Search

Understanding is a crucial yet elusive concept in artificial intelligence (AI). This work proposes a framework for analyzing understanding based on the notion of composability. Given any subject (e.g., a person or an AI), we suggest characterizing its understanding of an object in terms of its ability to process (compose) relevant inputs into satisfactory outputs from the perspective of a verifier. This highly universal framework can readily apply to non-human subjects, such as AIs, non-human animals, and institutions. Further, we propose methods for analyzing the inputs that enhance output quality in compositions, which we call catalysts. We show how the structure of a subject can be revealed by analyzing its components that act as catalysts and argue that a subject's learning ability can be regarded as its ability to compose inputs into its inner catalysts. Finally we examine the importance of learning ability for AIs to attain general intelligence. Our analysis indicates that models capable of generating outputs that can function as their own catalysts, such as language models, establish a foundation for potentially overcoming existing limitations in AI understanding., Comment: 13 pages, 3 figures

Molecular discovery, when formulated as an optimization problem, presents significant computational challenges because optimization objectives can be non-differentiable. Evolutionary Algorithms (EAs), often used to optimize black-box objectives in molecular discovery, traverse chemical space by performing random mutations and crossovers, leading to a large number of expensive objective evaluations. In this work, we ameliorate this shortcoming by incorporating chemistry-aware Large Language Models (LLMs) into EAs. Namely, we redesign crossover and mutation operations in EAs using LLMs trained on large corpora of chemical information. We perform extensive empirical studies on both commercial and open-source models on multiple tasks involving property optimization, molecular rediscovery, and structure-based drug design, demonstrating that the joint usage of LLMs with EAs yields superior performance over all baseline models across single- and multi-objective settings. We demonstrate that our algorithm improves both the quality of the final solution and convergence speed, thereby reducing the number of required objective evaluations. Our code is available at http://github.com/zoom-wang112358/MOLLEO

The industrial manufacturing of chemicals consumes a significant amount of energy and raw materials. In principle, the development of new catalysts could greatly improve the efficiency of chemical production. However, the discovery of viable catalysts can be exceedingly challenging because it is difficult to know the efficacy of a candidate without experimentally synthesizing and characterizing it. This study explores the feasibility of using fault-tolerant quantum computers to accelerate the discovery of homogeneous catalysts for nitrogen fixation, an industrially important chemical process. It introduces a set of ground-state energy estimation problems representative of calculations needed for the discovery of homogeneous catalysts and analyzes them on three dimensions: economic utility, classical hardness, and quantum resource requirements. For the highest utility problem considered, two steps of a catalytic cycle for the generation of cyanate anion from dinitrogen, the economic utility of running these computations is estimated to be $200,000, and the required runtime for double-factorized phase estimation on a fault-tolerant superconducting device is estimated under conservative assumptions to be 139,000 QPU-hours. The computational cost of an equivalent DMRG calculation is estimated to be about 400,000 CPU-hours. These results suggest that, with continued development, it will be feasible for fault-tolerant quantum computers to accelerate the discovery of homogeneous catalysts., Comment: 27 pages, 11 tables, 8 figures plus appendix

We present a set of 1194 Type-1 Rosseland-mean opacity tables for four different metallicity mixtures. These new Los Alamos OPLIB atomic radiative opacity tables are an order of magnitude larger in number than any previous opacity table release, and span regimes where previous opacity tables have not existed. For example, the new set of opacity tables expands the metallicity range to $Z$\,=\,10$^{-6}$ to $Z$\,=\,0.2 which allows improved accuracy of opacities at low and high metallicity, increases the table density in the metallicity range $Z$\,=\,10$^{-4}$ to $Z$\,=\,0.1 to enhance the accuracy of opacities drawn from interpolations across neighboring metallicities, and adds entries for hydrogen mass fractions between $X$\,=\,0 and $X$\,=\,0.1 including $X$\,=\,$10^{-2}, 10^{-3}, 10^{-4}, 10^{-5}, 10^{-6}$ that can improve stellar models of hydrogen deficient stars. We implement these new OPLIB radiative opacity tables in \MESA, and find that calibrated solar models agree broadly with previously published helioseismic and solar neutrino results. We find differences between using the new 1194 OPLIB opacity tables and the 126 OPAL opacity tables range from $\approx$\,20--80\% across individual chemical mixtures, up to $\approx$\,8\% and $\approx$\,15\% at the bottom and top of the solar convection zone respectively, and $\approx$\,7\% in the solar core. We also find differences between standard solar models using different opacity table sources that are on par with altering the initial abundance mixture. We conclude that this new, open-access set of OPLIB opacity tables does not solve the solar modeling problem, and suggest the investigation of physical mechanisms other than the atomic radiative opacity., Comment: 28 pages, 12 (13) figures. Accepted for Publication in ApJ

Stellar spin down is a critical yet poorly understood component of stellar evolution. In particular, results from the Kepler Mission imply that mature age, solar-type stars have inefficient magnetic braking, resulting in a stalled spin down rate. However, a large number of precise asteroseismic ages are needed for mature ($\geq$ 3Gyr) stars in order to probe the regime where traditional and stalled spin-down models differ. In this paper, we present a new asteroseismic benchmark star for gyrochronology discovered using reprocessed Kepler short cadence data. KIC 11029516 (Papayu) is a bright ($K_{p}$ = 9.6 mag) solar-type star with well-measured rotation period (21.1$\pm$0.8 days) from spot modulation using 4 years of Kepler long cadence data. We combine asteroseismology and spectroscopy to obtain $T_{eff}=5888\pm100$ K, $\rm{[Fe/H]} = 0.30 \pm 0.06\,$ dex, $M = 1.24 \pm 0.05 M_{\odot}$, $R = 1.34 \pm 0.02 R_{\odot}$ and age of 4.0 $\pm$ 0.4 Gyr, making Papayu one of the most similar stars to the Sun in terms of temperature and radius with an asteroseismic age and a rotation period measured from spot modulation. We find that Papayu sits at the transition of where traditional and weakened spin-down models diverge. A comparison with stars of similar zero-age main-sequence temperatures supports previous findings that weakened spin-down models are required to explain the ages and rotation periods of old solar-type stars.

Explainable Artificial Intelligence has gained significant attention due to the widespread use of complex deep learning models in high-stake domains such as medicine, finance, and autonomous cars. However, different explanations often present different aspects of the model's behavior. In this research manuscript, we explore the potential of ensembling explanations generated by deep classification models using convolutional model. Through experimentation and analysis, we aim to investigate the implications of combining explanations to uncover a more coherent and reliable patterns of the model's behavior, leading to the possibility of evaluating the representation learned by the model. With our method, we can uncover problems of under-representation of images in a certain class. Moreover, we discuss other side benefits like features' reduction by replacing the original image with its explanations resulting in the removal of some sensitive information. Through the use of carefully selected evaluation metrics from the Quantus library, we demonstrated the method's superior performance in terms of Localisation and Faithfulness, compared to individual explanations., Comment: accepted at 2nd World Conference on eXplainable Artificial Intelligence

Robust and highly accurate lung segmentation in X-rays is crucial in medical imaging. This study evaluates deep learning solutions for this task, ranking existing methods and analyzing their performance under diverse image modifications. Out of 61 analyzed papers, only nine offered implementation or pre-trained models, enabling assessment of three prominent methods: Lung VAE, TransResUNet, and CE-Net. The analysis revealed that CE-Net performs best, demonstrating the highest values in dice similarity coefficient and intersection over union metric., Comment: published at the Polish Conference on Artificial Intelligence (PP-RAI), 2024

As applications of machine learning proliferate, innovative algorithms inspired by specific real-world challenges have become increasingly important. Such work offers the potential for significant impact not merely in domains of application but also in machine learning itself. In this paper, we describe the paradigm of application-driven research in machine learning, contrasting it with the more standard paradigm of methods-driven research. We illustrate the benefits of application-driven machine learning and how this approach can productively synergize with methods-driven work. Despite these benefits, we find that reviewing, hiring, and teaching practices in machine learning often hold back application-driven innovation. We outline how these processes may be improved., Comment: 12 pages, 3 figures

Visual recognition of materials and their states is essential for understanding the physical world, from identifying wet regions on surfaces or stains on fabrics to detecting infected areas on plants or minerals in rocks. Collecting data that captures this vast variability is complex due to the scattered and gradual nature of material states. Manually annotating real-world images is constrained by cost and precision, while synthetic data, although accurate and inexpensive, lacks real-world diversity. This work aims to bridge this gap by infusing patterns automatically extracted from real-world images into synthetic data. Hence, patterns collected from natural images are used to generate and map materials into synthetic scenes. This unsupervised approach captures the complexity of the real world while maintaining the precision and scalability of synthetic data. We also present the first comprehensive benchmark for zero-shot material state segmentation, utilizing real-world images across a diverse range of domains, including food, soils, construction, plants, liquids, and more, each appears in various states such as wet, dry, infected, cooked, burned, and many others. The annotation includes partial similarity between regions with similar but not identical materials and hard segmentation of only identical material states. This benchmark eluded top foundation models, exposing the limitations of existing data collection methods. Meanwhile, nets trained on the infused data performed significantly better on this and related tasks. The dataset, code, and trained model are available. We also share 300,000 extracted textures and SVBRDF/PBR materials to facilitate future datasets generation.

We present a generative diffusion model specifically tailored to the discovery of surface structures. The generative model takes into account substrate registry and periodicity by including masked atoms and $z$-directional confinement. Using a rotational equivariant neural network architecture, we design a method that trains a denoiser-network for diffusion alongside a force-field for guided sampling of low-energy surface phases. An effective data-augmentation scheme for training the denoiser-network is proposed to scale generation far beyond structure sizes represented in the training data. We showcase the generative model by investigating multiple surface systems and propose an atomistic structure model for a previously unknown silver-oxide domain-boundary of unprecedented size.

Generative machine learning methods such as large-language models are revolutionizing the creation of text and images. While these models are powerful they also harness a large amount of computational resources. The transformer is a key component in large language models that aims to generate a suitable completion of a given partial sequence. In this work, we investigate transformer architectures under the lens of fault-tolerant quantum computing. The input model is one where trained weight matrices are given as block encodings and we construct the query, key, and value matrices for the transformer. We show how to prepare a block encoding of the self-attention matrix, with a new subroutine for the row-wise application of the softmax function. In addition, we combine quantum subroutines to construct important building blocks in the transformer, the residual connection and layer normalization, and the feed-forward neural network. Our subroutines prepare an amplitude encoding of the transformer output, which can be measured to obtain a prediction. Based on common open-source large-language models, we provide insights into the behavior of important parameters determining the run time of the quantum algorithm. We discuss the potential and challenges for obtaining a quantum advantage., Comment: 31 pages, 4 figures, 2 tables, comments are welcome

The discovery of small molecules with therapeutic potential is a long-standing challenge in chemistry and biology. Researchers have increasingly leveraged novel computational techniques to streamline the drug development process to increase hit rates and reduce the costs associated with bringing a drug to market. To this end, we introduce a quantum-classical generative model that seamlessly integrates the computational power of quantum algorithms trained on a 16-qubit IBM quantum computer with the established reliability of classical methods for designing small molecules. Our hybrid generative model was applied to designing new KRAS inhibitors, a crucial target in cancer therapy. We synthesized 15 promising molecules during our investigation and subjected them to experimental testing to assess their ability to engage with the target. Notably, among these candidates, two molecules, ISM061-018-2 and ISM061-22, each featuring unique scaffolds, stood out by demonstrating effective engagement with KRAS. ISM061-018-2 was identified as a broad-spectrum KRAS inhibitor, exhibiting a binding affinity to KRAS-G12D at $1.4 \mu M$. Concurrently, ISM061-22 exhibited specific mutant selectivity, displaying heightened activity against KRAS G12R and Q61H mutants. To our knowledge, this work shows for the first time the use of a quantum-generative model to yield experimentally confirmed biological hits, showcasing the practical potential of quantum-assisted drug discovery to produce viable therapeutics. Moreover, our findings reveal that the efficacy of distribution learning correlates with the number of qubits utilized, underlining the scalability potential of quantum computing resources. Overall, we anticipate our results to be a stepping stone towards developing more advanced quantum generative models in drug discovery.

Automation is one of the cornerstones of contemporary material discovery. Bayesian optimization (BO) is an essential part of such workflows, enabling scientists to leverage prior domain knowledge into efficient exploration of a large molecular space. While such prior knowledge can take many forms, there has been significant fanfare around the ancillary scientific knowledge encapsulated in large language models (LLMs). However, existing work thus far has only explored LLMs for heuristic materials searches. Indeed, recent work obtains the uncertainty estimate -- an integral part of BO -- from point-estimated, non-Bayesian LLMs. In this work, we study the question of whether LLMs are actually useful to accelerate principled Bayesian optimization in the molecular space. We take a sober, dispassionate stance in answering this question. This is done by carefully (i) viewing LLMs as fixed feature extractors for standard but principled BO surrogate models and by (ii) leveraging parameter-efficient finetuning methods and Bayesian neural networks to obtain the posterior of the LLM surrogate. Our extensive experiments with real-world chemistry problems show that LLMs can be useful for BO over molecules, but only if they have been pretrained or finetuned with domain-specific data., Comment: ICML 2024. Code: https://github.com/wiseodd/lapeft-bayesopt

This paper presents a comprehensive comparative analysis of explainable artificial intelligence (XAI) ensembling methods. Our research brings three significant contributions. Firstly, we introduce a novel ensembling method, NormEnsembleXAI, that leverages minimum, maximum, and average functions in conjunction with normalization techniques to enhance interpretability. Secondly, we offer insights into the strengths and weaknesses of XAI ensemble methods. Lastly, we provide a library, facilitating the practical implementation of XAI ensembling, thus promoting the adoption of transparent and interpretable deep learning models.

Simulating chemical systems is highly sought after and computationally challenging, as the simulation cost exponentially increases with the system size. Quantum computers have been proposed as a computational means to overcome this bottleneck. Most efforts recently have been spent on determining the ground states of chemical systems. Hardness results and the lack of efficient heuristics for initial-state generation sheds doubt on the feasibility. Here we propose an inherently efficient approach for solving chemical simulation problems, meaning it requires quantum circuits of size scaling polynomially in relevant system parameters. If a set of assumptions can be satisfied, our approach finds good initial states by assembling initial states for dynamical simulation in a scattering tree. We discuss a variety of quantities of chemical interest that can be measured based on quantum simulation, e.g. of a reaction, succeeding the initial state preparation., Comment: 12 pages, 4 figures

Advancements in large language models (LLMs) have demonstrated their potential in facilitating high-level reasoning, logical reasoning and robotics planning. Recently, LLMs have also been able to generate reward functions for low-level robot actions, effectively bridging the interface between high-level planning and low-level robot control. However, the challenge remains that even with syntactically correct plans, robots can still fail to achieve their intended goals due to imperfect plans or unexpected environmental issues. To overcome this, Vision Language Models (VLMs) have shown remarkable success in tasks such as visual question answering. Leveraging the capabilities of VLMs, we present a novel framework called Robotic Replanning with Perception and Language Models (RePLan) that enables online replanning capabilities for long-horizon tasks. This framework utilizes the physical grounding provided by a VLM's understanding of the world's state to adapt robot actions when the initial plan fails to achieve the desired goal. We developed a Reasoning and Control (RC) benchmark with eight long-horizon tasks to test our approach. We find that RePLan enables a robot to successfully adapt to unforeseen obstacles while accomplishing open-ended, long-horizon goals, where baseline models cannot, and can be readily applied to real robots. Find more information at https://replan-lm.github.io/replan.github.io/

We present the results of a systematic search of the Transiting Exoplanet Survey Satellite (TESS) 2-min cadence data for new rapidly oscillating Ap (roAp) stars observed during the Cycle 2 phase of its mission. We find seven new roAp stars previously unreported as such and present the analysis of a further 25 roAp stars that are already known. Three of the new stars show multiperiodic pulsations, while all new members are rotationally variable stars, leading to almost 70 per cent (22) of the roAp stars presented being $\alpha^2$ CVn-type variable stars. We show that targeted observations of known chemically peculiar stars are likely to overlook many new roAp stars, and demonstrate that multi-epoch observations are necessary to see pulsational behaviour changes. We find a lack of roAp stars close to the blue edge of the theoretical roAp instability strip, and reaffirm that mode instability is observed more frequently with precise, space-based observations. In addition to the Cycle 2 observations, we analyse TESS data for all known roAp stars. This amounts to 18 further roAp stars observed by TESS. Finally, we list six known roAp stars that TESS is yet to observe. We deduce that the incidence of roAp stars amongst the Ap star population is just 5.5 per cent, raising fundamental questions about the conditions required to excite pulsations in Ap stars. This work, coupled with our previous work on roAp stars in Cycle 1 observations, presents the most comprehensive, homogeneous study of the roAp stars in the TESS nominal mission, with a collection of 112 confirmed roAp stars in total., Comment: Accepted for publication in MNRAS. 32 Pages, 2 Tables, 77 Figures

Large Language Models (LLMs) have substantially driven scientific progress in various domains, and many papers have demonstrated their ability to tackle complex problems with creative solutions. Our paper introduces a new foundation model, nach0, capable of solving various chemical and biological tasks: biomedical question answering, named entity recognition, molecular generation, molecular synthesis, attributes prediction, and others. nach0 is a multi-domain and multi-task encoder-decoder LLM pre-trained on unlabeled text from scientific literature, patents, and molecule strings to incorporate a range of chemical and linguistic knowledge. We employed instruction tuning, where specific task-related instructions are utilized to fine-tune nach0 for the final set of tasks. To train nach0 effectively, we leverage the NeMo framework, enabling efficient parallel optimization of both base and large model versions. Extensive experiments demonstrate that our model outperforms state-of-the-art baselines on single-domain and cross-domain tasks. Furthermore, it can generate high-quality outputs in molecular and textual formats, showcasing its effectiveness in multi-domain setups., Comment: Accepted to Chemical Science Journal. Models are publicly available via https://huggingface.co/insilicomedicine/nach0_base and https://huggingface.co/insilicomedicine/nach0_large

Detailed measurements of the spectral structure of cosmic-ray electrons and positrons from 10.6 GeV to 7.5 TeV are presented from over 7 years of observations with the CALorimetric Electron Telescope (CALET) on the International Space Station. Because of the excellent energy resolution (a few percent above 10 GeV) and the outstanding e/p separation (10$^5$), CALET provides optimal performance for a detailed search of structures in the energy spectrum. The analysis uses data up to the end of 2022, and the statistics of observed electron candidates has increased more than 3 times since the last publication in 2018. By adopting an updated boosted decision tree analysis, a sufficient proton rejection power up to 7.5 TeV is achieved, with a residual proton contamination less than 10%. The observed energy spectrum becomes gradually harder in the lower energy region from around 30 GeV, consistently with AMS-02, but from 300 to 600 GeV it is considerably softer than the spectra measured by DAMPE and Fermi-LAT. At high energies, the spectrum presents a sharp break around 1 TeV, with a spectral index change from -3.15 to -3.91, and a broken power law fitting the data in the energy range from 30 GeV to 4.8 TeV better than a single power law with 6.9 sigma significance, which is compatible with the DAMPE results. The break is consistent with the expected effects of radiation loss during the propagation from distant sources (except the highest energy bin). We have fitted the spectrum with a model consistent with the positron flux measured by AMS-02 below 1 TeV and interpreted the electron + positron spectrum with possible contributions from pulsars and nearby sources. Above 4.8 TeV, a possible contribution from known nearby supernova remnants, including Vela, is addressed by an event-by-event analysis providing a higher proton-rejection power than a purely statistical analysis., Comment: main text: 7 pages, 4 figures; supplemental material: 10 pages, 8 figures, 1 table

Molecular modeling at the quantum level requires choosing a parameterization of the wavefunction that both respects the required particle symmetries, and is scalable to systems of many particles. For the simulation of fermions, valid parameterizations must be antisymmetric with respect to the exchange of particles. Typically, antisymmetry is enforced by leveraging the anti-symmetry of determinants with respect to the exchange of matrix rows, but this involves computing a full determinant each time the wavefunction is evaluated. Instead, we introduce a new antisymmetrization layer derived from sorting, the $\textit{sortlet}$, which scales as $O(N \log N)$ with regards to the number of particles -- in contrast to $O(N^3)$ for the determinant. We show numerically that applying this anti-symmeterization layer on top of an attention based neural-network backbone yields a flexible wavefunction parameterization capable of reaching chemical accuracy when approximating the ground state of first-row atoms and small molecules.

Sampling diverse, thermodynamically feasible molecular conformations plays a crucial role in predicting properties of a molecule. In this paper we propose to use GFlowNet for sampling conformations of small molecules from the Boltzmann distribution, as determined by the molecule's energy. The proposed approach can be used in combination with energy estimation methods of different fidelity and discovers a diverse set of low-energy conformations for highly flexible drug-like molecules. We demonstrate that GFlowNet can reproduce molecular potential energy surfaces by sampling proportionally to the Boltzmann distribution.

Structure determination is necessary to identify unknown organic molecules, such as those in natural products, forensic samples, the interstellar medium, and laboratory syntheses. Rotational spectroscopy enables structure determination by providing accurate 3D information about small organic molecules via their moments of inertia. Using these moments, Kraitchman analysis determines isotopic substitution coordinates, which are the unsigned $|x|,|y|,|z|$ coordinates of all atoms with natural isotopic abundance, including carbon, nitrogen, and oxygen. While unsigned substitution coordinates can verify guesses of structures, the missing $+/-$ signs make it challenging to determine the actual structure from the substitution coordinates alone. To tackle this inverse problem, we develop KREED (Kraitchman REflection-Equivariant Diffusion), a generative diffusion model that infers a molecule's complete 3D structure from its molecular formula, moments of inertia, and unsigned substitution coordinates of heavy atoms. KREED's top-1 predictions identify the correct 3D structure with >98% accuracy on the QM9 and GEOM datasets when provided with substitution coordinates of all heavy atoms with natural isotopic abundance. When substitution coordinates are restricted to only a subset of carbons, accuracy is retained at 91% on QM9 and 32% on GEOM. On a test set of experimentally measured substitution coordinates gathered from the literature, KREED predicts the correct all-atom 3D structure in 25 of 33 cases, demonstrating experimental applicability for context-free 3D structure determination with rotational spectroscopy., Comment: added software citations

Wavelet transforms are widely used in various fields of science and engineering as a mathematical tool with features that reveal information ignored by the Fourier transform. Unlike the Fourier transform, which is unique, a wavelet transform is specified by a sequence of numbers associated with the type of wavelet used and an order parameter specifying the length of the sequence. While the quantum Fourier transform, a quantum analog of the classical Fourier transform, has been pivotal in quantum computing, prior works on quantum wavelet transforms~(QWTs) were limited to the second and fourth order of a particular wavelet, the Daubechies wavelet. Here we develop a simple yet efficient quantum algorithm for executing any wavelet transform on a quantum computer. Our approach is to decompose the kernel matrix of a wavelet transform as a linear combination of unitaries (LCU) that are compilable by easy-to-implement modular quantum arithmetic operations and use the LCU technique to construct a probabilistic procedure to implement a QWT with a \textit{known} success probability. We then use properties of wavelets to make this approach deterministic by a few executions of the amplitude amplification strategy. We extend our approach to a multilevel wavelet transform and a generalized version, the packet wavelet transform, establishing computational complexities in terms of three parameters: the wavelet order $M$, the dimension $N$ of the transformation matrix, and the transformation level $d$. We show the cost is logarithmic in $N$, linear in $d$ and superlinear in $M$. Moreover, we show the cost is independent of $M$ for practical applications. Our proposed quantum wavelet transforms could be used in quantum computing algorithms in a similar manner to their well-established counterpart, the quantum Fourier transform., Comment: This version is identical in content to the published version. Presentation improved, typos fixed, and figures 1 and 2 added

Protein language models learn powerful representations directly from sequences of amino acids. However, they are constrained to generate proteins with only the set of amino acids represented in their vocabulary. In contrast, chemical language models learn atom-level representations of smaller molecules that include every atom, bond, and ring. In this work, we show that chemical language models can learn atom-level representations of proteins enabling protein generation unconstrained to the standard genetic code and far beyond it. In doing so, we show that language models can generate entire proteins atom by atom -- effectively learning the multiple hierarchical layers of molecular information that define proteins from their primary sequence to their secondary, and tertiary structure. We demonstrate language models are able to explore beyond protein space -- generating proteins with modified sidechains that form unnatural amino acids. Even further, we find that language models can explore chemical space and protein space simultaneously and generate novel examples of protein-drug conjugates. The results demonstrate the potential for biomolecular design at the atom level using language models.

Materials acceleration platforms (MAPs) combine automation and artificial intelligence to accelerate the discovery of molecules and materials. They have potential to play a role in addressing complex societal problems such as climate change. Solar chemicals and fuels generation via heterogeneous CO2 photo(thermal)catalysis is a relatively unexplored process that holds potential for contributing towards an environmentally and economically sustainable future, and therefore a very promising application for MAP science and engineering. Here, we present a brief overview of how design and innovation in heterogeneous CO2 photo(thermal)catalysis, from materials discovery to engineering and scale-up, could benefit from MAPs. We discuss relevant design and performance descriptors and the level of automation of state-of-the-art experimental techniques, and we review examples of artificial intelligence in data analysis. Based on these precedents, we finally propose a MAP outline for autonomous and accelerated discoveries in the emerging field of solar chemicals and fuels sourced from CO2 photo(thermal)catalysis.

Lung cancer and covid-19 have one of the highest morbidity and mortality rates in the world. For physicians, the identification of lesions is difficult in the early stages of the disease and time-consuming. Therefore, multi-task learning is an approach to extracting important features, such as lesions, from small amounts of medical data because it learns to generalize better. We propose a novel multi-task framework for classification, segmentation, reconstruction, and detection. To the best of our knowledge, we are the first ones who added detection to the multi-task solution. Additionally, we checked the possibility of using two different backbones and different loss functions in the segmentation task., Comment: presented at the Polish Conference on Artificial Intelligence (PP-RAI), 2023

Cepheids have long been used as standard candles to determine distances around the Milky Way and to nearby galaxies. A discrepancy still remains for Hubble Constant determinations using Cepheids vs. the cosmic microwave background or calibrations to the tip of the red-giant branch. Therefore, refinement of Cepheid period-luminosity relations continues to be an active topic of research. Cepheids are also important laboratories for testing stellar physics. This paper explores outstanding questions in Cepheid evolution and pulsation modeling. We examine the discrepancy between Cepheid masses determined from pulsation properties and binary orbital dynamics and those determined using stellar evolution models. We review attempts to resolve the discrepancy by including rotation, convective overshooting, and mass loss. We review the impact of uncertainties in nuclear reaction rates on Cepheid evolution and the extent of blue loops in the Hertzsprung-Russell diagram. We consider implications for Cepheids of stellar opacity revisions suggested in light of findings for the Sun and other types of variable stars. We apply the 1-D open-source MESA stellar evolution code and the MESA radial stellar pulsation (RSP) nonlinear hydrodynamics code to investigate changes in input physics for Cepheid models. We touch on progress in 2-D and 3-D stellar modeling applied to Cepheids. Additional areas in which Cepheid models are being tested against observations include: predicting the edges of the Cepheid pulsation instability strip; predicting period-change rates and implications for instability strip crossings; explaining period and amplitude modulations and periodicities that may be non-radial pulsation modes; discovering what can be learned from Cepheid observations in X-ray, ultraviolet, and radio wavelengths. We also show a few examples of Cepheid light curves from NASA TESS photometry., Comment: 10 pages, 17 figures. Revised and updated version of conference proceedings submitted for 2023 Society for Astronomical Sciences Symposium on Telescope Science, June 22-24, 2023, eds. J.C. Martin, R.K. Buchheim, R.M. Gill, W. Green, and J. Menke

Advances in artificial intelligence (AI) are fueling a new paradigm of discoveries in natural sciences. Today, AI has started to advance natural sciences by improving, accelerating, and enabling our understanding of natural phenomena at a wide range of spatial and temporal scales, giving rise to a new area of research known as AI for science (AI4Science). Being an emerging research paradigm, AI4Science is unique in that it is an enormous and highly interdisciplinary area. Thus, a unified and technical treatment of this field is needed yet challenging. This work aims to provide a technically thorough account of a subarea of AI4Science; namely, AI for quantum, atomistic, and continuum systems. These areas aim at understanding the physical world from the subatomic (wavefunctions and electron density), atomic (molecules, proteins, materials, and interactions), to macro (fluids, climate, and subsurface) scales and form an important subarea of AI4Science. A unique advantage of focusing on these areas is that they largely share a common set of challenges, thereby allowing a unified and foundational treatment. A key common challenge is how to capture physics first principles, especially symmetries, in natural systems by deep learning methods. We provide an in-depth yet intuitive account of techniques to achieve equivariance to symmetry transformations. We also discuss other common technical challenges, including explainability, out-of-distribution generalization, knowledge transfer with foundation and large language models, and uncertainty quantification. To facilitate learning and education, we provide categorized lists of resources that we found to be useful. We strive to be thorough and unified and hope this initial effort may trigger more community interests and efforts to further advance AI4Science.

Asteroseismology has transformed stellar astrophysics. Red giant asteroseismology is a prime example, with oscillation periods and amplitudes that are readily detectable with time-domain space-based telescopes. These oscillations can be used to infer masses, ages and radii for large numbers of stars, providing unique constraints on stellar populations in our galaxy. The cadence, duration, and spatial resolution of the Roman galactic bulge time-domain survey (GBTDS) are well-suited for asteroseismology and will probe an important population not studied by prior missions. We identify photometric precision as a key requirement for realizing the potential of asteroseismology with Roman. A precision of 1 mmag per 15-min cadence or better for saturated stars will enable detections of the populous red clump star population in the Galactic bulge. If the survey efficiency is better than expected, we argue for repeat observations of the same fields to improve photometric precision, or covering additional fields to expand the stellar population reach if the photometric precision for saturated stars is better than 1 mmag. Asteroseismology is relatively insensitive to the timing of the observations during the mission, and the prime red clump targets can be observed in a single 70 day campaign in any given field. Complementary stellar characterization, particularly astrometry tied to the Gaia system, will also dramatically expand the diagnostic power of asteroseismology. We also highlight synergies to Roman GBTDS exoplanet science using transits and microlensing., Comment: Roman Core Community Survey White Paper, 3 pages, 4 figures

Partial differential equations (PDEs) are ubiquitous in science and engineering. Prior quantum algorithms for solving the system of linear algebraic equations obtained from discretizing a PDE have a computational complexity that scales at least linearly with the condition number $\kappa$ of the matrices involved in the computation. For many practical applications, $\kappa$ scales polynomially with the size $N$ of the matrices, rendering a polynomial-in-$N$ complexity for these algorithms. Here we present a quantum algorithm with a complexity that is polylogarithmic in $N$ but is independent of $\kappa$ for a large class of PDEs. Our algorithm generates a quantum state that enables extracting features of the solution. Central to our methodology is using a wavelet basis as an auxiliary system of coordinates in which the condition number of associated matrices is independent of $N$ by a simple diagonal preconditioner. We present numerical simulations showing the effect of the wavelet preconditioner for several differential equations. Our work could provide a practical way to boost the performance of quantum-simulation algorithms where standard methods are used for discretization., Comment: 5 pages main text, 14 pages in total including appendix, 2 figures; updated references and fixed typos in this version

We present the observation of a charge-sign dependent solar modulation of galactic cosmic rays (GCRs) with the CALorimetric Electron Telescope onboard the International Space Station over 6 yr, corresponding to the positive polarity of the solar magnetic field. The observed variation of proton count rate is consistent with the neutron monitor count rate, validating our methods for determining the proton count rate. It is observed by the CALorimetric Electron Telescope that both GCR electron and proton count rates at the same average rigidity vary in anticorrelation with the tilt angle of the heliospheric current sheet, while the amplitude of the variation is significantly larger in the electron count rate than in the proton count rate. We show that this observed charge-sign dependence is reproduced by a numerical ``drift model'' of the GCR transport in the heliosphere. This is a clear signature of the drift effect on the long-term solar modulation observed with a single detector., Comment: main text: 6 pages, 3 figures, supplemental material: 8 pages, 3 figures, 1 table

Language models are powerful tools for molecular design. Currently, the dominant paradigm is to parse molecular graphs into linear string representations that can easily be trained on. This approach has been very successful, however, it is limited to chemical structures that can be completely represented by a graph -- like organic molecules -- while materials and biomolecular structures like protein binding sites require a more complete representation that includes the relative positioning of their atoms in space. In this work, we show how language models, without any architecture modifications, trained using next-token prediction -- can generate novel and valid structures in three dimensions from various substantially different distributions of chemical structures. In particular, we demonstrate that language models trained directly on sequences derived directly from chemical file formats like XYZ files, Crystallographic Information files (CIFs), or Protein Data Bank files (PDBs) can directly generate molecules, crystals, and protein binding sites in three dimensions. Furthermore, despite being trained on chemical file sequences -- language models still achieve performance comparable to state-of-the-art models that use graph and graph-derived string representations, as well as other domain-specific 3D generative models. In doing so, we demonstrate that it is not necessary to use simplified molecular representations to train chemical language models -- that they are powerful generative models capable of directly exploring chemical space in three dimensions for very different structures.

Estimation of the expectation value of observables is a key subroutine in quantum computing and is also the bottleneck of the performance of many near-term quantum algorithms. Many works have been proposed to reduce the number of measurements needed for this task and they provide different measurement schemes for generating the measurements to perform. In this paper, we propose a new approach, composite measurement scheme, which composes multiple measurement schemes by distributing shots to them with a trainable ratio. As an example of our method, we study the case where only Pauli measurements are allowed and propose Composite-LBCS (C-LBCS), a composite measurement scheme made by composing locally-biased classical shadows. We numerically demonstrate C-LBCS on molecular systems up to $\mathrm{CO}_2$ (30 qubits) and show that C-LBCS outperforms the previous state-of-the-art methods despite its simplicity. We also show that C-LBCS can be efficiently optimized by stochastic gradient descent and is trainable even when the observable contains a large number of terms. We believe our method opens up a reliable way toward efficient observable estimation on large quantum systems., Comment: 12 pages, 2 figures

We present the results of a direct measurement of the cosmic-ray helium spectrum with the CALET instrument in operation on the International Space Station since 2015. The observation period covered by this analysis spans from October 13, 2015 to April 30, 2022 (2392 days). The very wide dynamic range of CALET allowed to collect helium data over a large energy interval, from ~40 GeV to ~250 TeV, for the first time with a single instrument in Low Earth Orbit. The measured spectrum shows evidence of a deviation of the flux from a single power-law by more than 8$\sigma$ with a progressive spectral hardening from a few hundred GeV to a few tens of TeV. This result is consistent with the data reported by space instruments including PAMELA, AMS-02, DAMPE and balloon instruments including CREAM. At higher energy we report the onset of a softening of the helium spectrum around 30 TeV (total kinetic energy). Though affected by large uncertainties in the highest energy bins, the observation of a flux reduction turns out to be consistent with the most recent results of DAMPE. A Double Broken Power Law (DBPL) is found to fit simultaneously both spectral features: the hardening (at lower energy) and the softening (at higher energy). A measurement of the proton to helium flux ratio in the energy range from 60 GeV/n to about 60 TeV/n is also presented, using the CALET proton flux recently updated with higher statistics.

Generating low-level robot task plans from high-level natural language instructions remains a challenging problem. Although large language models have shown promising results in generating plans, the accuracy of the output remains unverified. Furthermore, the lack of domain-specific language data poses a limitation on the applicability of these models. In this paper, we propose CLAIRIFY, a novel approach that combines automatic iterative prompting with program verification to ensure programs written in data-scarce domain-specific language are syntactically valid and incorporate environment constraints. Our approach provides effective guidance to the language model on generating structured-like task plans by incorporating any errors as feedback, while the verifier ensures the syntactic accuracy of the generated plans. We demonstrate the effectiveness of CLAIRIFY in planning chemistry experiments by achieving state-of-the-art results. We also show that the generated plans can be executed on a real robot by integrating them with a task and motion planner.

The prediction of age is a challenging task with various practical applications in high-impact fields like the healthcare domain or criminology. Despite the growing number of models and their increasing performance, we still know little about how these models work. Numerous examples of failures of AI systems show that performance alone is insufficient, thus, new methods are needed to explore and explain the reasons for the model's predictions. In this paper, we investigate the use of Explainable Artificial Intelligence (XAI) for age prediction focusing on two specific modalities, EEG signal and lung X-rays. We share predictive models for age to facilitate further research on new techniques to explain models for these modalities., Comment: Presented at the Aging Hackathon at the 7th International workshop on Health Intelligence (W3PHIAI-23) at AAAI-23 Conference, http://w3phiai2023.w3phi.com/

Current quantum computing hardware is restricted by the availability of only few, noisy qubits which limits the investigation of larger, more complex molecules in quantum chemistry calculations on quantum computers in the near-term. In this work, we investigate the limits of their classical and near-classical treatment while staying within the framework of quantum circuits and the variational quantum eigensolver. To this end, we consider naive and physically motivated, classically efficient product ansatz for the parametrized wavefunction adapting the separable pair ansatz form. We combine it with post-treatment to account for interactions between subsystems originating from this ansatz. The classical treatment is given by another quantum circuit that has support between the enforced subsystems and is folded into the Hamiltonian. To avoid an exponential increase in the number of Hamiltonian terms, the entangling operations are constructed from purely Clifford or near-Clifford circuits. While Clifford circuits can be simulated efficiently classically, they are not universal. In order to account for missing expressibility, near-Clifford circuits with only few, selected non-Clifford gates are employed. The exact circuit structure to achieve this objective is molecule-dependent and is constructed using simulated annealing and genetic algorithms. We demonstrate our approach on a set of molecules of interest and investigate the extent of our methodology's reach. Empirical validation of our approach using numerical simulations shows a reduction of the qubit count of up to a 50\% at a similar accuracy as compared to the separable-pair ansatz., Comment: 12 pages, 9 figures plus 3 pages, 8 figures appendix

NGC 6819 is an open cluster of age 2.4 Gyr that was in the NASA Kepler spacecraft field of view from 2009 to 2013. The central part of the cluster was observed in a 200 x 200 pixel `superstamp' during these four years in 30-minute cadence photometry, providing a unique long time-series high-precision data set. The cluster contains 'blue straggler' stars, i.e., stars on the main sequence above the cluster turnoff that should have left the main sequence to become red giants. We present light curves and pulsation frequency analyses derived from custom photometric reductions for five confirmed cluster members--four blue stragglers and one star near the main-sequence turnoff. Two of these stars show a rich spectrum of $\delta$ Scuti pulsation modes, with 236 and 124 significant frequencies identified, respectively, while two stars show mainly low-frequency modes, characteristic of $\gamma$ Doradus variable stars. The fifth star, a known active x-ray binary, shows only several harmonics of two main frequencies. For the two $\delta$ Scuti stars, we use a frequency separation--mean-density relation to estimate mean density, and then use this value along with effective temperature to derive stellar mass and radius. For the two stars showing low frequencies, we searched for period-spacing sequences that may be representative of gravity-mode or Rossby-mode sequences, but found no clear sequences. The common age for the cluster members, considered along with the frequencies, will provide valuable constraints for asteroseismic analyses, and may shed light on the origin of the blue stragglers., Comment: Accepted for publication in The Astronomical Journal February 23, 2023. 36 pages, 14 figures, 4 tables in text, 5 in Appendix