Search

Segment Anything Model (SAM), a new AI model from Meta AI released in April 2023, is an ambitious tool designed to identify and separate individual objects within a given image through semantic interpretation. The advanced capabilities of SAM are the result of its training with millions of images and masks, and a few days after its release, several researchers began testing the model on medical images to evaluate its performance in this domain. With this perspective in focus -- i.e., optimizing work in the healthcare field -- this work proposes the use of this new technology to evaluate and study chest X-ray images. The approach adopted for this work, with the aim of improving the model's performance for lung segmentation, involved a transfer learning process, specifically the fine-tuning technique. After applying this adjustment, a substantial improvement was observed in the evaluation metrics used to assess SAM's performance compared to the masks provided by the datasets. The results obtained by the model after the adjustments were satisfactory and similar to cutting-edge neural networks, such as U-Net.

We apply evolutionary reinforcement learning to a simulation model in order to identify efficient time-dependent erasure protocols for a physical realization of a one-bit memory by an underdamped mechanical cantilever. We show that these protocols, when applied to the cantilever in the laboratory, are considerably more efficient than our best hand-designed protocols. The learned protocols allow reliable high-speed erasure by minimizing the heating of the memory during the operation. More generally, the combination of methods used here opens the door to the rational design of efficient protocols for a variety of physics applications.

Prognosis after intracranial hemorrhage (ICH) is influenced by a complex interplay between imaging and tabular data. Rapid and reliable prognosis are crucial for effective patient stratification and informed treatment decision-making. In this study, we aim to enhance image-based prognosis by learning a robust feature representation shared between prognosis and the clinical and demographic variables most highly correlated with it. Our approach mimics clinical decision-making by reinforcing the model to learn valuable prognostic data embedded in the image. We propose a 3D multi-task image model to predict prognosis, Glasgow Coma Scale and age, improving accuracy and interpretability. Our method outperforms current state-of-the-art baseline image models, and demonstrates superior performance in ICH prognosis compared to four board-certified neuroradiologists using only CT scans as input. We further validate our model with interpretability saliency maps. Code is available at https://github.com/MiriamCobo/MultitaskLearning_ICH_Prognosis.git., Comment: 16 pages. Accepted at Machine Learning in Medical Imaging Workshop @ MICCAI 2024 (MLMI2024). This is the submitted manuscript with added link to github repo, funding acknowledgements and authors' names and affiliations. No further post submission improvements or corrections were integrated. Final version not published yet

Understanding the mechanical properties of metals at extreme conditions is essential for the advancement of miniaturized technologies. As dimensions decrease, materials will experience higher strain rates at the same applied velocities. Moreover, the interplay effects of strain rates and temperatures are often overlooked and could have critical effects in applications. In this study, for the first time, the rate-dependent and temperature-dependent mechanical response of nickel microparticles have been investigated. The microparticles were obtained by solid-state dewetting of nickel thin films deposited on c-sapphire. They exhibit self-similar shapes with identical sets of planes, facilitating straightforward comparison between particles. This research represents the first in-depth analysis of the mechanical properties of nickel single crystal dewetted microparticles across six orders of magnitude at room temperature and three orders of magnitude at 128 K. Molecular dynamic simulations (MD) were conducted in parallel on particles with the same faceting. In this work, the gap between experiments and simulations has been reduced to over one order of magnitude in size and 3 orders of magnitude in the strain rates. The thermal activation parameter analysis and MD simulations were employed to ascertain whether homogeneous or heterogeneous dislocation nucleation was the dominant mechanism controlling deformation in the particles., Comment: 18 pages (29 in total with supplementary information), 4 figures, 21 Supplementary figures

We report a new scanning nanomechanical noise microscopy platform enabling to both heat and acquire the fluctuations of mechanical nanostructures with nanometric resolution. We use this platform to image the thermally activated nanomechanical dynamics of a model system consisting of a $40\,\mathrm{nm}$ diameter single-defect nanowire, while scanning a localized heat source across its surface. We develop a thermal backaction model, which we use to demonstrate a close connection between the structure of the nanowire, its thermal response, its dissipation and its fluctuations. We notably show that the defect behaves as a single fluctuation hub, whose e-beam excitation yields a far off-equilibrium vibrational state, largely dominated by the quantum fluctuations of the heating source. Our platform is of interest for future quantitative investigation of fundamental nanoscale dynamical phenomena, and appears as a new playground for investigating quantum thermodynamics in the strongly dissipative regime and at room temperature., Comment: 10 pages, 4 figures

We theoretically investigate the thermally induced back-action effects in absorption-sensitive micro- and nanomechanical resonators. We propose a unified approach, enabling to simultaneously address both the effective dynamics and non-equilibrium phononic state, depending on the position of a punctual sensing (and heating) probe at the surface of the mechanical device. We present an analytical solution in terms of green functions for a unidimensional resonator whose thermomechanical deformation profile generally follows that of the mechanical losses. In particular, we find that both the dynamics and the mechanical fluctuations strongly depend on the loss distribution. The effect of the quantum fluctuations of the heat source is also discussed. Our approach provides the first steps towards a thorough, general platform for analyzing thermal back-action effects and their consequences, which may be of significance for future development in ultrasensitive nanomechanical research.

We investigate the thermodynamic properties of an information engine that extracts work from thermal fluctuations, using a mechanical cantilever submitted to electrostatic feedback control. The cantilever's position is continuously measured, and feedback is triggered by the first passage of the system across a fixed threshold. The information $\Delta I$ associated with the feedback is based on the first-passage-time distribution. In this setting, we derive and experimentally verify two distinct fluctuation theorems that involve $\Delta I$ and give a tight bound on the work produced by the engine. Our results extend beyond the specific application to our experiment: we develop a general framework for obtaining fluctuation theorems and work bounds, formulated in terms of probability distributions of protocols rather than underlying measurement outcomes.

We describe an experiment on an underdamped mechanical oscillator used as an information engine. The system is equivalent to an inertial Brownian particle confined in a harmonic potential whose center is controlled by a feedback protocol which measures the particle position at a specific sampling frequency $1/{\tau}$. Several feedback protocols are applied and the power generated by the engine is measured as a function of the oscillator parameters and the sampling frequency. The optimal parameters are then determined. The results are compared to the theoretical predictions and numerical simulations on overdamped systems. We highlight the specific effects of inertia, which can be used to increase the amount of power extracted by the engine. In the regime of large $\tau$, we show that the produced work has a tight bound determined by information theories., Comment: 11 figures

The recording of planetary magnetic fields is often attributed to uniformly-magnetised nanoscopic iron oxides, called single-domain (SD). Yet, the main magnetic constituents of rocks are more complex, non-uniformly magnetised grains in single or multi-vortex states. We know little about their behaviour due to limitations in defining their precise shape and internal magnetic structure. We propose a novel approach combining non-destructive synchrotron-based ptychographic nanotomography with micromagnetic modelling to explore the magnetic stability of remanence-bearing minerals. Applied to a microscopic rock sample, we identified hundreds of nanoscopic grains of magnetite/maghemite with diverse morphologies. For some grains, shape irregularities near the transition from SD to the single-vortex state allow for multiple domain states, some unstable and unable to record the field for significant periods. Additionally, some other grains exhibit temperature-dependent occupancy probabilities, potentially hampering experiments to recover the intensity of past magnetic fields., Comment: 23 pages, 8 Figures, 1 Table, 1 Supplementary File

National Security Letters (NSLs) are similar to administrative subpoenas and can be issued directly by elements of the executive branch without requiring prior approval from a court or grand jury. Importantly, NSLs authorize the imposition of nondisclosure orders (aka "gag orders") on the receiving party. Controversy about potential abuses of this authority has driven a range of legal and policy discussions. To address these concerns, both the public sector and the private sector have sought to document the usage of NSLs in aggregated form. However, each data source is limited in scope, time, and kind. In this paper, we consolidate the available data around NSLs and answer two questions: (1) what can the public effectively learn from the reported data and does this information suffice to assess the NSL usage? (2) how accessible is this data collection? We show that longitudinal trends in the usage of NSLs can be observed. For instance, we find a significant increase in NSL requests for non-US persons and that the policy reforms to decrease the mandated nondisclosure period appear to be effective. The observed trends suggest that the current transparency mechanisms are viable safeguards against the excessive use of NSLs. However, aggregating and normalizing the data requires manual reviewing, parsing, and validating. We even find inconsistencies within and across official data sources. Overall, the laborious data collection process hinders external and internal auditing efforts and demonstrates the need for a unified and more usable dataset for NSLs., Comment: Accepted as short presentation at CSLAW 2024 (not in proceeding): https://computersciencelaw.org/2024/accepted-papers/

We demonstrate experimentally that, applying optimal protocols which drive the system between two equilibrium states characterized by a free energy difference $\Delta F$, we can maximize the probability of performing the transition between the two states with a work $W$ smaller than $\Delta F$. The second law holds only on average, resulting in the inequality $\langle W \rangle \geq \Delta F$. The experiment is performed using an underdamped oscillator evolving in a double-well potential. We show that with a suitable choice of parameters the probability of obtaining trajectories with $W \le \Delta F$ can be larger than 95%. Very fast protocols are a key feature to obtain these results, which are explained in terms of the Jarzynski equality.

Optical beam deflection is a popular method to measure the deformation of micromechanical devices. As it measures mostly a local slope, its sensitivity depends on the location and size of the optical spot. We present a method to evaluate precisely these parameters, using the relative amplitude of the thermal noise induced vibrations. With a case study of a microcantilever, we demonstrate the accuracy of the approach, as well as its ability to fully characterize the sensitivity of the detector and the parameters (mass, stiffness) of the resonator.

The mechanical performance of metallic metamaterials with 3-dimensional solid frames is typically a combination of the geometrical effect ("architecture") and the characteristic size effects of the base material ("microstructure"). In this study, for the first time, the temperature- and rate-dependent mechanical response of copper microlattices has been investigated. The microlattices were fabricated via a localized electrodeposition in liquid (LEL) process which enables high-precision additive manufacturing of metal at the micro-scale. The metal microlattices possess a unique microstructure with micron sized grains that are rich with randomly oriented growth twins and near-ideal nodal connectivity. Importantly, copper microlattices exhibited unique temperature (-150 and 25 degree C) and strain rate (0.001~100 s-1) dependent deformation behavior during in situ micromechanical testing. Systematic compression tests of fully dense copper micropillars, equivalent in diameter and length to the struts of the microlattice at comparable extreme loading conditions, allow us to investigate the intrinsic deformation mechanism of copper. Combined with the post-mortem microstructural analysis, substantial shifts in deformation mechanisms depending on the temperature and strain rate were revealed. On the one hand, at room temperature (25 degree C), dislocation slip based plastic deformation occurs and leads to a localized deformation of the micropillars. On the other hand, at cryogenic temperature (-150 degree C), mechanical twinning occurs and leads to relatively homogeneous deformation of the micropillars. Based on the intrinsic deformation mechanisms of copper, the temperature and strain rate dependent deformation behavior of microlattices could be explained.

Virtual potentials are an elegant, precise and flexible tool to manipulate small systems and explore fundamental questions in stochastic thermodynamics. In particular double-well potentials have applications in information processing, such as the demonstration of Landauer's principle. In this chapter, we detail the implementation of a feedback loop for an underdamped system, in order to build a tunable virtual double-well potential. This feedback behaves as a demon acting on the system depending on the outcome of a continuously running measurement. It can thus modify the energy exchanges with the thermostat and create an out-of-equilibrium state. To create a bi-stable potential, the feedback consists only in switching an external force between two steady values when the measured position crosses a threshold. We show that a small delay of the feedback loop in the switches between the two wells results in a modified velocity distribution. The latter can be interpreted as a cooling of the kinetic temperature of the system. Using a fast digital feedback, we successfully address all experimental issues to create a virtual potential that is statistically indistinguishable from a physical one, with a tunable barrier height and energy step between the two wells., Comment: Proceedings of the "Crossroad of Maxwell Demon" workshop in Toulouse, March 2023, to be published in Volume 15 of the Springer-Nature Series "Advances in Atom and Single Molecule Machines". arXiv admin note: text overlap with arXiv:2201.09870

Colloidal probes are often used in force microscopy when the geometry of the tip-sample interaction should be well controlled. Their calibration requires the understanding of their mechanical response, which is very sensitive to the details of the force sensor consisting of a cantilever and the attached colloid. We present analytical models to describe the dynamics of the cantilever and its load positioned anywhere along its length. The thermal noise calibration of such probes is then studied from a practical point of view, leading to correction coefficients that can be applied in standard force microscope calibration routines. Experimental measurements of resonance frequencies and thermal noise profiles of raw and loaded cantilevers demonstrate the validity of the approach.

The reliability of fast repeated erasures is studied experimentally and theoretically in a 1-bit underdamped memory. The bit is encoded by the position of a micro-mechanical oscillator whose motion is confined in a double well potential. To contain the energetic cost of fast erasures, we use a resonator with high quality factor $Q$: the erasure work $W$ is close to Landauer's bound, even at high speed. The drawback is the rise of the system's temperature $T$ due to a weak coupling to the environment. Repeated erasures without letting the memory thermalize between operations result in a continuous warming, potentially leading to a thermal noise overcoming the barrier between the potential wells. In such case, the reset operation can fail to reach the targeted logical state. The reliability is characterized by the success rate $R^s_i$ after $i$ successive operations. $W$, $T$ and $R^s_i$ are studied experimentally as a function of the erasure speed. Above a velocity threshold, $T$ soars while $R^s_i$ collapses: the reliability of too fast erasures is low. These experimental results are fully justified by two complementary models. We demonstrate that $Q\simeq 10$ is optimal to contain energetic costs and maintain high reliability standards for repeated erasures at any speed.

The NOT operation is a reversible transformation acting on a 1-bit logical state, and should be achievable in a physically reversible manner at no energetic cost. We experimentally demonstrate a bit-flip protocol based on the momentum of an underdamped oscillator confined in a double well potential. The protocol is designed to be reversible in the ideal dissipationless case, and the thermodynamic work required is inversely proportional to the quality factor of the system. Our implementation demonstrates an energy dissipation significantly lower than the minimal cost of information processing in logically irreversible operations. It is moreover performed at high speed: a fully equilibrated final state is reached in only half a period of the oscillator. The results are supported by an analytical model that takes into account the presence of irreversibility. The Letter concludes with a discussion of optimization strategies.

Landauer's principle makes a strong connection between information theory and thermodynamics by stating that erasing a one-bit memory at temperature $T_0$ requires an average energy larger than $W_{LB}=k_BT_0 \ln2$, with $k_B$ Boltzmann's constant. This tiny limit has been saturated in model experiments using quasi-static processes. For faster operations, an overhead proportional to the processing speed and to the memory damping appears. In this article, we show that underdamped systems are a winning strategy to reduce this extra energetic cost. We prove both experimentally and theoretically that, in the limit of vanishing dissipation mechanisms in the memory, the physical system is thermally insulated from its environment during fast erasures, i.e. fast protocols are adiabatic as no heat is exchanged with the bath. Using a fast optimal erasure protocol we also show that these adiabatic processes produce a maximum adiabatic temperature $T_a=2T_0$, and that Landauer's bound for fast erasures in underdamped systems becomes the adiabatic bound: $W_a = k_B T_0$.

Metal deposition with cryogenic cooling is widely utilized in the condensed matter community for producing ultra-thin epitaxial superconducting layers on semiconductors. However, a drawback arises when these films return to room temperature, as they tend to undergo dewetting. This issue is mitigated by capping the films with an amorphous layer. In this study, we examined the impact of different in-situ fabricated caps on the structural characteristics of Sn thin films deposited at 80 K on InSb substrates. Regardless of the type of capping, we observed that the films remained smooth upon returning to room temperature and were epitaxial on InSb in the cubic Sn ($\alpha$-Sn) phase. However, we noted a correlation between alumina capping with an electron beam evaporator and an increased presence of tetragonal Sn ($\beta$-Sn) grains. This suggests that heating from the alumina source may contribute to a partial phase transition in the Sn layer. The existence of the $\beta$-Sn phase induced superconducting behavior of the films by percolation effect. This study highlights the potential for modifying the structural properties of cryogenic Sn thin films through in-situ capping, paving the way for precise control in the production of superconducting Sn films for integration into quantum computing platforms.

Crystal structure, thermodynamic properties, and adiabatic demagnetization refrigeration (ADR) effect in the spin-7/2 triangular antiferromagnet KBaGd(BO3)2 are reported. With the average nearest-neighbor exchange coupling of 44 mK, this compound shows magnetic order below $T_N = 263$\,mK in zero field. The ADR tests reach the temperature of $T_{min} = 122$ mK, more than twice lower than $T_N$, along with the entropy storage capacity of 192 mJK$^{-1}$cm$^{-3}$ and the hold time of more than 8 hours in the PPMS setup, both significantly improved compared to the spin-1/2 Yb$^{3+}$ analog. We argue that KBaGd(BO3)2 shows a balanced interplay of exchange and dipolar couplings that together with structural randomness and geometrical frustration shift $T_{min}$ to well below the ordering temperature $T_N$, therefore facilitating the cooling., Comment: 8 pages, 6 figures, minor changes to the previous version, accepted for publication in Phys. Rev. B

We consider the stability of precipitates formed at grain boundaries (GBs) by radiation-induced segregation in dilute alloys subjected to irradiation. The effects of grain size and misorientation of symmetric-tilt GBs are quantified using phase field modeling. A novel regime is identified where, at long times, GBs are decorated by precipitate patterns that resist coarsening. Maps of the diffusional P\'eclet number indicate that arrested coarsening takes place when solute advection dominates over thermal diffusion right up to the precipitate/matrix interface, overwhelming capillary effects. This contrasts with liquid-solid mixtures where convection only accelerates coarsening., Comment: 29 pages, main text with 3 figures, 1 table, 1 supplemental information document

Transmission-constrained problems in power systems can be cast as polynomial optimization problems whose coefficients vary over time. We consider the complications therein and suggest several approaches. On the example of the alternating-current optimal power flows (ACOPFs), we illustrate one of the approaches in detail. For the time-varying ACOPF, we provide an upper bound for the difference between the optimal cost for a relaxation using the most recent data and the current approximate optimal cost generated by our algorithm. This bound is a function of the properties of the instance and the rate of change of the coefficients over time. Moreover, we also bound the number of floating-point operations to perform between two subsequent updates to ensure a bounded error., Comment: A journal version of Liu et al [arXiv:1710.07119, PSCC 2018] taking into account our recent work [arXiv:2104.05445 and arXiv:2210.08387]

We present an online algorithm for time-varying semidefinite programs (TV-SDPs), based on the tracking of the solution trajectory of a low-rank matrix factorization, also known as the Burer-Monteiro factorization, in a path-following procedure. There, a predictor-corrector algorithm solves a sequence of linearized systems. This requires the introduction of a horizontal space constraint to ensure the local injectivity of the low-rank factorization. The method produces a sequence of approximate solutions for the original TV-SDP problem, for which we show that they stay close to the optimal solution path if properly initialized. Numerical experiments for a time-varying max-cut SDP relaxation demonstrate the computational advantages of the proposed method for tracking TV-SDPs in terms of runtime compared to off-the-shelf interior point methods., Comment: 24 pages, 3 figures

For systems in equilibrium at a temperature $T$, thermal noise and energy damping are related to $T$ through the fluctuation-dissipation theorem (FDT). We study here an extension of the FDT to an out of equilibrium steady state: a microcantilever subject to a constant heat flux. The resulting thermal profile in this spatially extended system interplays with the local energy dissipation field to prescribe the amplitude of mechanical fluctuations. Using three samples with different damping profiles (localized or distributed), we probe this approach and experimentally demonstrate the link between fluctuations and dissipation. The thermal noise can therefore be predicted a priori from the measurement of the dissipation as a function of the maximum temperature of the micro-oscillator.

The parametric representation has been used since a long time for the evaluation of Feynman diagrams. As a dimension independent intermediate representation, it allows a clear description of singularities. Recently, it has become a choice tool for the investigation of the type of transcendent numbers appearing in the evaluation of Feynman diagrams. The inclusion of numerators has however stagnated since the ground work of Nakanishi. I here show how to greatly simplify the formulas through the use of Dodgson identities., Comment: 23 pages, TikZ pictures. Two additional result files, best read with Maple. v2: expanded explanations on some parts. Results and additional files unchanged

The two output signals of quadrature phase interferometers allow to benefit both from the high sensitivity of interferometry (working inside a fringe) and from an extended input range (counting fringes). Their calibration to reach a linear output is traditionally performed using Heydemann's correction, which involves fitting one output versus the other by an ellipse. Here we present two alternative methods based on the linear response of the measurement to a sinusoidal input in time, which enables a direct calibration with an excellent linearity. A ten fold improvement with respect to the usual technique is demonstrated on an optical interferometer measuring the deflection of scanning force microscopy cantilevers.

In most systems, thermal diffusion is intrinsically slow with respect to mechanical relaxation. We devise here a generic approach to accelerate the relaxation of the temperature field of a 1D object, in order to beat the mechanical time scales. This approach is applied to a micro-meter sized silicon cantilever, locally heated by a laser beam. A tailored driving protocol for the laser power is derived to quickly reach the thermal stationary state. The model is implemented experimentally yielding a significant acceleration of the thermal relaxation, up to a factor 30. An excellent agreement with the theoretical predictions is reported. This strategy allows a thermal steady state to be reached significantly faster than the natural mechanical relaxation.

Virtual potentials are a very elegant, precise and flexible tool to manipulate small systems and explore fundamental questions in stochastic thermodynamics. In particular double-well potentials have applications in information processing, such as the demonstration of Landauer's principle. Nevertheless, virtual double-well potentials had never been implemented in underdamped systems. In this article, we detail how to face the experimental challenge of creating a feedback loop for an underdamped system (exploring its potential energy landscape much faster than its over-damped counterpart), in order to build a tunable virtual double-well potential. To properly describe the system behavior in the feedback trap, we express the switching time in the double-well for all barrier heights, combining for the first time Kramer's description, valid at high barriers, with an adjusted model for lower ones. We show that a small hysteresis or delay of the feedback loop in the switches between the two wells results in a modified velocity distribution, interpreted as a cooling of the kinetic temperature of the system. We successfully address all issues to create experimentally a virtual potential that is statistically indistinguishable from a physical one, with a tunable barrier height and energy step between the two wells.

Atomic force microscopy cantilevers are often, intentionally or not, heated at their extremity. We describe a model to compute the resulting temperature field in the cantilever and in the surrounding fluid on a wide temperature range. In air and for common geometries, the heat fluxes in the cantilever and to the environment are of comparable magnitude. We then infer how the fluid-structure interaction is modified due to heating, and predict the induced changes in the dynamics of the system. In particular, we describe how the resonance frequencies of the cantilever shift with a temperature increase due to two competing processes: softening of the cantilever, and decrease of the fluid inertial effects. Our models are illustrated by experiments on a set of cantilevers spanning the relevant geometries to explore the relative importance of both effects.

Math performance is negatively related to math anxiety (MA), though MA may impact certain math skills more than others. We investigated whether the relation between MA and math performance is affected by task features, such as number type (e.g., fractions, whole numbers, percentages), number format (symbolic vs. nonsymbolic), and ratio component size (small vs. large). Across two large-scale studies (combined n = 3,822), the MA-performance relation was strongest for large whole numbers and fractions, and stronger for symbolic than nonsymbolic fractions. The MA-performance relation was also stronger for smaller relative to larger components, and MA relating to specific number types may be a better predictor of performance than general MA for certain tasks. The relation between MA and estimation performance changes depending on task features, which suggests that MA may relate to certain math skills more than others, which may have implications for how people reason with numerical information and may inform future interventions. [This article was published in "Journal of Experimental Psychology: General" v152 n7 p2094-2117 2023.]