Computational science

  • Article
    | Open Access

    Reservoir computers are artificial neural networks that can be trained on small data sets, but require large random matrices and numerous metaparameters. The authors propose an improved reservoir computer that overcomes these limitations and shows advantageous performance for complex forecasting tasks

    • Daniel J. Gauthier
    • , Erik Bollt
    •  & Wendson A. S. Barbosa
  • Article
    | Open Access

    Image-based simulation for obtaining physical quantities is limited by the uncertainty in the underlying image segmentation. Here, the authors introduce a workflow for efficiently quantifying segmentation uncertainty and creating uncertainty distributions of the resulting physics quantities.

    • Michael C. Krygier
    • , Tyler LaBonte
    •  & Scott A. Roberts
  • Article
    | Open Access

    Neural Networks are known to perform poorly outside of their training domain. Here the authors propose an inverse sampling strategy to train neural network potentials enabling to drive atomistic systems towards high-likelihood and high-uncertainty configurations without the need for molecular dynamics simulations.

    • Daniel Schwalbe-Koda
    • , Aik Rui Tan
    •  & Rafael Gómez-Bombarelli
  • Article
    | Open Access

    The analysis of networks and network processes can require low-dimensional representations, possible for specific structures only. The authors propose a geometric formalism which allows to unfold the mechanisms of dynamical processes propagation in various networks, relevant for control and community detection.

    • Adam Gosztolai
    •  & Alexis Arnaudon
  • Article
    | Open Access

    Quantitative methods to assess the quality of hPSC-derived organoids have not been developed. Here they present a prediction algorithm to assess the transcriptomic similarity between hPSC-derived organoids and the corresponding human target organs and perform validation on lung bud organoids, antral gastric organoids, and cardiomyocytes.

    • Mi-Ok Lee
    • , Su-gi Lee
    •  & Hyun-Soo Cho
  • Article
    | Open Access

    In-silico trials rely on virtual populations and interventions simulated using patient-specific models and may offer a solution to lower costs. Here, the authors present the flow diverter performance assessment in-silico trial, which models the treatment of intracranial aneurysms with a flow-diverting stent.

    • Ali Sarrami-Foroushani
    • , Toni Lassila
    •  & Alejandro F. Frangi
  • Article
    | Open Access

    Gravity waves are observed in Venus atmosphere, but their characteristics are not well-known. Here, the authors show spontaneous generation of gravity waves from the thermal tides in the Venus atmosphere as small-scale gravity waves are resolved in high-resolution general circulation model.

    • Norihiko Sugimoto
    • , Yukiko Fujisawa
    •  & Yoshi-Yuki Hayashi
  • Article
    | Open Access

    Generating new sensible molecular structures is a key problem in computer aided drug discovery. Here the authors propose a graph-based molecular generative model that outperforms previously proposed graph-based generative models of molecules and performs comparably to several SMILES-based models.

    • Omar Mahmood
    • , Elman Mansimov
    •  & Kyunghyun Cho
  • Article
    | Open Access

    In organic chemistry, synthetic routes for new molecules are often specified in terms of reacting molecules only. The current work reports an artificial intelligence model to predict the full sequence of experimental operations for an arbitrary chemical equation.

    • Alain C. Vaucher
    • , Philippe Schwaller
    •  & Teodoro Laino
  • Article
    | Open Access

    Transparent photodetectors based on graphene stacked vertically along the optical axis have shown promising potential for light field reconstruction. Here, the authors develop transparent photodetector arrays and implement a neural network for real-time 3D optical imaging and object tracking.

    • Dehui Zhang
    • , Zhen Xu
    •  & Theodore B. Norris
  • Article
    | Open Access

    The detection of the effects of spin symmetry in momentum distribution of an SU(N)-symmetric Fermi gas has remained challenging. Here, the authors use supervised machine learning to connect the spin multiplicity to thermodynamic quantities associated with different parts of the momentum distribution.

    • Entong Zhao
    • , Jeongwon Lee
    •  & Gyu-Boong Jo
  • Article
    | Open Access

    Machine learning algorithms offer new possibilities for automating reaction procedures. The present paper investigates automated reaction’s prediction with Molecular Transformer, the state-of-the-art model for reaction prediction, proposing a new debiased dataset for a realistic assessment of the model’s performance.

    • Dávid Péter Kovács
    • , William McCorkindale
    •  & Alpha A. Lee
  • Article
    | Open Access

    The accuracy of a machine-learned potential is limited by the quality and diversity of the training dataset. Here the authors propose an active learning approach to automatically construct general purpose machine-learning potentials here demonstrated for the aluminum case.

    • Justin S. Smith
    • , Benjamin Nebgen
    •  & Kipton Barros
  • Article
    | Open Access

    The role of children in the spread of COVID-19 is not fully understood, and the circumstances under which schools should be opened are therefore debated. Here, the authors demonstrate protocols by which schools in France can be safely opened without overwhelming the healthcare system.

    • Laura Di Domenico
    • , Giulia Pullano
    •  & Vittoria Colizza
  • Article
    | Open Access

    The dynamics of complex physical systems can be determined by the balance of a few dominant processes. Callaham et al. propose a machine learning approach for the identification of dominant regimes from experimental or numerical data with examples from turbulence, optics, neuroscience, and combustion.

    • Jared L. Callaham
    • , James V. Koch
    •  & Steven L. Brunton
  • Article
    | Open Access

    Metallization of pure hydrogen via overlapping of electronic bands requires high pressure above 3 Mbar. Here the authors study the Ba-H system and discover a unique superhydride BaH12 that contains molecular hydrogen, which demonstrates metallic properties and superconductivity below 1.5 Mbar.

    • Wuhao Chen
    • , Dmitrii V. Semenok
    •  & Tian Cui
  • Article
    | Open Access

    Standard benchmarking of single-molecule localization microscopy cannot quantify nanoscale accuracy of arbitrary datasets. Here, the authors present Wasserstein-induced flux, a method using a chosen perturbation and knowledge of the imaging system to measure confidence of individual localizations.

    • Hesam Mazidi
    • , Tianben Ding
    •  & Matthew D. Lew
  • Article
    | Open Access

    The authors present a passive meta-neural-network for real-time recognition of objects by analysis of acoustic scattering. It consists of unit cells termed meta-neurons, mimicking an analogous neural network for classical waves, and is shown to recognise handwritten digits and misaligned orbital-angular-momentum vortices.

    • Jingkai Weng
    • , Yujiang Ding
    •  & Jianchun Cheng
  • Article
    | Open Access

    Whether a turbulent flow would inevitably develop singular behavior at the smallest length scales is an ongoing intriguing debate. Using large-scale numerical simulations, Buaria et al. find an unexpected non-linear mechanism which counteracts local vorticity growth instead of enabling it.

    • Dhawal Buaria
    • , Alain Pumir
    •  & Eberhard Bodenschatz
  • Article
    | Open Access

    Accurate prediction of solubility represents a challenge for traditional computational approaches due to the complex nature of phenomena involved. Here the authors report a successful approach to solubility prediction in organic solvents and water using combination of machine learning and computational chemistry.

    • Samuel Boobier
    • , David R. J. Hose
    •  & Bao N. Nguyen
  • Article
    | Open Access

    Artificial intelligence (AI) has demonstrated promise in predicting acutekidney injury (AKI), however, clinical adoption of these models requires interpretability and transportability across sites. Here, the authors develop an AKI prediction model and a measure for model transportability across six independent health systems.

    • Xing Song
    • , Alan S. L. Yu
    •  & Mei Liu
  • Article
    | Open Access

    Distributed health data networks (DHDNs) leverage data from multiple healthcare systems, but often face major analytical challenges in the presence of missing data. This paper develops distributed multiple imputation methods that do not require sharing subject-level data across health systems.

    • Changgee Chang
    • , Yi Deng
    •  & Qi Long
  • Article
    | Open Access

    High-level ab initio quantum chemical methods carry a high computational burden, thus limiting their applicability. Here, the authors employ machine learning to generate coupled-cluster energies and forces at chemical accuracy for geometry optimization and molecular dynamics from DFT densities.

    • Mihail Bogojeski
    • , Leslie Vogt-Maranto
    •  & Kieron Burke
  • Article
    | Open Access

    RNA can be used as a programmable tool for detection of biological analytes. Here the authors use deep neural networks to predict toehold switch functionality in synthetic biology applications.

    • Nicolaas M. Angenent-Mari
    • , Alexander S. Garruss
    •  & James J. Collins
  • Article
    | Open Access

    The performance of a trained neural network may be biased even by generic features of its architecture. Yu et al. ask for the disordered lattice of atoms producing a certain wave localization and the network prefers to answer with power-law distributed displacements.

    • Sunkyu Yu
    • , Xianji Piao
    •  & Namkyoo Park
  • Article
    | Open Access

    Heterogenous ice nucleation is a ubiquitous phenomenon, but predicting the ice nucleation ability of a substrate is challenging. Here the authors develop a machine-learning data-driven approach to predict the ice nucleation ability of substrates, which is based on four descriptors related to physical properties of the interface.

    • Martin Fitzner
    • , Philipp Pedevilla
    •  & Angelos Michaelides
  • Article
    | Open Access

    Extracting central information from ever-growing data generated in our lives calls for new data mining methods. Ferreira et al. show a simple model, called chronnets, that can capture frequent patterns, spatial changes, outliers, and spatiotemporal clusters.

    • Leonardo N. Ferreira
    • , Didier A. Vega-Oliveros
    •  & Elbert E. N. Macau
  • Article
    | Open Access

    Designing efficient artificial networks able to quickly converge to optimal performance for a given task remains a challenge. Here, the authors demonstrate a relation between criticality, task-performance and information theoretic fingerprint in a spiking neuromorphic network with synaptic plasticity.

    • Benjamin Cramer
    • , David Stöckel
    •  & Viola Priesemann
  • Article
    | Open Access

    Electronic Health Records (EHR) are subject to noise, biases and missing data. Here, the authors present MixEHR, a multi-view Bayesian framework related to collaborative filtering and latent topic models for EHR data integration and modeling.

    • Yue Li
    • , Pratheeksha Nair
    •  & Manolis Kellis
  • Article
    | Open Access

    To advance the design of soft robots, novel computational frameworks that accurately model the dynamics of soft material systems are required. Here, the authors report a numerical framework for studying locomotion in limbed soft robots that is based on the discrete elastic rods algorithm.

    • Weicheng Huang
    • , Xiaonan Huang
    •  & M. Khalid Jawed
  • Article
    | Open Access

    The role of automatic electrocardiogram (ECG) analysis in clinical practice is limited by the accuracy of existing models. In that context, the authors present a Deep Neural Network (DNN) that recognizes different abnormalities in ECG recordings which matches or outperform cardiology and emergency resident medical doctors.

    • Antônio H. Ribeiro
    • , Manoel Horta Ribeiro
    •  & Antonio Luiz P. Ribeiro
  • Article
    | Open Access

    Phenomena like imitation, herding and positive feedbacks in the complex financial markets characterize the emergence of endogenous instabilities, which however is still understudied. Here the authors show that the graph-based approach is helpful to timely recognize phases of increasing instability that can drive the system to a new market configuration.

    • Alessandro Spelta
    • , Andrea Flori
    •  & Fabio Pammolli
  • Article
    | Open Access

    The authors propose a learning rule for a neuron model with dendrite. In their model, somatodendritic interaction implements self-supervised learning applicable to a wide range of sequence learning tasks, including spike pattern detection, chunking temporal input and blind source separation.

    • Toshitake Asabuki
    •  & Tomoki Fukai
  • Article
    | Open Access

    Polymer crosslinking in desalination membranes adds stability on the cost of molecular transportation rates through the membrane. Here the authors tailor crosslinking of desalination membranes to overcome the stability and transport trade-off, and demonstrate a pervaporation desalination thin-film composite membrane with high water flux.

    • Yun Long Xue
    • , Jin Huang
    •  & Pei Li
  • Article
    | Open Access

    Unconventional computing architectures might outperform current ones, but their realization has been limited to solving simple specific problems. Here, a network of interconnected Belousov-Zhabotinski reactions, operated by independent magnetic stirrers, performs encoding/decoding operations and data storage.

    • Juan Manuel Parrilla-Gutierrez
    • , Abhishek Sharma
    •  & Leroy Cronin
  • Article
    | Open Access

    Understanding the underlying mechanisms behind the successes of deep networks remains a challenge. Here, the author demonstrates an implicit regularization in training deep networks, showing that the control of complexity in the training is hidden within the optimization technique of gradient descent.

    • Tomaso Poggio
    • , Qianli Liao
    •  & Andrzej Banburski
  • Article
    | Open Access

    It is hard to design quantum neural networks able to work with quantum data. Here, the authors propose a noise-robust architecture for a feedforward quantum neural network, with qudits as neurons and arbitrary unitary operations as perceptrons, whose training procedure is efficient in the number of layers.

    • Kerstin Beer
    • , Dmytro Bondarenko
    •  & Ramona Wolf
  • Article
    | Open Access

    Chronic lymphocytic leukemia is an indolent disease, and many patients succumb to infection rather than the direct effects of the disease. Here, the authors use medical records and machine learning to predict the patients that may be at risk of infection, which may enable a change in the course of their treatment.

    • Rudi Agius
    • , Christian Brieghel
    •  & Carsten U. Niemann
  • Article
    | Open Access

    Discovery of hybrid dynamical models for real-world cyber-physical systems remains a challenge. This paper proposes a general framework for automating mechanistic modeling of hybrid dynamical systems from observed data with low computational complexity and noise resilience.

    • Ye Yuan
    • , Xiuchuan Tang
    •  & Jorge Goncalves
  • Article
    | Open Access

    The scale and dimensionality of imaging data means information is commonly overlooked. Here, using recurrent neural networks we understand temporal dependencies in hyperspectral imagery, enabling the observation of differences in ferroelectric switching mechanisms in PbZr0.2Ti0.8O3 thin films due to formation of charged domain walls.

    • Joshua C. Agar
    • , Brett Naul
    •  & Lane W. Martin