Web agents promise to automate complex browser tasks, but current methods remain brittle -- relying on step-by-step UI interactions and heavy LLM reasoning that break under dynamic layouts and long horizons. Humans, by contrast, exploit website-provided functionality through high-level operations like search, filter, and sort. We introduce WALT (Web Agents that Learn Tools), a framework that reverse-engineers latent website functionality into reusable invocable tools. Rather than hypothesizing ad-hoc skills, WALT exposes robust implementations of automations already designed into websites -- spanning discovery (search, filter, sort), communication (post, comment, upvote), and content management (create, edit, delete). Tools abstract away low-level execution: instead of reasoning about how to click and type, agents simply call search(query) or create(listing). This shifts the computational burden from fragile step-by-step reasoning to reliable tool invocation. On VisualWebArena and WebArena, WALT achieves higher success with fewer steps and less LLM-dependent reasoning, establishing a robust and generalizable paradigm for browser automation.
This study analyzes Walt Whitman's stylistic changes in his phenomenal work Leaves of Grass from a computational perspective and relates findings to standard literary criticism on Whitman. The corpus consists of all 7 editions of Leaves of Grass, ranging from the earliest 1855 edition to the 1891-92 "deathbed" edition. Starting from counting word frequencies, the simplest stylometry technique, we find consistent shifts in word choice. Macro-etymological analysis reveals Whitman's increasing preference for words of specific origins, which is correlated to the increasing lexical complexity in Leaves of Grass. Principal component analysis, an unsupervised learning algorithm, reduces the dimensionality of tf-idf vectors to 2 dimensions, providing a straightforward view of stylistic changes. Finally, sentiment analysis shows the evolution of Whitman's emotional state throughout his writing career.
Identify, Transform, Benchmark, Show Quantum Advantage (ITBQ): Evaluating use cases for quantum computers. We introduce a four-step framework for assessing quantum computing applications -- from identifying relevant industry problems to demonstrating quantum advantage -- addressing steps often overlooked in the literature, such as rigorous benchmarking against classical solutions and the challenge of translating real-world tasks onto quantum hardware. Applying this framework to cases like NMR, multireference chemistry, and radicals reveals both significant opportunities and key barriers on the path to practical advantage. Our results highlight the need for transparent, structured criteria to focus research, guide investment, and accelerate meaningful quantum progress.
We study (strong) first countability of locally solid convergence structures on Archimedean vector lattices. Among other results, we characterise those vector lattices for which relatively unform-, order-, and $σ$-order convergence, respectively, is (strongly) first countable. The implications for the validity of sequential arguments in the contexts of these convergences are pointed out.
Soft continuum arms (SCAs) soft and deformable nature presents challenges in modeling and control due to their infinite degrees of freedom and non-linear behavior. This work introduces a reinforcement learning (RL)-based framework for visual servoing tasks on SCAs with zero-shot sim-to-real transfer capabilities, demonstrated on a single section pneumatic manipulator capable of bending and twisting. The framework decouples kinematics from mechanical properties using an RL kinematic controller for motion planning and a local controller for actuation refinement, leveraging minimal sensing with visual feedback. Trained entirely in simulation, the RL controller achieved a 99.8% success rate. When deployed on hardware, it achieved a 67% success rate in zero-shot sim-to-real transfer, demonstrating robustness and adaptability. This approach offers a scalable solution for SCAs in 3D visual servoing, with potential for further refinement and expanded applications.
Quantum computing has the potential to revolutionize quantum chemistry and material science by offering solutions to complex problems unattainable with classical computers. However, the development of efficient quantum algorithms that are efficient under noisy conditions remains a major challenge. This paper introduces the truncated Variational Hamiltonian Ansatz (tVHA), a novel circuit design for conducting quantum calculations on Noisy Intermediate-Scale Quantum (NISQ) devices. tVHA provides a promising approach for a broad range of applications by utilizing principles from the adiabatic theorem in solid state physics. Our proposed ansatz significantly reduces the parameter count and can decrease circuit size substantially, with a trade-off in accuracy. Thus, tVHA facilitates easier convergence within the variational quantum eigensolver framework compared to state-of-the-art ansätze such as Unitary Coupled Cluster (UCC) and Hardware-Efficient Ansatz (HEA). While this paper concentrates on the practical applications of tVHA in quantum chemistry, demonstrating its suitability for both weakly and strongly correlated systems and its compatibility with active space calculations, its u
Effective and efficient agricultural manipulation and harvesting depend on accurately understanding the current state of the grasp. The agricultural environment presents unique challenges due to its complexity, clutter, and occlusion. Additionally, fruit is physically attached to the plant, requiring precise separation during harvesting. Selecting appropriate sensors and modeling techniques is critical for obtaining reliable feedback and correctly identifying grasp states. This work investigates a set of key sensors, namely inertial measurement units (IMUs), infrared (IR) reflectance, tension, tactile sensors, and RGB cameras, integrated into a compliant gripper to classify grasp states. We evaluate the individual contribution of each sensor and compare the performance of two widely used classification models: Random Forest and Long Short-Term Memory (LSTM) networks. Our results demonstrate that a Random Forest classifier, trained in a controlled lab environment and tested on real cherry tomato plants, achieved 100% accuracy in identifying slip, grasp failure, and successful picks, marking a substantial improvement over baseline performance. Furthermore, we identify a minimal viabl
We present a scalable large language model (LLM)-based system that combines aspect-based sentiment analysis (ABSA) with guided summarization to generate concise and interpretable product review summaries for the Wayfair platform. Our approach first extracts and consolidates aspect-sentiment pairs from individual reviews, selects the most frequent aspects for each product, and samples representative reviews accordingly. These are used to construct structured prompts that guide the LLM to produce summaries grounded in actual customer feedback. We demonstrate the real-world effectiveness of our system through a large-scale online A/B test. Furthermore, we describe our real-time deployment strategy and release a dataset of 11.8 million anonymized customer reviews covering 92,000 products, including extracted aspects and generated summaries, to support future research in aspect-guided review summarization.
The questions of the rarity of the 4.8 GHz formaldehyde masers and the non-detection of 14.5 GHz formaldehyde masers are addressed from a theoretical point of view. The pumping free-free radiation fields were obtained using the photo-ionization code Cloudy to simulate hyper-compact HII regions. Implementation of these free-free radiation fields in solving the rate equations shows that the free-free radiation fields of some hyper-compact HII regions are ineffective to invert the 4.8 GHz transition. Investigation of the variation of the inversion of the 4.8 GHz and 14.5 GHz transitions with radial distance revealed that there are regions where only the 4.8 GHz transition is inverted. It is also shown that the projection of the masing region toward the edge of a hyper-compact may explain the non-detection of 14.5 GHz masers. The attenuation of the 4.8 GHz and 14.5 GHz masers in the molecular envelope is investigated. It is found that attenuation can be significant. Comparison of the velocities of the 4.8 GHz masers with the centre velocities of associated 4.8 GHz absorption features shows that the maser emission lies at the edge of the absorption features. This suggests that the maser
We develop integration theory for integrating functions taking values into a Dedekind complete unital $f$-algebra $\mathbb{L}$ with respect to $\mathbb{L}$-valued measures. We then discuss and prove completeness results of $\mathbb{L}$-valued $L^p$-spaces.
We present new results of a numerical study of the pumping of 4.8 GHz and 14.5 GHz maser of o-Formaldehyde in the presence of a free-free radiation field. It is shown that in the presence of a free-free radiation field inversion of not only the 4.8 GHz transition, but also the 14.5 GHz transition and other doublet state transitions occur. Further results are presented to illustrate how, as a consequence of the pumping scheme, the inversion of the 4.8 GHz and 14.5 GHz transitions respond to the free-free radiation fields associated with HII regions with different emission measures and levels of geometric dilution with respect to the masing region. We also discuss the criticism raised in the past by various authors against the pumping of the 4.8 GHz Formaldehyde masers by a free-free radiation field. It is argued that the rarity of the Formaldehyde masers is not to be ascribed to the pumping scheme but to other factors such as, e.g., the evolution of the associated HII region or the chemical evolution of the star forming region which determines the Formaldehyde abundance or a combination of both.
Inspired by the theories of Kaplansky-Hilbert modules and probability theory in vector lattices, we generalise functional analysis by replacing the scalars $\mathbb{R}$ or $\mathbb{C}$ by a real or complex Dedekind complete unital $f$-algebra $\mathbb{L}$; such an algebra can be represented as a suitable space of continuous functions. We set up the basic theory of $\mathbb{L}$-normed and $\mathbb{L}$-Banach spaces and bounded operators between them, we discuss the $\mathbb{L}$-valued analogues of the classical $\ell^p$-spaces, and we prove the analogue of the Hahn-Banach theorem. We also discuss the basics of the theory of $\mathbb{L}$-Hilbert spaces, including projections onto convex subsets, the Riesz Representation theorem, and representing $\mathbb{L}$-Hilbert spaces as a direct sum of $\ell^2$-spaces.
(Abridged) Molecular lines are commonly detected towards protostellar sources. However, to get a better understanding of the chemistry of these sources we need unbiased molecular surveys over a wide frequency range for as many sources as possible to shed light on the origin of this chemistry, particularly any influence from the external environment. We present results from the PILS-Cygnus survey of ten intermediate- to high-mass protostellar sources in the nearby Cygnus-X complex, through high angular resolution interferometric observations over a wide frequency range. Using the Submillimeter Array (SMA), a spectral line survey of ten sources was performed in the frequency range 329-361 GHz, with an angular resolution of $\sim$1\farcs5, ($\sim$2000 AU, source distance of 1.3 kpc). Spectral modelling was performed to identify molecular emission and determine column densities and excitation temperatures for each source. We detect CH$_3$OH towards nine of the ten sources, CH$_3$OCH$_3$ and CH$_3$OCHO towards three sources, and CH$_3$CN towards four sources. Towards five sources the chemistry is spatially differentiated (different species peak at different positions and are offset from
Nuclear parton distribution functions (nPDFs) can be determined in a global QCD analysis using a wide range of experimental data. In addition to older fixed-target deep inelastic scattering and Drell-Yan (DY) dilepton production data, several analyses from p+Pb collisions at the LHC provide further constraints and extend the kinematic reach of applicable data. Here we present an update of our previous TUJU19 analysis where we now include also electroweak-boson production data recently measured by ATLAS and CMS. For the first time, LHC data are included in a nPDF analysis performed at next-to-next-to-leading order (NNLO) in perturbative QCD. As before, our setup is based on the open-source analysis framework xFitter and we fit our own proton baseline, ensuring a fully consistent setup. We find good agreement with the applied data and that the resulting $χ^2/N_{\mathrm{df}}$ is significantly smaller in case of the NNLO analysis (0.84) compared to our NLO analysis (0.94). Also, we present comparisons between our NNLO calculations and electroweak-boson production data in Pb+Pb collisions from ATLAS and CMS and DY data recently measured by CMS where NNLO corrections are found significan
It is well known that the bidual of $\mathrm C(X)$ for a compact space $X$, supplied with the Arens product, is isometrically isomorphic as a Banach algebra to $\mathrm C(\tilde X)$ for some compact space $\tilde X$. The space $\tilde X$ is unique up to homeomorphism. We establish a similar result for realcompact spaces: The order bidual of $\mathrm C(X)$ for a realcompact space $X$, when supplied with the Arens product, is isomorphic as an $f$-algebra to $\mathrm C(\tilde X)$ for some realcompact space $\tilde X$. The space $\tilde X$ is unique up to homeomorphism.
Heart failure remains a major public health challenge with growing costs. Ejection fraction (EF) is a key metric for the diagnosis and management of heart failure however estimation of EF using echocardiography remains expensive for the healthcare system and subject to intra/inter operator variability. While chest x-rays (CXR) are quick, inexpensive, and require less expertise, they do not provide sufficient information to the human eye to estimate EF. This work explores the efficacy of computer vision techniques to predict reduced EF solely from CXRs. We studied a dataset of 3488 CXRs from the MIMIC CXR-jpg (MCR) dataset. Our work establishes benchmarks using multiple state-of-the-art convolutional neural network architectures. The subsequent analysis shows increasing model sizes from 8M to 23M parameters improved classification performance without overfitting the dataset. We further show how data augmentation techniques such as CXR rotation and random cropping further improves model performance another ~5%. Finally, we conduct an error analysis using saliency maps and Grad-CAMs to better understand the failure modes of convolutional models on this task.
If $K$ is a compact Hausdorff space so that the Banach lattice $C(K)$ is isometrically lattice isomorphic to a dual of some Banach lattice, then $C(K)$ can be decomposed as the $\ell^\infty$-direct sum of the carriers of a maximal singular family of order continuous functionals on $C(K)$. In order to generalise this result to the vector lattice $C(X)$ of continuous, real valued functions on a realcompact space $X$, we consider direct and inverse limits in suitable categories of vector lattices. We develop a duality theory for such limits and apply this theory to show that $C(X)$ is lattice isomorphic to the order dual of some vector lattice $F$ if and only if $C(X)$ can be decomposed as the inverse limit of the carriers of all order continuous functionals on $C(X)$. In fact, we obtain a more general result: A Dedekind complete vector lattice $E$ is perfect if and only if it is lattice isomorphic to the inverse limit of the carriers of a suitable family of order continuous functionals on $E$. A number of other applications are presented, including a decomposition theorem for order dual spaces in terms of spaces of Radon measures.
Graphene nanoelectronics potential was limited by the lack of an intrinsic bandgap[1] and attempts to tailor a bandgap either by quantum confinement or by chemical functionalization failed to produce a semiconductor with a large enough band gap and a sufficient mobility. It is well known that by evaporating silicon from commercial electronics grade silicon carbide crystals an epitaxial graphene layer forms on the surfaces [2]. The first epigraphene layer to form on the silicon terminated face, known as the buffer layer, is insulating. It is chemically bonded to the SiC and spectroscopic measurements [3] have identified semiconducting signatures on the microscopic domains. However, the bonding to the SiC is disordered and the mobilities are small. Here we demonstrate a quasi-equilibrium annealing method that produces macroscopic atomically flat terraces covered with a well ordered epigraphene buffer layer that has a 0.6 eV bandgap. Room temperature mobilities exceed 5000 cm2/Vs which is much larger than silicon and 20 times larger than the phonon scattering imposed limit of current 2D semiconductors. Critical for nanotechnology, its lattice is aligned with the SiC substrate, it is c
Memristive crossbars have become a popular means for realizing unsupervised and supervised learning techniques. In previous neuromorphic architectures with leaky integrate-and-fire neurons, the crossbar itself has been separated from the neuron capacitors to preserve mathematical rigor. In this work, we sought to simplify the design, creating a fast circuit that consumed significantly lower power at a minimal cost of accuracy. We also showed that connecting the neurons directly to the crossbar resulted in a more efficient sparse coding architecture, and alleviated the need to pre-normalize receptive fields. This work provides derivations for the design of such a network, named the Simple Spiking Locally Competitive Algorithm, or SSLCA, as well as CMOS designs and results on the CIFAR and MNIST datasets. Compared to a non-spiking model which scored 33% on CIFAR-10 with a single-layer classifier, this hardware scored 32% accuracy. When used with a state-of-the-art deep learning classifier, the non-spiking model achieved 82% and our simplified, spiking model achieved 80%, while compressing the input data by 92%. Compared to a previously proposed spiking model, our proposed hardware co
A topological space $X$ is called resolvable if it contains a dense subset with dense complement. Using only basic principles, we show that whenever the space $X$ has a resolving subset that can be written as an at most countably infinite union of subsets, in such a way that a given vector lattice of (not necessarily bounded) continuous functions on $X$ separates every point outside the resolving subset from each of its constituents, then the order continuous dual of this lattice is trivial. In order to apply this result in specific cases, we show that several spaces have resolving subsets that can be written as at most countably infinite unions of closed nowhere dense subsets. An appeal to the main result then yields, for example, that, under appropriate conditions, vector lattices of continuous functions on separable spaces, metric spaces, and topological vector spaces have trivial order continuous duals if they separate points and closed nowhere dense subsets. Our results in this direction extend known results in the literature. We also show that, under reasonably mild separation conditions, vector lattices of continuous functions on locally connected $T_1$ Baire spaces without