Search

First-order Policy Gradient (FoPG) algorithms such as Backpropagation through Time and Analytical Policy Gradients leverage local simulation physics to accelerate policy search, significantly improving sample efficiency in robot control compared to standard model-free reinforcement learning. However, FoPG algorithms can exhibit poor learning dynamics in contact-rich tasks like locomotion. Previous approaches address this issue by alleviating contact dynamics via algorithmic or simulation innovations. In contrast, we propose guiding the policy search by learning a residual over a simple baseline policy. For quadruped locomotion, we find that the role of residual policy learning in FoPG-based training (FoPG RPL) is primarily to improve asymptotic rewards, compared to improving sample efficiency for model-free RL. Additionally, we provide insights on applying FoPG's to pixel-based local navigation, training a point-mass robot to convergence within seconds. Finally, we showcase the versatility of FoPG RPL by using it to train locomotion and perceptive navigation end-to-end on a quadruped in minutes.

Bosonic two-mode squeezed states are paradigmatic entangled states in continuous variable systems, which have broad applications in quantum information processing. In this work, we propose a photon-phonon squeezing protocol assisted by a Kerr magnon within a hybrid cavity magnomechanical system. We construct an effective Hamiltonian that accounts for photon-phonon squeezing through strong photon-magnon interaction and precise modulation over the driving frequency on the photon mode. The effective Hamiltonian can be confirmed by a fascinating method about the diagonalization of the system's Liouvilian superoperator. This method can address the level attractions rather than avoided level crossings in the energy diagram of the whole system. With the effective Hamiltonian and quantum Langevin equation, we provide a rigorous theoretical solution for the dynamical process of squeezing generation. Our finding indicates that asymptotic stationary squeezing can be obtained by optimizing the squeezing quadrature operator, even when the covariance matrix of the system still varies with time. This squeezing level can exceed the maximum value under stable conditions. Moreover, our analysis also reveals that the Kerr nonlinearity of the magnon can further promote the squeezing generation. Our work provides an extendable framework for generating squeezed states that entangle two Gaussian modes with indirect coupling.

Detecting cracks with pixel-level precision for key structures is a significant challenge, as existing methods struggle to effectively integrate local textures and pixel dependencies of cracks. Furthermore, these methods often possess numerous parameters and substantial computational requirements, complicating deployment on edge control devices. In this paper, we propose a staircase cascaded fusion crack segmentation network (CrackSCF) that generates high-quality crack segmentation maps using minimal computational resources. We constructed a staircase cascaded fusion module that effectively captures local patterns of cracks and long-range dependencies of pixels, and it can suppress background noise well. To reduce the computational resources required by the model, we introduced a lightweight convolution block, which replaces all convolution operations in the network, significantly reducing the required computation and parameters without affecting the network's performance. To evaluate our method, we created a challenging benchmark dataset called TUT and conducted experiments on this dataset and five other public datasets. The experimental results indicate that our method offers significant advantages over existing methods, especially in handling background noise interference and detailed crack segmentation. The F1 and mIoU scores on the TUT dataset are 0.8382 and 0.8473, respectively, achieving state-of-the-art (SOTA) performance while requiring the least computational resources. The code and dataset is available at https://github.com/Karl1109/CrackSCF.

Time series classification with missing data is a prevalent issue in time series analysis, as temporal data often contain missing values in practical applications. The traditional two-stage approach, which handles imputation and classification separately, can result in sub-optimal performance as label information is not utilized in the imputation process. On the other hand, a one-stage approach can learn features under missing information, but feature representation is limited as imputed errors are propagated in the classification process. To overcome these challenges, this study proposes an end-to-end neural network that unifies data imputation and representation learning within a single framework, allowing the imputation process to take advantage of label information. Differing from previous methods, our approach places less emphasis on the accuracy of imputation data and instead prioritizes classification performance. A specifically designed multi-scale feature learning module is implemented to extract useful information from the noise-imputation data. The proposed model is evaluated on 68 univariate time series datasets from the UCR archive, as well as a multivariate time series dataset with various missing data ratios and 4 real-world datasets with missing information. The results indicate that the proposed model outperforms state-of-the-art approaches for incomplete time series classification, particularly in scenarios with high levels of missing data.

We infer the growth of large scale structure over the redshift range $0.4\lesssim z \lesssim 1$ from the cross-correlation of spectroscopically calibrated Luminous Red Galaxies (LRGs) selected from the Dark Energy Spectroscopic Instrument (DESI) legacy imaging survey with CMB lensing maps reconstructed from the latest Planck and ACT data. We adopt a hybrid effective field theory (HEFT) model that robustly regulates the cosmological information obtainable from smaller scales, such that our cosmological constraints are reliably derived from the (predominantly) linear regime. We perform an extensive set of bandpower- and parameter-level systematics checks to ensure the robustness of our results and to characterize the uniformity of the LRG sample. We demonstrate that our results are stable to a wide range of modeling assumptions, finding excellent agreement with a linear theory analysis performed on a restricted range of scales. From a tomographic analysis of the four LRG photometric redshift bins we find that the rate of structure growth is consistent with $\Lambda$CDM with an overall amplitude that is $\simeq5-7\%$ lower than predicted by primary CMB measurements with modest $(\sim2\sigma)$ statistical significance. From the combined analysis of all four bins and their cross-correlations with Planck we obtain $S_8 = 0.765\pm0.023$, which is less discrepant with primary CMB measurements than previous DESI LRG cross Planck CMB lensing results. From the cross-correlation with ACT we obtain $S_8 = 0.790^{+0.024}_{-0.027}$, while when jointly analyzing Planck and ACT we find $S_8 = 0.775^{+0.019}_{-0.022}$ from our data alone and $\sigma_8 = 0.772^{+0.020}_{-0.023}$ with the addition of BAO data. These constraints are consistent with the latest Planck primary CMB analyses at the $\simeq 1.6-2.2\sigma$ level, and are in excellent agreement with galaxy lensing surveys., Comment: 60 pages, 26 figures, comments welcome

We present a high-significance cross-correlation of CMB lensing maps from the Atacama Cosmology Telescope (ACT) Data Release 6 (DR6) with spectroscopically calibrated luminous red galaxies (LRGs) from the Dark Energy Spectroscopic Instrument (DESI). We detect this cross-correlation at a significance of 38$\sigma$; combining our measurement with the Planck Public Release 4 (PR4) lensing map, we detect the cross-correlation at 50$\sigma$. Fitting this jointly with the galaxy auto-correlation power spectrum to break the galaxy bias degeneracy with $\sigma_8$, we perform a tomographic analysis in four LRG redshift bins spanning $0.4 \le z \le 1.0$ to constrain the amplitude of matter density fluctuations through the parameter combination $S_8^\times = \sigma_8 \left(\Omega_m / 0.3\right)^{0.4}$. Prior to unblinding, we confirm with extragalactic simulations that foreground biases are negligible and carry out a comprehensive suite of null and consistency tests. Using a hybrid effective field theory (HEFT) model that allows scales as small as $k_{\rm max}=0.6$ $h/{\rm Mpc}$, we obtain a 3.3% constraint on $S_8^\times = \sigma_8 \left(\Omega_m / 0.3\right)^{0.4} = 0.792^{+0.024}_{-0.028}$ from ACT data, as well as constraints on $S_8^\times(z)$ that probe structure formation over cosmic time. Our result is consistent with the early-universe extrapolation from primary CMB anisotropies measured by Planck PR4 within 1.2$\sigma$. Jointly fitting ACT and Planck lensing cross-correlations we obtain a 2.7% constraint of $S_8^\times = 0.776^{+0.019}_{-0.021}$, which is consistent with the Planck early-universe extrapolation within 2.1$\sigma$, with the lowest redshift bin showing the largest difference in mean. The latter may motivate further CMB lensing tomography analyses at $z<0.6$ to assess the impact of potential systematics or the consistency of the $\Lambda$CDM model over cosmic time., Comment: Prepared for submission to JCAP (47 pages, 13 figures)

We present a new perturbative full-shape analysis of BOSS galaxy clustering data, including the full combination of the galaxy power spectrum and bispectrum multipoles, baryon acoustic oscillations, and cross-correlations with the gravitational lensing of cosmic microwave background measured from \textit{Planck}. Assuming the $\Lambda$CDM model, we constrain the matter density fraction $\Omega_m = 0.3154\pm 0.0089$, the Hubble constant $H_0=68.34\pm 0.77\,\mathrm{km}\,\mathrm{s}^{-1}\mathrm{Mpc}^{-1}$, and the mass fluctuation amplitude $\sigma_8=0.686\pm 0.027$ (equivalent to $S_8 = 0.704\pm 0.031$). Cosmic structure at low redshifts appears suppressed with respect to the Planck $\Lambda$CDM concordance model at $4.5\sigma$. We explore whether this tension can be explained by the recent DESI preference for dynamical dark energy (DDE): the BOSS data combine with DESI BAO and PantheonPlus supernovae competitively compared to the CMB, yielding no preference for DDE, but the same $\sim 10\%$ suppression of structure, with dark energy being consistent with a cosmological constant at 68\% CL. Our results suggest that either the data contains residual systematics, or more model-building efforts may be required to restore cosmological concordance., Comment: 9 pages, 2 figures, to be submitted to PRL

We study the bispectrum in Lagrangian perturbation theory. Extending past results for the power spectrum, we describe a method to efficiently compute the bispectrum in LPT, focusing on the Zeldovich approximation, in which contributions due to linear displacements are captured to all orders in a manifestly infrared (IR) safe way. We then isolate the effects of these linear displacements on oscillatory components of the power spectrum like baryon acoustic oscillations or inflationary primordial features and show that the Eulerian perturbation theory (EPT) prescription wherein their effects are resummed by a Gaussian damping of the oscillations arise as a saddle-point approximation of our calculation. These two methods of IR resummation are in excellent agreement at 1-loop in the bispectrum. At tree level, resummed EPT does less well to capture the nonlinear damping of the oscillations, and the LPT calculation does not require an artificial split of the power spectrum into smooth and oscillatory components, making the latter particularly useful for modeling exotic features. We finish by extending our analysis of IR resummation in LPT to N-point functions of arbitrary order., Comment: 47 pages, 13 figures, updated to match version accepted by JCAP

We derive new constraints on the $E_G$ statistic as a test of gravity, combining the CMB lensing map estimated from Data Release 6 (DR6) of the Atacama Cosmology Telescope with SDSS BOSS CMASS and LOWZ galaxy data. We develop an analysis pipeline to measure the cross-correlation between CMB lensing maps and galaxy data, following a blinding policy and testing the approach through null and consistency checks. By testing the equivalence of the spatial and temporal gravitational potentials, the $E_G$ statistic can distinguish $\Lambda$CDM from alternative models of gravity. We find $E_G= 0.31^{+0.06}_{-0.05}$ for ACT and CMASS data at 68.28\% confidence level, and $E_G = 0.49^{+0.14}_{-0.11}$ for ACT and LOWZ. Systematic errors are estimated to be 3\% and 4\% respectively. Including CMB lensing information from Planck PR4 results in $E_G = 0.34^{+0.05}_{-0.05}$ with CMASS and $E_G= 0.43^{+0.11}_{-0.09}$ with LOWZ. These are consistent with predictions for the $\Lambda$CDM model that best fits the Planck CMB anisotropy and SDSS BOSS BAO, where $E_G^{\rm GR} (z_{\rm eff} = 0.555) = 0.401\pm 0.005$ for CMB lensing combined with CMASS and $E_G^{\rm GR} (z_{\rm eff} = 0.316) = 0.452\pm0.005$ combined with LOWZ. We also find $E_G$ to be scale independent, with PTE $>5\%$, as predicted by general relativity. The methods developed in this work are also applicable to improved future analyses with upcoming spectroscopic galaxy samples and CMB lensing measurements., Comment: 33 pages, 21 figures, prepared for submission to PRD

Legged locomotion has recently achieved remarkable success with the progress of machine learning techniques, especially deep reinforcement learning (RL). Controllers employing neural networks have demonstrated empirical and qualitative robustness against real-world uncertainties, including sensor noise and external perturbations. However, formally investigating the vulnerabilities of these locomotion controllers remains a challenge. This difficulty arises from the requirement to pinpoint vulnerabilities across a long-tailed distribution within a high-dimensional, temporally sequential space. As a first step towards quantitative verification, we propose a computational method that leverages sequential adversarial attacks to identify weaknesses in learned locomotion controllers. Our research demonstrates that, even state-of-the-art robust controllers can fail significantly under well-designed, low-magnitude adversarial sequence. Through experiments in simulation and on the real robot, we validate our approach's effectiveness, and we illustrate how the results it generates can be used to robustify the original policy and offer valuable insights into the safety of these black-box policies. Project page: https://fanshi14.github.io/me/rss24.html, Comment: RSS 2024

Massive particles leave imprints on primordial non-Gaussianity via couplings to the inflaton, even despite their exponential dilution during inflation: practically, the Universe acts as a Cosmological Collider. We present the first dedicated search for spin-zero particles using BOSS redshift-space galaxy power spectrum and bispectrum multipoles, as well as Planck CMB non-Gaussianity data. We demonstrate that some Cosmological Collider models are well approximated by the standard equilateral and orthogonal parametrization; assuming negligible inflaton self-interactions, this facilitates us translating Planck non-Gaussianity constraints into bounds on Collider models. Many models have signatures that are not degenerate with equilateral and orthogonal non-Gaussianity and thus require dedicated searches. Here, we constrain such models using BOSS three-dimensional redshift-space galaxy clustering data, focusing on spin-zero particles in the principal series and constraining their couplings to the inflaton at varying speed and mass, marginalizing over the unknown inflaton self-interactions. This is made possible through an improvement in Cosmological Bootstrap techniques and the combination of perturbation theory and halo occupation distribution models for galaxy clustering. Our work sets the standard for inflationary spectroscopy with cosmological observations, providing the ultimate link between physics on the largest and smallest scales., Comment: 35 pages, 15 figures, 6 tables

This paper provides a comprehensive overview of how fitting of Baryon Acoustic Oscillations (BAO) is carried out within the upcoming Dark Energy Spectroscopic Instrument's (DESI) 2024 results using its DR1 dataset, and the associated systematic error budget from theory and modelling of the BAO. We derive new results showing how non-linearities in the clustering of galaxies can cause potential biases in measurements of the isotropic ($\alpha_{\mathrm{iso}}$) and anisotropic ($\alpha_{\mathrm{ap}}$) BAO distance scales, and how these can be effectively removed with an appropriate choice of reconstruction algorithm. We then demonstrate how theory leads to a clear choice for how to model the BAO and develop, implement and validate a new model for the remaining smooth-broadband (i.e., without BAO) component of the galaxy clustering. Finally, we explore the impact of all remaining modelling choices on the BAO constraints from DESI using a suite of high-precision simulations, arriving at a set of best-practices for DESI BAO fits, and an associated theory and modelling systematic error. Overall, our results demonstrate the remarkable robustness of the BAO to all our modelling choices and motivate a combined theory and modelling systematic error contribution to the post-reconstruction DESI BAO measurements of no more than $0.1\%$ ($0.2\%$) for its isotropic (anisotropic) distance measurements. We expect the theory and best-practices laid out to here to be applicable to other BAO experiments in the era of DESI and beyond., Comment: 30 pages, 18 figures, 1 table, updated to match version accepted by MNRAS

The $E_G$ statistic is a discriminating probe of gravity developed to test the prediction of general relativity (GR) for the relation between gravitational potential and clustering on the largest scales in the observable universe. We present a novel high-precision estimator for the $E_G$ statistic using CMB lensing and galaxy clustering correlations that carefully matches the effective redshifts across the different measurement components to minimize corrections. A suite of detailed tests is performed to characterize the estimator's accuracy, its sensitivity to assumptions and analysis choices and the non-Gaussianity of the estimator's uncertainty is characterized. After finalization of the estimator, it is applied to $\textit{Planck}$ CMB lensing and SDSS CMASS and LOWZ galaxy data. We report the first harmonic space measurement of $E_G$ using the LOWZ sample and CMB lensing and also updated constraints using the final CMASS sample and the latest $\textit{Planck}$ CMB lensing map. We find $E_G^{Planck+CMASS} = 0.36^{+0.06}_{-0.05}$ (68.27%) and $E_G^{\rm \textit{Planck}+LOWZ} = 0.40^{+0.11}_{-0.09} $ (68.27%), with additional subdominant systematic error budget estimates of 2% and 3% respectively. Using $\Omega_{\rm m,0}$ constraints from $\textit{Planck}$ and SDSS BAO observations, $\Lambda$CDM-GR predicts $E_G^{\rm GR} (z = 0.555) = 0.401 \pm 0.005$ and $E_G^{\rm GR} (z = 0.316) = 0.452 \pm 0.005$ at the effective redshifts of the CMASS and LOWZ based measurements. We report the measurement to be in good statistical agreement with the $\Lambda$CDM-GR prediction, and report that the measurement is also consistent with the more general GR prediction of scale-independence for $E_G$. This work provides a carefully constructed and calibrated statistic with which $E_G$ measurements can be confidently and accurately obtained with upcoming survey data., Comment: 36 pages, 20 figures, Accepted for publication in PRD

Endowing machines with abstract reasoning ability has been a long-term research topic in artificial intelligence. Raven's Progressive Matrix (RPM) is widely used to probe abstract visual reasoning in machine intelligence, where models will analyze the underlying rules and select one image from candidates to complete the image matrix. Participators of RPM tests can show powerful reasoning ability by inferring and combining attribute-changing rules and imagining the missing images at arbitrary positions of a matrix. However, existing solvers can hardly manifest such an ability in realistic RPM tests. In this paper, we propose a deep latent variable model for answer generation problems through Rule AbstractIon and SElection (RAISE). RAISE can encode image attributes into latent concepts and abstract atomic rules that act on the latent concepts. When generating answers, RAISE selects one atomic rule out of the global knowledge set for each latent concept to constitute the underlying rule of an RPM. In the experiments of bottom-right and arbitrary-position answer generation, RAISE outperforms the compared solvers in most configurations of realistic RPM datasets. In the odd-one-out task and two held-out configurations, RAISE can leverage acquired latent concepts and atomic rules to find the rule-breaking image in a matrix and handle problems with unseen combinations of rules and attributes.

This comprehensive paper delves into the forefront of personalized voice synthesis within artificial intelligence (AI), spotlighting the Dynamic Individual Voice Synthesis Engine (DIVSE). DIVSE represents a groundbreaking leap in text-to-voice (TTS) technology, uniquely focusing on adapting and personalizing voice outputs to match individual vocal characteristics. The research underlines the gap in current AI-generated voices, which, while technically advanced, fall short in replicating the unique individuality and expressiveness intrinsic to human speech. It outlines the challenges and advancements in personalized voice synthesis, emphasizing the importance of emotional expressiveness, accent and dialect variability, and capturing individual voice traits. The architecture of DIVSE is meticulously detailed, showcasing its three core components: Voice Characteristic Learning Module (VCLM), Emotional Tone and Accent Adaptation Module (ETAAM), and Dynamic Speech Synthesis Engine (DSSE). The innovative approach of DIVSE lies in its adaptive learning capability, which evolves over time to tailor voice outputs to specific user traits. The paper presents a rigorous experimental setup, utilizing accepted datasets and personalization metrics like Mean Opinion Score (MOS) and Emotional Alignment Score, to validate DIVSE's superiority over mainstream models. The results depict a clear advancement in achieving higher personalization and emotional resonance in AI-generated voices., Comment: 8 pages

In this research, we introduce RefineNet, a novel architecture designed to address resolution limitations in text-to-image conversion systems. We explore the challenges of generating high-resolution images from textual descriptions, focusing on the trade-offs between detail accuracy and computational efficiency. RefineNet leverages a hierarchical Transformer combined with progressive and conditional refinement techniques, outperforming existing models in producing detailed and high-quality images. Through extensive experiments on diverse datasets, we demonstrate RefineNet's superiority in clarity and resolution, particularly in complex image categories like animals, plants, and human faces. Our work not only advances the field of image-to-text conversion but also opens new avenues for high-fidelity image generation in various applications., Comment: 8 pages

We formulate the Lagrangian perturbation theory of galaxy intrinsic alignments and compute the resulting auto and cross power spectra of galaxy shapes, densities and matter to 1-loop order. Our model represents a consistent effective-theory description of galaxy shape including the resummation of long-wavelength displacements which damp baryon acoustic oscillations, and includes one linear, three quadratic and two cubic dimensionless bias coefficients at this order, along with counterterms and stochastic contributions whose structure we derive. We compare this Lagrangian model against the three-dimensional helicity spectra of halo shapes measured in N-body simulations by Akitsu et al (2023) and find excellent agreement on perturbative scales while testing a number of more restrictive bias parametrizations. The calculations presented are immediately relevant to analyses of both cosmic shear surveys and spectroscopic shape measurements, and we make a fast FFTLog-based code spinosaurus publicly available with this publication., Comment: 41 pages, 8 figures, updated to match version accepted in JCAP. Fast Python code to predict IA correlations computed in the text can be found at https://github.com/sfschen/spinosaurus

Transferring human motion skills to humanoid robots remains a significant challenge. In this study, we introduce a Wasserstein adversarial imitation learning system, allowing humanoid robots to replicate natural whole-body locomotion patterns and execute seamless transitions by mimicking human motions. First, we present a unified primitive-skeleton motion retargeting to mitigate morphological differences between arbitrary human demonstrators and humanoid robots. An adversarial critic component is integrated with Reinforcement Learning (RL) to guide the control policy to produce behaviors aligned with the data distribution of mixed reference motions. Additionally, we employ a specific Integral Probabilistic Metric (IPM), namely the Wasserstein-1 distance with a novel soft boundary constraint to stabilize the training process and prevent mode collapse. Our system is evaluated on a full-sized humanoid JAXON in the simulator. The resulting control policy demonstrates a wide range of locomotion patterns, including standing, push-recovery, squat walking, human-like straight-leg walking, and dynamic running. Notably, even in the absence of transition motions in the demonstration dataset, robots showcase an emerging ability to transit naturally between distinct locomotion patterns as desired speed changes.

In order to address fundamental questions related to the expansion history of the Universe and its primordial nature with the next generation of galaxy experiments, we need to model reliably large-scale structure observables such as the correlation function and the power spectrum. Cosmological $N$-body simulations provide a reference through which we can test our models, but their output suffers from sample variance on large scales. Fortunately, this is the regime where accurate analytic approximations exist. To reduce the variance, which is key to making optimal use of these simulations, we can leverage the accuracy and precision of such analytic descriptions using Control Variates (CV). The power of control variates stems from utilizing inexpensive but highly correlated surrogates of the statistics one wishes to measure. The stronger the correlation between the surrogate and the statistic of interest, the larger the variance reduction delivered by the method. We apply two control variate formulations to mock catalogs generated in anticipation of upcoming data from the Dark Energy Spectroscopic Instrument (DESI) to test the robustness of its analysis pipeline. Our CV-reduced measurements offer a factor of 5-10 improvement in the measurement error compared with the raw measurements. We explore the relevant properties of the galaxy samples that dictate this reduction and comment on the improvements we find on some of the derived quantities relevant to Baryon Acoustic Oscillation (BAO) analysis. We also provide an optimized package for computing the power spectra and other two-point statistics of an arbitrary galaxy catalog as well as a pipeline for obtaining CV-reduced measurements on any of the AbacusSummit cubic box outputs. We make our scripts publicly available and report a speed improvement of $\sim$10 for a grid size of $N_{\rm mesh} = 256^3$ compared with \texttt{nbodykit}., Comment: 17 pages, 9 figures, public package (for power spectrum and control variates estimation)

In addition to the intrinsic clustering of galaxies themselves, the spatial distribution of galaxies observed in surveys is modulated by the presence of weak lensing due to matter in the foreground. This effect, known as magnification bias, is a significant contaminant to analyses of galaxy-lensing cross-correlations and must be carefully modelled. We present a method to estimate the magnification bias in spectroscopically confirmed galaxy samples based on finite differences of galaxy catalogues while marginalizing over errors due to finite step size. We use our estimator to measure the magnification biases of the CMASS and LOWZ samples in the SDSS BOSS galaxy survey, analytically taking into account the dependence on galaxy shape for fiber and PSF magnitudes, finding $\alpha_{\rm CMASS} = 2.71 \pm 0.02$ and $\alpha_{\rm LOWZ} = 2.45 \pm 0.02$ and quantify modelling uncertainties in these measurements. Finally, we quantify the redshift evolution of the magnification bias within the CMASS and LOWZ samples, finding a difference of up to a factor of three between the lower and upper redshift bounds for the former. We discuss how to account for this evolution in modelling and its interaction with commonly applied redshift-dependent weights. Our method should be readily-applicable to upcoming surveys and we make our code publicly available as part of this work., Comment: 14 pages, 9 figures, 3 tables, Accepted in MNRAS

The abstract visual reasoning ability in human intelligence benefits discovering underlying rules in the novel environment. Raven's Progressive Matrix (RPM) is a classic test to realize such ability in machine intelligence by selecting from candidates. Recent studies suggest that solving RPM in an answer-generation way boosts a more in-depth understanding of rules. However, existing generative solvers cannot discover the global concept-changing rules without auxiliary supervision (e.g., rule annotations and distractors in candidate sets). To this end, we propose a deep latent variable model for Concept-changing Rule ABstraction (CRAB) by learning interpretable concepts and parsing concept-changing rules in the latent space. With the iterative learning process, CRAB can automatically abstract global rules shared on the dataset on each concept and form the learnable prior knowledge of global rules. CRAB outperforms the baselines trained without auxiliary supervision in the arbitrary-position answer generation task and achieves comparable and even higher accuracy than the compared models trained with auxiliary supervision. Finally, we conduct experiments to illustrate the interpretability of CRAB in concept learning, answer selection, and global rule abstraction.

Accurate and reliable optical remote sensing image-based small-ship detection is crucial for maritime surveillance systems, but existing methods often struggle with balancing detection performance and computational complexity. In this paper, we propose a novel lightweight framework called \textit{HSI-ShipDetectionNet} that is based on high-order spatial interactions and is suitable for deployment on resource-limited platforms, such as satellites and unmanned aerial vehicles. HSI-ShipDetectionNet includes a prediction branch specifically for tiny ships and a lightweight hybrid attention block for reduced complexity. Additionally, the use of a high-order spatial interactions module improves advanced feature understanding and modeling ability. Our model is evaluated using the public Kaggle marine ship detection dataset and compared with multiple state-of-the-art models including small object detection models, lightweight detection models, and ship detection models. The results show that HSI-ShipDetectionNet outperforms the other models in terms of recall, and mean average precision (mAP) while being lightweight and suitable for deployment on resource-limited platforms.

We present the Aemulus $\nu$ simulations: a suite of 150 $(1.05 h^{-1}\rm Gpc)^3$ $N$-body simulations with a mass resolution of $3.51\times 10^{10} \frac{\Omega_{cb}}{0.3} ~ h^{-1} M_{\odot}$ in a $w\nu$CDM cosmological parameter space. The simulations have been explicitly designed to span a broad range in $\sigma_8$ to facilitate investigations of tension between large scale structure and cosmic microwave background cosmological probes. Neutrinos are treated as a second particle species to ensure accuracy to $0.5\, \rm eV$, the maximum neutrino mass that we have simulated. By employing Zel'dovich control variates, we increase the effective volume of our simulations by factors of $10-10^5$ depending on the statistic in question. As a first application of these simulations, we build new hybrid effective field theory and matter power spectrum surrogate models, demonstrating that they achieve $\le 1\%$ accuracy for $k\le 1\, h\,\rm Mpc^{-1}$ and $0\le z \le 3$, and $\le 2\%$ accuracy for $k\le 4\, h\,\rm Mpc^{-1}$ for the matter power spectrum. We publicly release the trained surrogate models, and estimates of the surrogate model errors in the hope that they will be broadly applicable to a range of cosmological analyses for many years to come., Comment: 37 pages, 15 figures, matching version accepted by JCAP

The growth of large-scale structure, as revealed in the anisotropic of clustering of galaxies in the low redshift Universe, provides a stringent test of our cosmological model. The strongest current constraints come from the BOSS and eBOSS surveys, with uncertainties on the amplitude of clustering of less than 10 per cent. A number of different approaches have been taken to fitting this signal, leading to apparently discrepant conclusions about the amplitude of fluctuations at late times. We compare in some detail two of the leading approaches, one based on a fitting a template cosmology whose amplitude and length scales are allowed to float with one based on a more traditional forward modeling approach, when fitting to the BOSS DR12 data. Holding the input data, scale cuts, window functions and modeling framework fixed we are able to isolate the cause of the differences and discuss the implications for future surveys.

This paper shows the achievement of a sensing and navigation system of aerial robot for measuring location and size of trees in a forest environment autonomously. Although forestry is an important industry in Japan, the working population of forestry is decreasing. Then, as an application of mechanization of forestry, we propose tree data collection system by aerial robots which have high mobility in three-dimensional space. First, we develop tree recognition and measurement method, along with algorithm to generate tree database. Second, we describe aerial robot navigation system based on tree recognition. Finally, we present an experimental result in which an aerial robot flies in a forest and collects tree data.

Entangled states are self-evidently important to a wide range of applications in quantum communication and quantum information processing. We propose an efficient and convenient two-step protocol for generating Bell states and NOON states of two microwave resonators from merely coherent states. In particular, we derive an effective Hamiltonian for resonators coupled to a superconducting $\Lambda$-type qutrit in the dispersive regime. By the excitation-number-dependent Stark shifts of the qutrit transition frequencies, we are able to individually control the amplitudes of specified Fock states of the resonators associated with relevant qutrit transition, using carefully tailored microwave drive signals. Thereby an arbitrary bipartite entangled state in Fock space can be generated by a typical evolution-and-measurement procedure. We analysis the undesired state transitions and the robustness of our protocol against the systematic errors from the microwave driving intensity and frequency, the quantum decoherence of all components, and the crosstalk of two resonators. In addition, we demonstrate that our protocol can be extended to a similar scenario with a $\Xi$-type qutrit., Comment: 13 pages, 11 figures, 1 table

Numerical simulations in cosmology require trade-offs between volume, resolution and run-time that limit the volume of the Universe that can be simulated, leading to sample variance in predictions of ensemble-average quantities such as the power spectrum or correlation function(s). Sample variance is particularly acute at large scales, which is also where analytic techniques can be highly reliable. This provides an opportunity to combine analytic and numerical techniques in a principled way to improve the dynamic range and reliability of predictions for clustering statistics. In this paper we extend the technique of Zel'dovich control variates, previously demonstrated for 2-point functions in real space, to reduce the sample variance in measurements of 2-point statistics of biased tracers in redshift space. We demonstrate that with this technique, we can reduce the sample variance of these statistics down to their shot-noise limit out to $k \sim 0.2\, h\rm Mpc^{-1}$. This allows a better matching with perturbative models and improved predictions for the clustering of e.g.~quasars, galaxies and neutral Hydrogen measured in spectroscopic redshift surveys at very modest computational expense. We discuss the implementation of ZCV, give some examples and provide forecasts for the efficacy of the method under various conditions., Comment: 17 pages main text, 5 pages of appendices, 9 figures

Human cognition has compositionality. We understand a scene by decomposing the scene into different concepts (e.g., shape and position of an object) and learning the respective laws of these concepts, which may be either natural (e.g., laws of motion) or man-made (e.g., laws of a game). The automatic parsing of these laws indicates the model's ability to understand the scene, which makes law parsing play a central role in many visual tasks. This paper proposes a deep latent variable model for Compositional LAw Parsing (CLAP), which achieves the human-like compositionality ability through an encoding-decoding architecture to represent concepts in the scene as latent variables. CLAP employs concept-specific latent random functions instantiated with Neural Processes to capture the law of concepts. Our experimental results demonstrate that CLAP outperforms the baseline methods in multiple visual tasks such as intuitive physics, abstract visual reasoning, and scene representation. The law manipulation experiments illustrate CLAP's interpretability by modifying specific latent random functions on samples. For example, CLAP learns the laws of position-changing and appearance constancy from the moving balls in a scene, making it possible to exchange laws between samples or compose existing laws into novel laws.

We propose a concise and deterministic protocol to generate NOON states in a hybrid system consisting of a superconducting qubit, a circuit resonator mode, and two magnonic modes, based on Floquet engineering. In particular, we construct a time-reversal-symmetry broken Hamiltonian for chiral state propagation of the three continuous-variable modes depending on qubit state, by the time modulation over qubit-resonator interaction and magnon frequency. Then an arbitrary magnonic NOON state can be generated by a typical preparing-and-measurement procedure. We analyze the robustness of our protocol against the systematic errors in the qubit-magnon coupling strength, the Floquet-driving intensity, the frequency mismatch of the magnons, and the counter-rotating interactions. We can obtain a high-fidelity NOON state in the presence of the quantum dissipation on all components.