Search

The detection of phase transitions is a fundamental challenge in condensed matter physics, traditionally addressed through analytical methods and direct numerical simulations. In recent years, machine learning techniques have emerged as powerful tools to complement these standard approaches, offering valuable insights into phase and structure determination. Additionally, they have been shown to enhance the application of traditional methods. In this work, we review recent advancements in this area, with a focus on our contributions to phase and structure determination using supervised and unsupervised learning methods in several systems: (a) 2D site percolation, (b) the 3D Anderson model of localization, (c) the 2D $J_1$-$J_2$ Ising model, and (d) the prediction of large-angle convergent beam electron diffraction patterns., Comment: 15 two-column pages and 8 figures, invited review to the JPSJ issue of Special Topics "Machine Learning Physics"

We develop a new approach for efficient and scalable simulations of measurement and control of quantum systems built upon existing phase-space methods, namely the Truncated Wigner Approximation (TWA). We benchmark against existing particle-filter methods by simulating measurement based feedback cooling in a two-mode system, whose low-dimensional nature permits a computation of an exact solution. The advantage of our method is multi-mode scalability, which we demonstrate through the first successful simulation of measurement-based feedback cooling of an incoherent quasi-1D thermal ensemble to quantum degeneracy. As the underlying principle of our approach exploits a general correspondence between measurement and coherent feedback, we anticipate it is also applicable across a broad range of other quantum control scenarios., Comment: 11 pages, 4 figures. 03/09/24 Version: Minor edits in references, fixed author list formatting

Context. Most stars form in clusters or associations but only a small number of these groups are expected to remain bound for longer than a few Myr. Once star formation has ended and the molecular gas around young stellar objects has been expelled via feedback processes, most initially bound young clusters lose the majority of their binding mass and begin to disperse into the Galactic field. Aims. This process can be investigated by analysing the structure and kinematic trends in nearby young clusters, particularly expansion, the tell-tale sign that a cluster is no longer gravitationally bound but is dispersing into the field. Methods. We combine Gaia DR3 5-parameter astrometry with calibrated radial velocities for members of the nearby young cluster {\lambda} Ori (Collinder 69). Results. We characterise the plane-of-sky substructure of the cluster using the Q-parameter and Angular Dispersion parameter. We find evidence that the cluster contains significant substructure, but that this is preferentially located away from the central cluster core, which is smooth and likely remains bound. We find strong evidence for expansion in {\lambda} Ori in the plane-of-sky using a number of metrics, but also that the trends are asymmetric at the 5{\sigma} significance level. with the maximum rate of expansion being directed nearly parallel to the Galactic plane. We then invert the maximum rate of expansion of 0.144^{+0.003}_{-0.003} kms^{-1}pc^{-1} to give an expansion timescale of 6.944^{+0.148}_{-0.142} Myr, which is slightly larger than typical literature age estimates for the cluster. We also find asymmetry in the velocity dispersion, potential signatures of cluster rotation, and calculate kinematic ages for individual cluster members by tracing their motion back in time to their closest approach to the cluster center., Comment: 20 pages, 17 figures, submitted to A&A

Most reinforcement learning (RL) methods focus on learning optimal policies over low-level action spaces. While these methods can perform well in their training environments, they lack the flexibility to transfer to new tasks. Instead, RL agents that can act over useful, temporally extended skills rather than low-level actions can learn new tasks more easily. Prior work in skill-based RL either requires expert supervision to define useful skills, which is hard to scale, or learns a skill-space from offline data with heuristics that limit the adaptability of the skills, making them difficult to transfer during downstream RL. Our approach, EXTRACT, instead utilizes pre-trained vision language models to extract a discrete set of semantically meaningful skills from offline data, each of which is parameterized by continuous arguments, without human supervision. This skill parameterization allows robots to learn new tasks by only needing to learn when to select a specific skill and how to modify its arguments for the specific task. We demonstrate through experiments in sparse-reward, image-based, robot manipulation environments that EXTRACT can more quickly learn new tasks than prior works, with major gains in sample efficiency and performance over prior skill-based RL. Website at https://www.jessezhang.net/projects/extract/., Comment: 22 pages, 13 figures

Galaxy clusters are useful cosmological probes and interesting astrophysical laboratories. With growing cluster samples, a deeper understanding of the sample characteristics and improved control of systematics becomes more crucial. In this analysis we create a new and larger ACT-DR5-based thermal Sunyaev-Zel'dovich Effect (tSZE) selected galaxy cluster catalog with improved control over sample purity and completeness. We employ the red sequence based cluster redshift and confirmation tool MCMF together with optical imaging data from the Legacy Survey DR-10 and infrared data from the WISE satellite to systematicallyidentify true clusters from a new cluster candidate detection run on the ACT-DR5 dataset. The resulting ACT-DR5 MCMF sample contains 6,237 clusters with a residual contamination of 10.7%. This is an increase of 51% compared to the previous ACT-DR5 cluster catalog, making this catalog the largest tSZE-selected cluster catalog to date. The z>1 subsample contains 703 clusters, three times more than in the previous ACT-DR5 catalog. Matching the ACT-DR5 MCMF cluster catalog with a deeper tSZE sample from SPTpol 500d allows us to confirm the completeness and purity of the new ACT-DR5 MCMF sample. Cross-matching to the two largest X-ray selected cluster samples, the all-sky RASS MCMF and the half-sky eRASS1, confirms the sample purity of the RASS MCMF sample and in the case of eRASS1 reveals that 43% of the matched clusters are designated in eRASS1 as X-ray point sources rather than clusters. Cross-correlating the ACT-DR5 MCMF cluster catalog with ACT-DR6 lensing maps results in a 16.4\sigma detection of CMB lensing around the clusters, corresponding to the strongest signal found so far for a galaxy cluster sample. Repeating the measurement for the z>1 cluster subsample yields a significance of 4.3\sigma, which is the strongest CMB lensing detection in a z>1 cluster sample to date., Comment: 18 pages, 15 figures, submitted to A&A

Survival analysis, as a challenging task, requires integrating Whole Slide Images (WSIs) and genomic data for comprehensive decision-making. There are two main challenges in this task: significant heterogeneity and complex inter- and intra-modal interactions between the two modalities. Previous approaches utilize co-attention methods, which fuse features from both modalities only once after separate encoding. However, these approaches are insufficient for modeling the complex task due to the heterogeneous nature between the modalities. To address these issues, we propose a Biased Progressive Encoding (BPE) paradigm, performing encoding and fusion simultaneously. This paradigm uses one modality as a reference when encoding the other. It enables deep fusion of the modalities through multiple alternating iterations, progressively reducing the cross-modal disparities and facilitating complementary interactions. Besides modality heterogeneity, survival analysis involves various biomarkers from WSIs, genomics, and their combinations. The critical biomarkers may exist in different modalities under individual variations, necessitating flexible adaptation of the models to specific scenarios. Therefore, we further propose a Mixture of Multimodal Experts (MoME) layer to dynamically selects tailored experts in each stage of the BPE paradigm. Experts incorporate reference information from another modality to varying degrees, enabling a balanced or biased focus on different modalities during the encoding process. Extensive experimental results demonstrate the superior performance of our method on various datasets, including TCGA-BLCA, TCGA-UCEC and TCGA-LUAD. Codes are available at https://github.com/BearCleverProud/MoME., Comment: 8 + 1/2 pages, early accepted to MICCAI2024

Real-world cellular invasion processes often take place in curved geometries. Such problems are frequently simplified in models to neglect the curved geometry in favour of computational simplicity, yet doing so risks inaccuracy in any model-based predictions. To quantify the conditions under which neglecting a curved geometry are justifiable, we examined solutions to the Fisher-Kolmogorov-Petrovsky-Piskunov (Fisher-KPP) model, a paradigm nonlinear reaction-diffusion equation typically used to model spatial invasion, on an annular geometry. Defining $\epsilon$ as the ratio of the annulus thickness $\delta$ and radius $r_0$ we derive, through an asymptotic expansion, the conditions under which it is appropriate to ignore the domain curvature, a result that generalises to other reaction-diffusion equations with constant diffusion coefficient. We further characterise the nature of the solutions through numerical simulation for different $r_0$ and $\delta$. Thus, we quantify the size of the deviation from an analogous simulation on the rectangle, and how this deviation changes across the width of the annulus. Our results grant insight into when it is appropriate to neglect the domain curvature in studying travelling wave behaviour in reaction-diffusion equations., Comment: 6 pages, 4 figures

The flattening of single-particle band structures plays an important role in the quest for novel quantum states of matter due to the crucial role of interactions. Recent advances in theory and experiment made it possible to construct and tune systems with nearly flat bands, ranging from graphene multilayers and moire' materials to kagome' metals and ruthenates. While theoretical models predict exactly flat bands under certain ideal conditions, evidence was provided that these systems host high-order Van Hove points, i.e., points of high local band flatness and power-law divergence in energy of the density of states. In this review, we examine recent developments in engineering and realising such weakly dispersive bands. We focus on high-order Van Hove singularities and explore their connection to exactly flat bands. We provide classification schemes and discuss interaction effects. We also review experimental evidence for high-order Van Hove singularities and point out future research directions., Comment: Review article, to appear in Annual Review of Condensed Matter Physics

Digital trace data provide potentially valuable resources for understanding human behaviour, but their value has been limited by issues of unclear measurement. The growth of large language models provides an opportunity to address this limitation in the case of text data. Specifically, recognizing cases where their responses are a form of psychological measurement (the use of observable indicators to assess an underlying construct) allows existing measures and accuracy assessment frameworks from psychology to be re-purposed to use with large language models. Based on this, we offer four methodological recommendations for using these models to quantify text features: (1) identify the target of measurement, (2) use multiple prompts, (3) assess internal consistency, and (4) treat evaluation metrics (such as human annotations) as expected correlates rather than direct ground-truth measures. Additionally, we provide a workflow for implementing this approach., Comment: 12 pages, 2 figures, 1 table

We demonstrate an atom interferometer measurement protocol compatible with operation on a dynamic platform. Our method employs two open interferometers, derived from the same atomic source, with different interrogation times to eliminate initial velocity dependence while retaining precision, accuracy, and long term stability. We validate the protocol by measuring gravitational tides, achieving a precision of 4.5 {\mu}Gal in 2000 runs, marking the first demonstration of inertial quantity measurement with open atom interferometry that achieves long-term phase stability.

Although measuring the deterministic waveform of a weak classical force is a well-studied problem, estimating a random waveform, such as the spectral density of a stochastic signal field, is much less well-understood despite it being a widespread task at the frontier of experimental physics. State-of-the-art precision sensors of random forces must account for the underlying quantum nature of the measurement, but the optimal quantum protocol for interrogating such linear sensors is not known. We derive the fundamental precision limit, the extended channel quantum Cram\'er-Rao bound, and the optimal protocol that attains it. In the experimentally relevant regime where losses dominate, we prove that non-Gaussian state preparation and measurements are required for optimality. We discuss how this non-Gaussian protocol could improve searches for signatures of quantum gravity, stochastic gravitational waves, and axionic dark matter., Comment: 32 pages (main text: 17 pages, appendices: 15 pages), 8 figures (main text: 6 figures, appendices: 2 figures). Presently in LIGO internal review as Document No. P2400069. To be submitted to PRX on April 29th, 2024

Laser-light sails are a spacecraft concept wherein lightweight "sails" are propelled to high speeds by lasers with high intensities. The sails must comprise materials with low optical loss, to minimize the risk of laser damage. Stoichiometric silicon nitride (Si$_3$N$_4$) is a candidate material with low loss in the near infrared, but the precise absorption coefficient has not been characterized in the membrane form-factor needed for sails. We use photothermal common-path interferometry (PCI), a sensitive pump-probe technique, to measure the absorption coefficient of stoichiometric and nonstoichiometric silicon nitride. To calibrate PCI measurements of membranes, we developed a self-referencing technique where a measurement is performed twice: once on a bare membrane, and a second time with a monolayer of graphene deposited on the membrane. The absorption of the sample with graphene can be measured by both PCI and more-conventional spectroscopic techniques, enabling the calibration of the PCI measurement. We find that with an absorption coefficient of (2.09 $\pm$ 0.76) $\times$ 10$^{-2}$ cm$^{-1}$ at 1064 nm, Si$_3$N$_4$ is a suitable laser-sail material for laser intensities as high as ~10 GW/m$^{2}$, which have been proposed for some laser-sail missions, while silicon-rich SiN$_x$ (x~1), with an absorption coefficient of 7.94 $\pm$ 0.50 cm$^{-1}$, is unlikely to survive such high laser intensities., Comment: Main text + supplementary

We consider the problem of evaluating risk for a system that is modeled by a complex stochastic simulation with many possible input parameter values. Two sources of computational burden can be identified: the effort associated with extensive simulation runs required to accurately represent the tail of the loss distribution for each set of parameter values, and the computational cost of evaluating multiple candidate parameter values. The former concern can be addressed by using Extreme Value Theory (EVT) estimations, which specifically concentrate on the tails. Meta-modeling approaches are often used to tackle the latter concern. In this paper, we propose a framework for constructing a particular meta-modeling framework, stochastic kriging, that is based on EVT-based estimation for a class of coherent measures of risk. The proposed approach requires an efficient estimator of the intrinsic variance, and so we derive an EVT-based expression for it. It then allows us to avoid multiple replications of the risk measure in each design point, which was required in similar previously proposed approaches, resulting in a substantial reduction in computational effort. We then perform a case study, outlining promising use cases, and conditions when the EVT-based approach outperforms simpler empirical estimators.

The creation of large, diverse, high-quality robot manipulation datasets is an important stepping stone on the path toward more capable and robust robotic manipulation policies. However, creating such datasets is challenging: collecting robot manipulation data in diverse environments poses logistical and safety challenges and requires substantial investments in hardware and human labour. As a result, even the most general robot manipulation policies today are mostly trained on data collected in a small number of environments with limited scene and task diversity. In this work, we introduce DROID (Distributed Robot Interaction Dataset), a diverse robot manipulation dataset with 76k demonstration trajectories or 350 hours of interaction data, collected across 564 scenes and 84 tasks by 50 data collectors in North America, Asia, and Europe over the course of 12 months. We demonstrate that training with DROID leads to policies with higher performance and improved generalization ability. We open source the full dataset, policy learning code, and a detailed guide for reproducing our robot hardware setup., Comment: Project website: https://droid-dataset.github.io/

In their early, formative stages star clusters can undergo rapid dynamical evolution leading to strong gravitational interactions and ejection of ``runaway'' stars at high velocities. While O/B runaway stars have been well studied, lower-mass runaways are so far very poorly characterised, even though they are expected to be much more common. We carried out spectroscopic observations with MAG2-MIKE to follow-up 27 high priority candidate runaways consistent with having been ejected from the Orion Nebula Cluster (ONC) $>2.5$ Myr ago, based on Gaia astrometry. We derive spectroscopic youth indicators (Li \& H$\alpha$) and radial velocities, enabling detection of bona fide runaway stars via signatures of youth and 3D traceback. We successfully confirmed 10 of the candidates as low-mass Young Stellar Objects (YSOs) on the basis of our spectroscopic criteria and derived radial velocities (RVs) with which we performed 3D traceback analysis. Three of these confirmed YSOs have kinematic ejection ages $>4\:$Myr, with the oldest being 4.7~Myr. This yields an estimate for the overall formation time of the ONC to be at least $\sim 5\:$Myr, i.e., about 10 free-fall times, and with a mean star formation efficiency per free-fall time of $\bar{\epsilon}_{\rm ff}\lesssim0.05$. These results favor a scenario of slow, quasi-equilibrium star cluster formation, regulated by magnetic fields and/or protostellar outflow feedback., Comment: submitted to MNRAS, comments welcome

The ability of quantum states to be in superposition is one of the key features that sets them apart from the classical world. This `coherence' is rigorously quantified by resource theories, which aim to understand how such properties may be exploited in quantum technologies. There has been much research on what the resource theory of coherence can reveal about quantum metrology, almost all of which has been from the viewpoint of Fisher information. We prove however that the relative entropy of coherence, and its recent generalisation to POVMs, naturally quantify the performance of Bayesian metrology. In particular, we show how a coherence measure can be applied to an ensemble of states. We then prove that during parameter estimation, the ensemble relative entropy of coherence is equal to the difference between the information gained, and the optimal Holevo information. We call this relation the CXI equality. The ensemble coherence lets us visualise how much information is locked away in superposition inaccessible with a given measurement scheme, and quantify the advantage that would be gained by using a joint measurement on multiple states. Our results hold regardless of how the parameter is encoded in the state, encompassing unitary, dissipative, and discrete settings. We consider both projective measurements, and general POVMs. This work suggests new directions for research in quantum resource theories, provides a novel operational interpretation for the relative entropy of coherence and its POVM generalisation, and introduces a new tool to study the role of quantum features in metrology., Comment: 10 pages main + 4 pages supplementary

We present forecasts for constraints on the Hu \& Sawicki $f(R)$ modified gravity model using realistic mock data representative of future cluster and weak lensing surveys. We create mock thermal Sunyaev-Zel'dovich effect selected cluster samples for SPT-3G and CMB-S4 and the corresponding weak gravitational lensing data from next-generation weak-lensing (ngWL) surveys like Euclid and Rubin. We employ a state-of-the-art Bayesian likelihood approach that includes all observational effects and systematic uncertainties to obtain constraints on the $f(R)$ gravity parameter $\log_{10}|f_{R0}|$. In this analysis we vary the cosmological parameters $[\Omega_{\rm m}, \Omega_\nu h^2, h^2, A_s, n_s, \log_{10}|f_{R0}|]$, which allows us to account for possible degeneracies between cosmological parameters and $f(R)$ modified gravity. The analysis accounts for $f(R)$ gravity via its effect on the halo mass function which is enhanced on cluster mass scales compared to the expectations within general relativity (GR). Assuming a fiducial GR model, the upcoming cluster dataset SPT-3G$\times$ngWL is expected to obtain an upper limit of $\log_{10}|f_{R0}| < -5.95$ at $95\,\%$ credibility, which significantly improves upon the current best bounds. The CMB-S4$\times$ngWL dataset is expected to improve this even further to $\log_{10}|f_{R0}| < -6.23$. Furthermore, $f(R)$ gravity models with $\log_{10}|f_{R0}| \geq -6$, which have larger numbers of clusters, would be distinguishable from GR with both datasets. We also report degeneracies between $\log_{10}|f_{R0}|$ and $\Omega_{\mathrm{m}}$ as well as $\sigma_8$ for $\log_{10}|f_{R0}| > -6$ and $\log_{10}|f_{R0}| > -5$ respectively. Our forecasts indicate that future cluster abundance studies of $f(R)$ gravity will enable substantially improved constraints that are competitive with other cosmological probes., Comment: 24 pages, 12 figures, minor changes in version 2 to match version in PRD, published in PRD: https://journals.aps.org/prd/abstract/10.1103/PhysRevD.109.123503