Search

Laminar separation bubbles around airfoils lead to the growth of instability waves, which enhances to acoustic scattering at the trailing-edge, forming a feedback loop that produces to tonal noise. To reduce the trailing-edge tonal noise, an array of roughness elements was used over a NACA0012 airfoil, at low angles of attack and moderate Reynolds number. Aeroacoustics and flow visualization experiments were performed for four configurations: a baseline smooth airfoil, two configurations with roughness elements on either airfoil surface (pressure side or suction side), and a fourth configuration with roughness elements at both airfoil surfaces. The roughness elements are made up of a row of spanwise periodically spaced cylinders, which were placed close to the mid-chord position. It is expected the proposed array of roughness elements stabilizes the Kelvin-Helmholtz instability mechanism that is present on the separated shear layer through the induction of streak structures. The separation bubble is assessed using oil flow visualizations, and the cases with roughness elements show the generation of streaks downstream of the array cylinders, disrupting the separation bubble. Acoustic measurements results show a decrease, and in some cases the total suppression, of the tonal noise at Reynolds numbers ranging from 0.6$\times$10$^5$ to 2.5$\times$10$^5$, and angles of attack ranging from 0 to 4 degrees., Comment: 20 pages, 17 figures

In a combined experimental and numerical effort we investigate airfoil tonal noise generation and reduction. The means of noise control are streak generators in form of cylindrical roughness elements. These elements are placed periodically along the span of airfoil at the mid chord streamwise position. Experiments are performed for a wide range of Reynolds number and angle of attack in a companion work (Alva et al. 2024). In the present work we concentrate on numerical investigations for a further investigation of selected cases. We have performed wall-resolved large-eddy simulations for a NACA 0012 airfoil at zero angle of attack and Mach 0.3. Two Reynolds numbers (80,000 and 100,000) have been investigated, showing acoustic results consistent with experiments at the same Reynolds but lower Mach numbers. Roughness elements attenuate tones in the acoustic field and, for the higher Reynolds number, suppress them. Through Fourier decomposition and spectral POD analysis of streamwise velocity data, dominating structures have been identified. Further, the coupling between structures generated by surface roughness and instability modes (Kelvin-Helmholtz) of shear layer has been identified through stability analysis, suggesting stabilisation mechanisms by which the sound generation by the airfoil is reduced by the roughness elements., Comment: 24 pages, 15 figures

We explore a reduced-order model (ROM) of plane Couette flow with a view to performing turbulence control. The ROM is derived through Galerkin projections of the incompressible Navier-Stokes (NS) system onto a basis composed of controllability modes, truncated to a few streamwise and spanwise wavenumbers. Such ROMs were found to reproduce key aspects of nonlinear turbulence dynamics in Couette flow with only a few hundreds degrees of freedom. Here we use the ROM to devise a control strategy. For that, we consider a ROM with an extra forcing term, consisting in a steady body force. The spatial structure of the forcing is given by a linear combination of Stokes modes, optimised using a gradient-descent algorithm in order to minimise the total fluctuation energy. The optimisation is performed at different Reynolds numbers, with the optimal forcing leading to laminarisation of the flow in all cases. The forcing mechanism acts by strongly reducing the shear in a large central portion of the channel. This disrupts the dynamics of large-scale streaks and rolls and hinders the main energy input to the system. When the forcing is active, the flow reaches a new laminar state which is linearly stable and whose linear transient growth is substantially reduced with respect to that of laminar Couette flow. These features prompt the flow to return to the laminar Couette state when the forcing is switched off. Body forces optimised in the ROM are subsequently applied to the full NS system in direct numerical simulations (DNS) for the same flow configurations. The same control mechanisms are observed in the DNS, where laminarisation is also achieved. The present work opens up interesting possibilities for turbulence control. We show that the ROMs provide an effective framework to design turbulence control strategies, despite the high degree of truncation with respect to the full system.

Modern NLP tasks increasingly rely on dense retrieval methods to access up-to-date and relevant contextual information. We are motivated by the premise that retrieval benefits from segments that can vary in size such that a content's semantic independence is better captured. We propose LumberChunker, a method leveraging an LLM to dynamically segment documents, which iteratively prompts the LLM to identify the point within a group of sequential passages where the content begins to shift. To evaluate our method, we introduce GutenQA, a benchmark with 3000 "needle in a haystack" type of question-answer pairs derived from 100 public domain narrative books available on Project Gutenberg. Our experiments show that LumberChunker not only outperforms the most competitive baseline by 7.37% in retrieval performance (DCG@20) but also that, when integrated into a RAG pipeline, LumberChunker proves to be more effective than other chunking methods and competitive baselines, such as the Gemini 1.5M Pro. Our Code and Data are available at https://github.com/joaodsmarques/LumberChunker

We propose a new observable derived from a centrality-dependent scaling of transverse particle spectra. By removing the global scales of total particle number and mean transverse momentum, we isolate the shape of the spectrum. In hydrodynamic simulations, while the multiplicity and mean transverse momentum fluctuate significantly, the scaled spectrum is found to be almost constant even at an event-by-event level and after resonance decays. This universality survives when averaging over events in each centrality bin before scaling. We then investigate the presence of this scaling in experimental data from the ALICE collaboration in Pb-Pb, Xe-Xe, and p-Pb collisions. We find a remarkable universality in the experimentally observed scaled spectra at low transverse momentum, compatible with hydrodynamic predictions. The data show a minor breaking of universality at large transverse momentum and hints of evolution with the system size that are not seen in simulations. Our results motivate further theoretical and experimental investigations of this new observable to bring to light the collective and non-collective behavior encoded in the transverse particle spectrum of different collision systems., Comment: V1: 6 pages, 3 figures

Automated clinical text anonymization has the potential to unlock the widespread sharing of textual health data for secondary usage while assuring patient privacy and safety. Despite the proposal of many complex and theoretically successful anonymization solutions in literature, these techniques remain flawed. As such, clinical institutions are still reluctant to apply them for open access to their data. Recent advances in developing Large Language Models (LLMs) pose a promising opportunity to further the field, given their capability to perform various tasks. This paper proposes six new evaluation metrics tailored to the challenges of generative anonymization with LLMs. Moreover, we present a comparative study of LLM-based methods, testing them against two baseline techniques. Our results establish LLM-based models as a reliable alternative to common approaches, paving the way toward trustworthy anonymization of clinical text.

The impact of the non-linear effects in the QCD dynamics on the observables is directly related to the magnitude of the saturation scale $Q_s$, which is predicted to increase with the energy, rapidity and multiplicity. In this paper, we investigate the $D^0$ meson production in $pp$ collisions at forward rapidities and/or high multiplicities considering the Color Glass Condensate (CGC) formalism and the solutions of the running coupling Balitsky - Kovchegov (BK) equation. The contributions of gluon - and charm - initiated processes are taken into account, and a comparison with the current LHCb data is performed. The impact of an intrinsic charm component in the proton's wave function is also estimated. Predictions for the self-normalized yields of $D^0$ mesons as a function of the multiplicity of coproduced charged hadrons are presented, considering $pp$ collisions at $\sqrt{s} = 13$ TeV and different values of the meson rapidity. A comparison with the predictions for the kaon and isolated photon production is performed. Our results indicate that a future experimental analysis of the $D^0$ meson production at forward rapidities and high multiplicities can be useful to probe the CGC formalism and to disentangle the contribution of initial - and final - state effects., Comment: 14 pages, 13 figures

It has been argued that the speed of sound of the strong interaction at high temperature can be measured using the variation of the mean transverse momentum with the particle multiplicity in ultracentral heavy-ion collisions. We test this correspondence by running hydrodynamic simulations at zero impact parameter with several equations of state, at several colliding energies from 0.2 TeV to 15 TeV per nucleon pair. The correspondence is found to be precise and robust for a smooth, boost-invariant fluid and an ideal detector. We discuss the differences between this ideal setup and an actual experiment. We conclude that the extraction of the speed of sound from data is reliable, and that the main uncertainty comes from our poor knowledge of the distribution of density fluctuations at the early stages of the collision., Comment: v2: 8 pages, 4 figures. Added new section IV - E. Matches the published version

The production of $B$ mesons at forward rapidities is strongly sensitive to the behavior of the gluon and bottom distribution functions for small and large values of the Bjorken - $x$ variable. In this exploratory study, we estimate the cross - section for the $B^{\pm}$ meson production in the kinematic range probed by the LHCb detector and that will be analyzed by the future Forward Physics Facility (FPF) considering the hybrid formalism, the solution of the running coupling Balitsky - Kovchegov equation and distinct descriptions for the bottom distribution function. We assume an ansatz for the intrinsic bottom component in the proton wave function, and estimate its impact on the transverse momentum, rapidity and Feynman - $x$ distributions. Our results indicate that the presence of an intrinsic bottom strongly modifies the magnitude of the cross - section at ultra - forward rapidities ($y \ge 6$), which implies an enhancement of the $B^{\pm}$ production at the FPF. Possible implications on the prompt neutrino flux at ultra-high energies are also briefly discussed., Comment: 8 pages, 5 figures. Version to be published in European Physical Journal C

Reactive flow control has been shown as a promising tool to improve, among other aspects, the aerodynamic characteristics of an aircraft. This paper focuses on the use of reactive flow control to attenuate Tollmien-Schlichting (TS) waves over a wing profile. TS waves are an instability mechanism that is one of the first stages of boundary layer transition to turbulence. The Wiener-Hopf technique was used in this work for the experimental boundary layer control. The approach improves previous wave-cancellation techniques that, by constructing control kernels in the frequency domain, lead to control kernels with a non-causal part, i.e., actuation would need future sensor information to be constructed. In practical applications, it is unfeasible to access this type of information. Ignoring the non-causal part of the kernel leads to suboptimal solutions that might significantly degrade the performance of the controller. The Wiener-Hopf formalism allows us to take into account causality constraints in the formulation of the control problem, leading to an optimal realistic solution and a control kernel that is causal by construction. Moreover, it is possible to construct the control strategy based only on the power and cross-spectra obtained experimentally, in a data-driven approach. The present work shows how to apply experimentally the Wiener-Hopf resolvent-based formalism using signals from a wind tunnel experiment, demonstrating that the Tollmien-Schlichting waves can be effectively attenuated via a Wiener-Hopf based controller, which yielded better results than a typical wave-cancellation approach., Comment: American Physical Society (APS) copyright

We present an experimental study of reactive control of turbulent jets. We target axisymmetric disturbances associated with coherent structures, which are known to underpin the peak sound radiation of turbulent jets. We first consider a forced jet flow case, such that the coherent structures can be amplified above background levels, which makes it easier to detect them by the sensors. We then consider the more challenging case of a natural jet, i.e., without artificial forcing. The control strategy explores linear convective mechanisms in the initial jet region, which justifies application of linear control theory. The control law is constructed in the frequency domain, based on empirically determined transfer functions. The Wiener-Hopf formalism is used to enforce causality, providing an optimal causal solution, thus, preventing the drop in performance that may be observed in flow control applications that use simpler wave-cancellation methods. With this approach, we could improve the control performance of forced turbulent jets compared to results obtained in previous studies, attaining order-of-magnitude attenuation of power spectra of velocity fluctuations. Furthermore, we could obtain substantial levels of attenuation of natural turbulent jets, of about 60% in power spectra for the most amplified frequencies. These results open new directions for the control of turbulent flows.

Large-eddy simulations of a flat-plate boundary layer, without a leading edge, subject to multiple levels of incoming free stream turbulence are considered in the present work. Within an input-output model where non-linear terms of the incompressible Navier-Stokes equations are treated as an external forcing, we manage to separate inputs related to perturbations coming through the intake of the numerical domain, whose evolution represents a linear mechanism, and the volumetric non-linear forcing due to triadic interactions. With these, we perform the full reconstruction of the statistics of the flow, as measured in the simulations, to quantify pairs of wavenumbers and frequencies more affected by either linear or non-linear receptivity mechanisms. Inside the boundary layer, different wavenumbers at near-zero frequency reveal streaky structures. Those that are amplified predominantly via linear interactions with the incoming vorticity occur upstream and display transient growth, while those generated by the non-linear forcing are the most energetic and appear in more downstream positions. The latter feature vortices growing proportionally to the laminar boundary layer thickness, along with a velocity profile that agrees with the optimal amplification obtained by linear transient growth theory. The numerical approach presented is general and could potentially be extended to any simulation for which receptivity to incoming perturbations needs to be assessed.

How can we detect if copyrighted content was used in the training process of a language model, considering that the training data is typically undisclosed? We are motivated by the premise that a language model is likely to identify verbatim excerpts from its training text. We propose DE-COP, a method to determine whether a piece of copyrighted content was included in training. DE-COP's core approach is to probe an LLM with multiple-choice questions, whose options include both verbatim text and their paraphrases. We construct BookTection, a benchmark with excerpts from 165 books published prior and subsequent to a model's training cutoff, along with their paraphrases. Our experiments show that DE-COP surpasses the prior best method by 9.6% in detection performance (AUC) on models with logits available. Moreover, DE-COP also achieves an average accuracy of 72% for detecting suspect books on fully black-box models where prior methods give approximately 4% accuracy. The code and datasets are available at https://github.com/LeiLiLab/DE-COP.

Spectral proper orthogonal decomposition (SPOD) is an increasingly popular modal analysis method in the field of fluid dynamics due to its specific properties: a linear system forced with white noise should have SPOD modes identical to response modes from resolvent analysis. The SPOD, coupled with the Welch method for spectral estimation, may require long time-resolved data sets. In this work, a linearised Ginzburg-Landau model is considered in order to study the method's convergence. SPOD modes of the white-noise forced equation are computed and compared with corresponding response resolvent modes. The quantified error is shown to be related to the time length of Welch blocks (spectral window size) normalised by a convective time. Subsequently, an algorithm based on a temporal data shift is devised to further improve SPOD convergence and is applied to the Ginzburg-Landau system. Next, its efficacy is demonstrated in a numerical database of a boundary layer subject to bypass transition. The proposed approach achieves substantial improvement in mode convergence with smaller spectral window sizes with respect to the standard method. Furthermore, SPOD modes display growing wall-normal and span-wise velocity components along the stream-wise direction, a feature which had not yet been observed and is also predicted by a global resolvent calculation. The shifting algorithm for the SPOD opens the possibility for using the method on datasets with time series of moderate duration, often produced by large simulations., Comment: GitHub repository listed in the acknowledgements

We study generalised quasilinear (GQL) approximations applied to turbulent plane Couette flow. The GQL framework is explored in conjunction with a Galerkin reduced-order model (ROM) recently developed by Cavalieri & Nogueira (Phys. Rev. Fluids 7, L102601, 2022), which considers controllability modes of the linearised Navier-Stokes system as basis functions, representing coherent structures in the flow. The velocity field is decomposed into two groups: one composed by high-controllability modes and the other by lowcontrollability modes. The former group is solved with the full nonlinear equations, whereas the equations for the latter are linearised.We also consider a new GQL framework wherein the linearised equations for the lowcontrollability modes are driven by nonlinear interactions of modes in the first group, which are characterised by large-scale coherent structures. It is shown that GQL-ROMs successfully recover the statistics of the full model with relatively high controllability thresholds and sparser nonlinear operators. Driven GQL-ROMs were found to converge more rapidly than standard GQL approximations, providing accurate description of the statistics with a larger number of linearised modes. This indicates that the forcing of linearised flow structures by large-scale coherent structures is an important feature of turbulence dynamics that should be considered in GQL models. The results presented here reveal that further model reductions are attainable with GQL-ROMs, which can be valuable to extend these models to larger Reynolds numbers.

Heavy-ion collisions, such as Pb-Pb or p-Pb, produce extreme conditions in temperature and density that make the hadronic matter transition to a new state, called quark-gluon plasma (QGP). Simulations of heavy-ion collisions provide a way to improve our understanding of the QGP's properties. These simulations are composed of a hybrid description that results in final observables in agreement with accelerators like LHC and RHIC. However, recent works pointed out that these hydrodynamic simulations can display acausal behavior during the evolution in certain regions, indicating a deviation from a faithful representation of the underlying QCD dynamics. To pursue a better understanding of this problem and its consequences, this work simulated two different collision systems, Pb-Pb and p-Pb at $\sqrt{s_{NN}} = 5.02$ TeV. In this context, our results show that causality violation, even though always present, typically occurs on a small part of the system, quantified by the total energy fraction residing in the acausal region. In addition, the acausal behavior can be reduced with changes in the pre-hydrodynamic factors and the definition of the bulk-viscous relaxation time. Since these aspects are fairly arbitrary in current simulation models, without solid guidance from the underlying theory, it is reasonable to use the disturbing presence of acausal behavior in current simulations to guide improvements towards more realistic modeling. While this work does not solve the acausality problem, it sheds more light on this issue and also proposes a way to solve this problem in simulations of heavy-ion collisions., Comment: 15 pages, 17 figures

The widest spanwise scales in turbulent channel flows are studied through the use of three periodic channel-flow simulations at friction Reynolds number $\mathrm{Re}_{\tau}=550$. The length and height of the channels are the same in all cases ($L_x/h=8\pi$ and $L_y/h=2$ respectively), while the width is progressively doubled: $L_z/h = \{4\pi, 8\pi, 16\pi\}$. The effects of increasing the domain can not be determined with statistical significance in our simulations, since the difference in the statistics between the simulations is of the same order as the errors of convergence. A channel flow similar to the smaller one ($\textit{J. Fluid Mech.}$, vol. 500, 2004, pp. 135--144), which was averaged over a very long time, was used as a reference. The one-dimensional spanwise spectrum of the streamwise velocity is computed with the aim of assessing the domain-size effect on the widest scales. Our results indicate that $90\%$ of the total streamwise energetic fluctuations is recovered without a significant influence of the size of the domain. The remaining $10\%$ of the energy reflects that the widest scales in the outer layer are the ones most significantly affected by the spanwise length of the domain. The power-spectral density for $k_z = 0$ remains constant even if the size of the domain in the spanwise direction is increased up to 4 times the standard spanwise length, indicating that wide, spanwise coherent structures are not an artifact of domain truncation.

Matching addresses is a critical task for companies and post offices involved in the processing and delivery of packages. The ramifications of incorrectly delivering a package to the wrong recipient are numerous, ranging from harm to the company's reputation to economic and environmental costs. This research introduces a deep learning-based model designed to increase the efficiency of address matching for Portuguese addresses. The model comprises two parts: (i) a bi-encoder, which is fine-tuned to create meaningful embeddings of Portuguese postal addresses, utilized to retrieve the top 10 likely matches of the un-normalized target address from a normalized database, and (ii) a cross-encoder, which is fine-tuned to accurately rerank the 10 addresses obtained by the bi-encoder. The model has been tested on a real-case scenario of Portuguese addresses and exhibits a high degree of accuracy, exceeding 95% at the door level. When utilized with GPU computations, the inference speed is about 4.5 times quicker than other traditional approaches such as BM25. An implementation of this system in a real-world scenario would substantially increase the effectiveness of the distribution process. Such an implementation is currently under investigation., Comment: To be published in the 22nd EPIA Conference on Artificial Intelligence, EPIA 2023, Faial Island - Azores, Portugal, 5-8 September 2023, Proceedings

Plane Couette flow at Re=1200 (based on the channel half-height and half the velocity difference between the top and bottom plates) is investigated with a spatial domain designed to retain only two spanwise integral length scales. In this system, the computation of invariant solutions that are physically representative of the turbulent state has been understood to be challenging. To address this challenge, our approach is to employ an accurate reduced-order model with 600 degrees of freedom (Cavalieri & Nogueira, {Phys. Rev. Fluids}, vol. 7, 2022, L102601). Using the two-scale energy budget and the temporal cross-correlation of key observables, it is first demonstrated that the model contains most of the multi-scale physical processes identified recently (Doohan et al., J. Fluid Mech., vol. 913, 2021, A8): i.e. the large- and small-scale self-sustaining processes, the energy cascade for turbulent dissipation, and an energy-cascade mediated small-scale production mechanism. Invariant solutions of the reduced-order model are subsequently computed, including 96 equilibria and 43 periodic orbits. It is found that none of the computed equilibrium solutions are able to reproduce an accurate energy balance associated with the multi-scale dynamics of turbulent state. Incorporation of unsteadiness into invariant solutions is seen to be essential for a sensible description of the multi-scale turbulent dynamics and the related energetics, at least in this type of flow, as periodic orbits with a sufficiently long period are mainly able to describe the complex spatiotemporal dynamics associated with the known multi-scale phenomena., Comment: In press in the Journal of Fluid Mechanics

Ultrarelativistic isobar collisions have been proposed as a useful tool to investigate nuclear structure. These systems are not created in equilibrium, rather undergo a pre-thermalization stage. In this stage, some of the initial structure information may be lost and additional effects introduced. The objective of this paper is to study this possibility in the extreme case of a ``free-streaming'' pre-equilibrium stage. We do this by computing estimators for ratios of various measured (or measurable) quantities (elliptic and triangular flows, mean transverse momentum and associated cumulants, correlators between elliptic or triangular flows and mean transverse momentum, symmetric cumulant and two-plane correlator) and study their sensitivity to the duration of the free-streaming stage. We find that the correlators between elliptic or triangular flows and mean transverse momentum, the so-called $\rho_2$ and $\rho_3$, are indeed sensitive to the duration of the free-streaming stage and that the normalized symmetric cumulant, $\varepsilon NSC(2,3)$ might also depend on this duration., Comment: v3: 27 pages, 12 figures

The production of isolated photons in high multiplicity events is investigated considering the Color Glass Condensate (CGC) formalism. The associated cross-section for proton - proton collisions is estimated considering three distinct solutions of the Balitsky - Kovchegov (BK) equation and predictions for the normalized photon yield as a function of the multiplicities of co - produced charged particles are presented. We predict the increasing of the yield with the multiplicity, with the slope being smaller for larger rapidities. As the isolated photon production is not affected by the fragmentation process, a future experimental investigation of this process in current high energy hadronic colliders is ideal to test the treatment of high multiplicity events using the CGC formalism, previously applied only for the production of hadronic final states., Comment: 9 pages, 6 figures. Improved version published in European Physical Journal A

Coherent structures are found in many different turbulent flows and are known to drive self-sustaining processes in wall turbulence. Identifying the triadic interactions which generate coherent structures can provide insights beyond what is possible with linearized models. There are infinite possible interactions that may generate a given structure. Thus, a method to systematically study those, ranking them in terms of their contribution to the structure of interest, is essential. We here use the resolvent-based extended spectral proper orthogonal decomposition (RESPOD) approach (Karban, U. et al. 2022 Self-similar mechanisms in wall turbulence studied using resolvent analysis. Journal of Fluid Mechanics 969, A36) to rank the triadic interactions which give rise to the dominant coherent structures in minimal Couette flows at Reynolds number 400 and 1000. Our analysis identifies that six triadic interactions dominate the most energetic coherent structure, revealing the capability of the methodology to identify and rank nonlinear interactions. The approach can be used to analyse the energy exchange in turbulent flows and may guide the construction of reduced-order models based on the interplay between different flow modes. Based on this framework, we introduce a modelling strategy where the interactions increasing or reducing the energy of a given mode are grouped as sources and sinks, respectively. The effect of the sinks is embedded in the resolvent operator by using an eddy viscosity model. The sources are used for driving this modified resolvent operator and are shown to yield accurate flow predictions at zero frequency. We discuss that this strategy can be useful when analysing nonlinear interactions or modelling forcing at high-Reynolds-number flows., Comment: 25 pages, 20 figures

We extend our previous investigation of the effects of pre-hydrodynamic evolution on final-state observables in heavy-ion collisions to smaller systems. We use a state-of-the-art hybrid model for the numerical simulations with optimal parameters obtained from a previous Bayesian study. By studying p-Pb collisions, we find that the effects due to the assumption of a conformal evolution in the pre-hydrodynamical stage are even more important in small systems. We also show that this effect depends on the time duration of the pre-equilibrium stage, which is further enhanced in small systems. Finally, we show that the recent proposal of a free-streaming with subluminal velocity for the pre-equilibrium stage, thus effectively breaking conformal invariance, can alleviate the contamination of final state observables. Our study further reinforces the need for moving beyond conformal approaches in pre-equilibrium dynamics modeling, especially when extracting transport coefficients from hybrid models in the high-precision era of heavy-ion collisions., Comment: 15 pages, 14 figures

The production of the $K_S^0$ meson in high multiplicity $pp$ collisions at $\sqrt{s} =$ 13 TeV is investigated considering the hybrid formalism and the solution of the running coupling Balitsky - Kovchegov equation. The associated cross section is estimated and compared with the experimental data for the transverse momentum spectrum. Moreover, we analyze the self-normalized yields of $K_S^0$ mesons as a function of the multiplicity of coproduced charged hadrons and demonstrate that a steep increasing is theoretically predicted. A comparison with the ALICE data is presented considering two distinct solutions of the BK equation., Comment: 8 pages, 5 figures. Improved version to be published in Physical Review C

Reduced-order models were derived for plane Couette flow using Galerkin projection, with orthonormal basis functions taken as the leading controllability modes of the linearised Navier-Stokes system for a few low wavenumbers. Resulting Galerkin systems comprise ordinary differential equations, with a number of degrees of freedom ranging from 144 to 600, which may be integrated to large times without sign of numerical instability. The reduced-order models so obtained are also found to match statistics of direct numerical simulations at Reynolds number 500 and 1200 with reasonable accuracy, despite a truncation of orders of magnitude in the degrees of freedom of the system. The present models offer thus an interesting compromise between simplicity and accuracy in a canonical wall-bounded flow, with relatively few modes representing coherent structures in the flow and their dominant dynamics., Comment: 10 pages, 4 figures

Imagery collected from outdoor visual environments is often degraded due to the presence of dense smoke or haze. A key challenge for research in scene understanding in these degraded visual environments (DVE) is the lack of representative benchmark datasets. These datasets are required to evaluate state-of-the-art vision algorithms (e.g., detection and tracking) in degraded settings. In this paper, we address some of these limitations by introducing the first realistic hazy image benchmark, from both aerial and ground view, with paired haze-free images, and in-situ haze density measurements. This dataset was produced in a controlled environment with professional smoke generating machines that covered the entire scene, and consists of images captured from the perspective of both an unmanned aerial vehicle (UAV) and an unmanned ground vehicle (UGV). We also evaluate a set of representative state-of-the-art dehazing approaches as well as object detectors on the dataset. The full dataset presented in this paper, including the ground truth object classification bounding boxes and haze density measurements, is provided for the community to evaluate their algorithms at: https://a2i2-archangel.vision. A subset of this dataset has been used for the ``Object Detection in Haze'' Track of CVPR UG2 2022 challenge at http://cvpr2022.ug2challenge.org/track1.html., Comment: This paper has been ACCEPTED for publication as a REGULAR paper in the IEEE Transactions on Image Processing (TIP)

The application of control tools to complex flows frequently requires approximations, such as reduced-order models and/or simplified forcing assumptions, where these may be considered low-rank or defined in terms of simplified statistics (e.g. white noise). In this work, we propose a resolvent-based control methodology with causality imposed via a Wiener-Hopf formalism. Linear optimal causal estimation and control laws are obtained directly from full-rank, globally stable systems with arbitrary disturbance statistics, circumventing many drawbacks of alternative methods. We use efficient, matrix-free methods to construct the matrix Wiener-Hopf problem, and we implement a tailored method to solve the problem numerically. The approach naturally handles forcing terms with space-time colour; it allows inexpensive parametric investigation of sensor/actuator placement in scenarios where disturbances/targets are low rank; it is directly applicable to complex flows disturbed by high-rank forcing; it has lower cost in comparison to standard methods; it can be used in scenarios where an adjoint solver is not available; or it can be based exclusively on experimental data. The method is particularly well-suited for the control of amplifier flows, for which optimal control approaches are typically robust. Validation of the approach is performed using the linearized Ginzburg-Landau equation. Flow over a backward-facing step perturbed by high-rank forcing is then considered. Sensor and actuator placement are investigated for this case, and we show that while the flow response downstream of the step is dominated by the Kelvin-Helmholtz mechanism, it has a complex, high-rank receptivity to incoming upstream perturbations, requiring multiple sensors for control., Comment: 46 pages, 20 figures. Submitted to the Journal of Fluid Mechanics