their sizes on prediction accuracy. We illustrate that the innovation, defined as the difference between the true velocity value and the prediction value, highlights particularly unpredictable events that we directly link with extreme events of the velocity gradients and so to intermittency. This result indicates that while the estimator effectively seizes linear correlations, it fails to fully capture higher-order dependencies. The innovation underscores the presence of intermittency, revealing the limitations of current predictive models and suggesting directions for future improvements in turbulence forecasting.

Giardia lamblia is a flagellate intestinal protozoan with global distribution causing the disease known as giardiasis. This parasite is responsable for 35.1% of outbreaks of diarrhea caused by contaminated water which and mainly affects children in whom it can cause physical and cognitive impairment. In this paper, we consider a model of population dynamics to represent the behavior of Giardia lamblia in vitro, taking into account its mutation characteristic that guarantees to the protozoan resistance to the drug metronidazole. Different from what is found in the literature, it is pursued as the control objective the extermination of the protozoan considering that the parameters of the model are uncertain and only the partial measurement of the state vector is possible. On these assumptions, a control law is designed and the stability of the closed-loop system is rigorously proved. Simulation and experimental results illustrate the benefits of the proposed population control method of Giardia lamblia., Comment: 7 pages, 5 figures, 1 Table

Reverberation as supervision (RAS) is a framework that allows for training monaural speech separation models from multi-channel mixtures in an unsupervised manner. In RAS, models are trained so that sources predicted from a mixture at an input channel can be mapped to reconstruct a mixture at a target channel. However, stable unsupervised training has so far only been achieved in over-determined source-channel conditions, leaving the key determined case unsolved. This work proposes enhanced RAS (ERAS) for solving this problem. Through qualitative analysis, we found that stable training can be achieved by leveraging the loss term to alleviate the frequency-permutation problem. Separation performance is also boosted by adding a novel loss term where separated signals mapped back to their own input mixture are used as pseudo-targets for the signals separated from other channels and mapped to the same channel. Experimental results demonstrate high stability and performance of ERAS., Comment: Accepted to Interspeech 2024

Time-frequency (TF) domain dual-path models achieve high-fidelity speech separation. While some previous state-of-the-art (SoTA) models rely on RNNs, this reliance means they lack the parallelizability, scalability, and versatility of Transformer blocks. Given the wide-ranging success of pure Transformer-based architectures in other fields, in this work we focus on removing the RNN from TF-domain dual-path models, while maintaining SoTA performance. This work presents TF-Locoformer, a Transformer-based model with LOcal-modeling by COnvolution. The model uses feed-forward networks (FFNs) with convolution layers, instead of linear layers, to capture local information, letting the self-attention focus on capturing global patterns. We place two such FFNs before and after self-attention to enhance the local-modeling capability. We also introduce a novel normalization for TF-domain dual-path models. Experiments on separation and enhancement datasets show that the proposed model meets or exceeds SoTA in multiple benchmarks with an RNN-free architecture., Comment: Accepted to IWAENC 2024

Modern artificial intelligence (AI) systems are powered by foundation models. This paper presents a new set of foundation models, called Llama 3. It is a herd of language models that natively support multilinguality, coding, reasoning, and tool usage. Our largest model is a dense Transformer with 405B parameters and a context window of up to 128K tokens. This paper presents an extensive empirical evaluation of Llama 3. We find that Llama 3 delivers comparable quality to leading language models such as GPT-4 on a plethora of tasks. We publicly release Llama 3, including pre-trained and post-trained versions of the 405B parameter language model and our Llama Guard 3 model for input and output safety. The paper also presents the results of experiments in which we integrate image, video, and speech capabilities into Llama 3 via a compositional approach. We observe this approach performs competitively with the state-of-the-art on image, video, and speech recognition tasks. The resulting models are not yet being broadly released as they are still under development.

This paper presents the design of an extremum seeking controller based on sliding modes and cyclic search for real-time optimization of non-linear multivariable dynamic systems. These systems have arbitrary relative degree, compensated by the technique of time-scaling. The resulting approach guarantees global convergence of the system output to a small neighborhood of the optimum point. To corroborate with the theoretical results, numerical simulations are presented considering a system with two inputs and one output, which rapidly converges to the optimal parameters of the objective function., Comment: 6 pages, 5 figures

This letter illustrates the opinion of the molecular dynamics (MD) community on the need to adopt a new FAIR paradigm for the use of molecular simulations. It highlights the necessity of a collaborative effort to create, establish, and sustain a database that allows findability, accessibility, interoperability, and reusability of molecular dynamics simulation data. Such a development would democratize the field and significantly improve the impact of MD simulations on life science research. This will transform our working paradigm, pushing the field to a new frontier. We invite you to support our initiative at the MDDB community (https://mddbr.eu/community/)

We put forward a Quantum Amplitude Estimation algorithm delivering superior performance (lower quantum computational complexity and faster classical computation parts) compared to the approaches available to-date. The algorithm does not relay on the Quantum Fourier Transform and its quantum computational complexity is of order $O(\frac{1}{\varepsilon})$ in terms of the target accuracy $\varepsilon>0$. The $O(\frac{1}{\varepsilon})$ bound on quantum computational complexity is also superior compared to those in the earlier approaches due to smaller constants. Moreover, a much tighter bound is obtained by means of computer-assisted estimates for the expected value of quantum computational complexity. The correctness of the algorithm and the $O(\frac{1}{\varepsilon})$ bound on quantum computational complexity are supported by precise proofs.

While finetuning language models from pairwise preferences has proven remarkably effective, the underspecified nature of natural language presents critical challenges. Direct preference feedback is uninterpretable, difficult to provide where multidimensional criteria may apply, and often inconsistent, either because it is based on incomplete instructions or provided by diverse principals. To address these challenges, we consider the two-step preference modeling procedure that first resolves the under-specification by selecting a context, and then evaluates preference with respect to the chosen context. We decompose reward modeling error according to these two steps, which suggests that supervising context in addition to context-specific preference may be a viable approach to aligning models with diverse human preferences. For this to work, the ability of models to evaluate context-specific preference is critical. To this end, we contribute context-conditioned preference datasets and accompanying experiments that investigate the ability of language models to evaluate context-specific preference. We use our datasets to (1) show that existing preference models benefit from, but fail to fully consider, added context, (2) finetune a context-aware reward model with context-specific performance exceeding that of GPT-4 and Llama 3 70B on tested datasets, and (3) investigate the value of context-aware preference modeling., Comment: NeurIPS 2024. 10 pages (29 with references and appendix)

We study the problem of multimodal physical scene understanding, where an embodied agent needs to find fallen objects by inferring object properties, direction, and distance of an impact sound source. Previous works adopt feed-forward neural networks to directly regress the variables from sound, leading to poor generalization and domain adaptation issues. In this paper, we illustrate that learning a disentangled model of acoustic formation, referred to as disentangled acoustic field (DAF), to capture the sound generation and propagation process, enables the embodied agent to construct a spatial uncertainty map over where the objects may have fallen. We demonstrate that our analysis-by-synthesis framework can jointly infer sound properties by explicitly decomposing and factorizing the latent space of the disentangled model. We further show that the spatial uncertainty map can significantly improve the success rate for the localization of fallen objects by proposing multiple plausible exploration locations.

Single-channel speech dereverberation aims at extracting a dry speech signal from a recording affected by the acoustic reflections in a room. However, most current deep learning-based approaches for speech dereverberation are not interpretable for room acoustics, and can be considered as black-box systems in that regard. In this work, we address this problem by regularizing the training loss using a novel physical coherence loss which encourages the room impulse response (RIR) induced by the dereverberated output of the model to match the acoustic properties of the room in which the signal was recorded. Our investigation demonstrates the preservation of the original dereverberated signal alongside the provision of a more physically coherent RIR.

High kinetic inductance superconductors are gaining increasing interest for the realisation of qubits, amplifiers and detectors. Moreover, thanks to their high impedance, quantum buses made of such materials enable large zero-point fluctuations of the voltage, boosting the coupling rates to spin and charge qubits. However, fully exploiting the potential of disordered or granular superconductors is challenging, as their inductance and, therefore, impedance at high values are difficult to control. Here we have integrated a granular aluminium resonator, having a characteristic impedance exceeding the resistance quantum, with a germanium double quantum dot and demonstrate strong charge-photon coupling with a rate of $g_\text{c}/2\pi= (566 \pm 2)$ MHz. This was achieved due to the realisation of a wireless ohmmeter, which allows \emph{in situ} measurements during film deposition and, therefore, control of the kinetic inductance of granular aluminium films. Reproducible fabrication of circuits with impedances (inductances) exceeding 13 k$\Omega$ (1 nH per square) is now possible. This broadly applicable method opens the path for novel qubits and high-fidelity, long-distance two-qubit gates.

We explore the relation of weak conjugacy in the group of homeomorphisms isotopic to the identity, for surfaces., Comment: 18 pages, 2 figures

Optical properties of multilayer semi-conductor nano-emitters are crucially dependent on the relative energy levelsof their different components. For core/shell quantum dots, the relative energy difference between conduction bandedge of core and shell materials induces, depending on its value, either a confinement of the electron within the coreor a delocalization of its wave function within the whole quantum dot. This results in drastic consequences on theenergy and the oscillator strength of the fundamental transition. Surprisingly, the literature currently lacks a definitivevalue for the energy difference between CdSe and CdS conduction band edges as most of the experimental studiesprovide values corresponding to specific geometries of quantum dots. Here, we develop a full theoretical modelexpressing energy levels considering core/shell interface pressure, ligands and enabling the accurate prediction ofthe bandgap value with the nanocrystal size. It allows to reliably determine the energy difference between theconduction band edge of CdSe and CdS materials, known as the conduction band offset, in such a way that this valuecan later be used to model quantum dots of any geometry. This value is determined using our model and two differentexperimental methods: optical spectroscopy and X-ray photoemission (XPS) experiments.

Thin film interference is integral to modern photonics and optoelectronics, e.g. allowing for precise design of high performance optical filters, efficiency enhancements in photovoltaics and light-emitting devices, as well as the realization of microlasers and high-performance photodetectors. However, interference inevitably leads to a change of spectral characteristics with angle, which is generally undesired and can limit the usefulness of thin-film coatings and devices. Here, we introduce a strategy to overcome this fundamental limit in optics by utilizing and tuning the exciton-polariton modes arising in ultra-strongly coupled microcavities. We demonstrate optical filters with narrow pass bands that shift by less than their half width (<15 nm) even at extreme angles. Our filters cover the entire visible range and surpass comparable metal-dielectric-metal filters in all relevant metrics. By expanding this strategy to strong coupling with the photonic sidebands of dielectric multilayer stacks, we also obtain filters with high extinction ratios and up to 98% peak transmission. Based on these findings, we realize ultrathin and flexible narrowband filter films, monolithically integrate our filters with organic photodiodes, and demonstrate polarization-sensitive polariton filters. These results illustrate how strong coupling provides additional degrees of freedom in thin film optics that will enable a multitude of exciting new applications in micro-optics, sensing, and biophotonics.

Sound event detection is the task of recognizing sounds and determining their extent (onset/offset times) within an audio clip. Existing systems commonly predict sound presence confidence in short time frames. Then, thresholding produces binary frame-level presence decisions, with the extent of individual events determined by merging consecutive positive frames. In this paper, we show that frame-level thresholding degrades the prediction of the event extent by coupling it with the system's sound presence confidence. We propose to decouple the prediction of event extent and confidence by introducing SEBBs, which format each sound event prediction as a tuple of a class type, extent, and overall confidence. We also propose a change-detection-based algorithm to convert legacy frame-level outputs into SEBBs. We find the algorithm significantly improves the performance of DCASE 2023 Challenge systems, boosting the state of the art from .644 to .686 PSDS1., Comment: Accepted for publication at Interspeech 2024

The innovative bladeless design of the AirMultiplier, patented by Dyson in 2009, has generated significant interest due to its unique approach to air circulation that leverages the Coanda effect. In this study, we present a comprehensive parametric analysis of a simplified bladeless fan geometry, with a specific focus on the discharge ratio and generated thrust force for small to medium radii (5 to 200mm) and a range of flow rates spanning from 1 to 100g/s. In addition to the radii and mass flow rates, this study also addresses the effect of the slit nozzle thickness on the discharge ratio and thrust force of the AirMultiplier. By exploring the influence of this additional parameter, we aim to provide a more complete understanding of the performance characteristics of the geometry and to offer insights that could inform the design and optimization of similar bladeless geometries.

This paper deals with the gradient extremum seeking control for static scalar maps with actuators governed by distributed diffusion partial differential equations (PDEs). To achieve the real-time optimization objective, we design a compensation controller for the distributed diffusion PDE via backstepping transformation in infinite dimensions. A further contribution of this paper is the appropriate motion planning design of the so-called probing (or perturbation) signal, which is more involved than in the non-distributed counterpart. Hence, with these two design ingredients, we provide an averaging-based methodology that can be implemented using the gradient and Hessian estimates. Local exponential stability for the closed-loop equilibrium of the average error dynamics is guaranteed through a Lyapunov-based analysis. By employing the averaging theory for infinite-dimensional systems, we prove that the trajectory converges to a small neighborhood surrounding the optimal point. The effectiveness of the proposed extremum seeking controller for distributed diffusion PDEs in cascade of nonlinear maps to be optimized is illustrated by means of numerical simulations., Comment: 8 pages and 7 figures

This paper discusses the design of an extremum seeking controller that relies on a monitoring function for a class of SISO uncertain nonlinear systems characterized by arbitrary and uncertain relative degree. Our demonstration illustrates the feasibility of achieving an arbitrarily small proximity to the desired optimal point through output feedback. The core concept involves integrating a monitoring function with a norm state observer for the unitary relative degree case and its expansion to arbitrary relative degrees by means of the employment of a time-scaling technique. Significantly, our proposed scheme attains the extremum of an unknown nonlinear mapping across the entire domain of initial conditions, ensuring global convergence and stability for the real-time optimization algorithm. Furthermore, we provide tuning rules to ensure convergence to the global maximum in the presence of local extrema. To validate the effectiveness of the proposed approach, we present a numerical example and apply it to a source-seeking problem involving a cart-track linear positioning servomechanism. Notably, the cart lacks the ability to sense its velocity or the source's position, but can detect the source of a light signal of unknown concentration field., Comment: 27 pages, 14 figures

This paper introduces a new method to achieve stable convergence to Nash equilibrium in duopoly noncooperative games. Inspired by the recent fixed-time Nash Equilibrium seeking (NES) as well as prescribed-time extremum seeking (ES) and source seeking schemes, our approach employs a distributed sliding mode control (SMC) scheme, integrating extremum seeking with sinusoidal perturbation signals to estimate the pseudogradients of quadratic payoff functions. Notably, this is the first attempt to address noncooperative games without relying on models, combining classical extremum seeking with relay components instead of proportional control laws. We prove finite-time convergence of the closed-loop average system to Nash equilibrium using stability analysis techniques such as time-scaling, Lyapunov's direct method, and averaging theory for discontinuous systems. Additionally, we quantify the size of residual sets around the Nash equilibrium and validate our theoretical results through simulations., Comment: 8 pages and 2 figures. arXiv admin note: substantial text overlap with arXiv:2404.07287

The growing number of parameter-efficient adaptations of a base large language model (LLM) calls for studying whether we can reuse such trained adapters to improve performance for new tasks. We study how to best build a library of adapters given multi-task data and devise techniques for both zero-shot and supervised task generalization through routing in such library. We benchmark existing approaches to build this library and introduce model-based clustering, MBC, a method that groups tasks based on the similarity of their adapter parameters, indirectly optimizing for transfer across the multi-task dataset. To re-use the library, we present a novel zero-shot routing mechanism, Arrow, which enables dynamic selection of the most relevant adapters for new inputs without the need for retraining. We experiment with several LLMs, such as Phi-2 and Mistral, on a wide array of held-out tasks, verifying that MBC-based adapters and Arrow routing lead to superior generalization to new tasks. We make steps towards creating modular, adaptable LLMs that can match or outperform traditional joint training.

One of the main interests of 2D materials is their ability to be assembled with many degrees of freedom for tuning and manipulating excitonic properties. There is a need to understand how the structure of the interfaces between atomic layers influences exciton properties. Here we use cathodoluminescence (CL) and time-resolved CL experiments to study how excitons interact with the interface between two twisted hexagonal boron nitride (hBN) crystals with various angles. An efficient capture of free excitons by the interface is demonstrated, which leads to a population of long lived and interface-localized (2D) excitons. Temperature dependent experiments indicate that for high twist angles, these excitons localized at the interface further undergo a self-trapping. It consists in a distortion of the lattice around the exciton on which the exciton traps itself. Our results suggest that this exciton-interface interaction causes a broad optical emission of highly twisted hBN-hBN structures around 300 nm (4 eV). Exciton self-trapping is finally discussed as a common feature of sp2 hybridized boron nitride polytypes and nanostructures due to the ionic nature of the B-N bond and their compact excitons., Comment: 15 pages, 13 figures

On-device machine learning (ODML) enables powerful edge applications, but power consumption remains a key challenge for resource-constrained devices. To address this, developers often face a trade-off between model accuracy and power consumption, employing either computationally intensive models on high-power cores or pared-down models on low-power cores. Both approaches typically lead to a compromise in user experience (UX). This work focuses on the use of Gated Compression (GC) layer to enhance ODML model performance while conserving power and maximizing cost-efficiency, especially for always-on use cases. GC layers dynamically regulate data flow by selectively gating activations of neurons within the neural network and effectively filtering out non-essential inputs, which reduces power needs without compromising accuracy, and enables more efficient execution on heterogeneous compute cores. These improvements enhance UX through prolonged battery life, improved device responsiveness, and greater user comfort. In this work, we have integrated GC layers into vision and speech domain models including the transformer-based ViT model. Our experiments demonstrate theoretical power efficiency gains ranging from 158x to 30,000x for always-on scenarios. This substantial improvement empowers ODML applications with enhanced UX benefits., Comment: Initial Submission

Advanced Krylov subspace methods are investigated for the solution of large sparse linear systems arising from stiff adjoint-based aerodynamic shape optimization problems. A special attention is paid to the flexible inner-outer GMRES strategy combined with most relevant preconditioning and deflation techniques. The choice of this specific class of Krylov solvers for challenging problems is based on its outstanding convergence properties. Typically in our implementation the efficiency of the preconditioner is enhanced with a domain decomposition method with overlapping. However, maintaining the performance of the preconditioner may be challenging since scalability and efficiency of a preconditioning technique are properties often antagonistic to each other. In this paper we demonstrate how flexible inner-outer Krylov methods are able to overcome this critical issue. A numerical study is performed considering either a Finite Volume (FV), or a high-order Discontinuous Galerkin (DG) discretization which affect the arithmetic intensity and memory-bandwith of the algebraic operations. We consider test cases of transonic turbulent flows with RANS modelling over the two-dimensional supercritical OAT15A airfoil and the three-dimensional ONERA M6 wing. Benefits in terms of robustness and convergence compared to standard GMRES solvers are obtained. Strong scalability analysis shows satisfactory results. Based on these representative problems a discussion of the recommended numerical practices is proposed., Comment: 22 pages, 10 figures