Article
Statistics & Probability
Alice Martin, Marie-Pierre Etienne, Pierre Gloaguen, Sylvain Le Corff, Jimmy Olsson
Summary: This article proposes a new Sequential Monte Carlo algorithm to perform online estimation when certain densities are intractable in state space models. It introduces a pseudo-marginal backward importance sampling step to estimate expectations. The proposed algorithm significantly reduces computational time and broadens the class of eligible models. Its performance is assessed in different scenarios. Appendices for this article are available online.
JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS
(2023)
Article
Chemistry, Physical
Runfang Mao, Kevin D. Dorfman
Summary: We used Langevin dynamics simulations to study the knot diffusion mechanisms and time scales governing the untying of trefoil knots in DNA molecules confined in nanochannels. Knot untying follows a process of expanding and fluctuating before annihilation. The average knot size increases with chain length, and knot diffusion in nanochannel-confined DNA molecules is subdiffusive. The identified scaling exponent and knot conformations suggest a combination of self-reptation and knot region breathing.
JOURNAL OF CHEMICAL PHYSICS
(2023)
Article
Automation & Control Systems
Kaikai Zheng, Dawei Shi, Ling Shi
Summary: This work attempts to approximate a linear Gaussian system with a finite-state hidden Markov model (HMM) and address challenges in designing networked control systems. An indirect approach is developed where a state-space model (SSM) is identified for the Gaussian system and used as an emulator for learning an HMM. Parameter learning algorithms are designed based on the periodic structural characteristics of the HMM, leading to convergence and asymptotic properties of the proposed algorithms. The HMM learned using the proposed algorithms is successfully applied to event-triggered state estimation, demonstrating their validity through numerical results on model learning and state estimation.
Article
Environmental Sciences
Liming Xing, Diogo Bolster, Haifei Liu, Thomas Sherman, David H. Richter, Kyle Rocha-Brownell, Zhiming Ru
Summary: This study investigates the transport of microplastics in open-channel flows by implementing three Markov models. The models are validated using numerical simulations and laboratory experiments, demonstrating their effectiveness and high efficiency. The research provides new insights into preventing and reducing the environmental hazards of microplastics.
WATER RESOURCES RESEARCH
(2022)
Article
Engineering, Electrical & Electronic
Benjamin Cox, Victor Elvira
Summary: State-space models (SSMs) are a powerful statistical tool for modelling time-varying systems via a latent state. The estimation of the parameters in these models is challenging but essential for inference and prediction. In this work, we propose SpaRJ, a fully probabilistic Bayesian approach that explores sparsity in the transition matrix of a linear-Gaussian state-space model. This approach enhances interpretability and has strong theoretical guarantees.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Sharmin Kibria, Jinsub Kim, Raviv Raich
Summary: This study focuses on joint nonlinear state estimation with multi-period measurement vectors potentially corrupted by sparse gross errors. A nonlinear sparse optimization formulation is used for joint sparse error correction and robust state estimation, exploiting the sparsity and short-term invariance of error locations. A sequential convex approximation approach is introduced to solve the nonlinear sparse optimization problem with a convergence guarantee. An identifiability-aware version of the proposed algorithm is presented to improve the accuracy of gross error localization using a necessary rank condition for identifiable gross error matrix. The efficacy of the approach is demonstrated through application to power system nonlinear state estimation in IEEE 14-bus and 118-bus networks.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Computer Science, Interdisciplinary Applications
Abhinav Subramanian, Sankaran Mahadevan
Summary: This paper investigates the impact of errors in formulating the system physics model on the prediction accuracy of system responses, and proposes a Bayesian state estimation-based framework for estimating these model errors and extending the model error estimation to black-box models.
JOURNAL OF COMPUTATIONAL PHYSICS
(2022)
Article
Chemistry, Physical
John Strahan, Adam Antoszewski, Chatipat Lorpaiboon, Bodhi P. Vani, Jonathan Weare, Aaron R. Dinner
Summary: This study introduces a new dynamical approximation method, the dynamical Galerkin approximation, to reduce dependence on lag time selection and provide improved estimators for rates and committors. Simple procedures are also proposed for constructing suitable smoothly varying basis functions from arbitrary molecular features, and these methods are evaluated using a dataset.
JOURNAL OF CHEMICAL THEORY AND COMPUTATION
(2021)
Article
Chemistry, Multidisciplinary
Emilia P. Barros, Ozlem Demir, Jenaro Soto, Melanie J. Cocco, Rommie E. Amaro
Summary: The study demonstrates the potential significance of mutated p53 in cancer therapy by reactivating it. The mutations induce conformational changes in the important loops L6 and L1 of the p53 protein, suggesting potential new therapeutic targets.
Article
Mathematics, Interdisciplinary Applications
Xue Zhang, Chun Wang
Summary: The study utilized a two-stage divide-and-conquer strategy and proposed a novel method to correct measurement biases and errors, achieving better recovery of structural parameters.
BRITISH JOURNAL OF MATHEMATICAL & STATISTICAL PSYCHOLOGY
(2021)
Article
Biochemistry & Molecular Biology
Mueed Ur Rahman, Kaiyuan Song, Lin-Tai Da, Hai-Feng Chen
Summary: By using molecular dynamics simulations and Markov state model, the early oligomerization mechanism of A beta(16-22) peptides was investigated. It was found that the dimeric form of the peptides adopted globular random-coil or extended beta-strand like conformations, and intermolecular antiparallel beta-sheets were the most common type of secondary structure in the observed dimers. The states containing beta secondary structure and extended coiled structures were majorly involved in aggregation.
INTERNATIONAL JOURNAL OF BIOLOGICAL MACROMOLECULES
(2022)
Article
Computer Science, Information Systems
Luis Lopez Paya, Pedro Cordoba, Angela Sanchez Perez, Javier Barrachina, Manuel Benavent-Lledo, David Mulero-Perez, Jose Garcia-Rodriguez
Summary: This paper reviews and analyzes the biases in facial recognition quality estimation techniques, and experiments on several quality estimators are conducted. The research findings help identify biases within facial recognition models.
Article
Computer Science, Artificial Intelligence
Jeova F. S. Rocha Neto, Pedro Felzenszwalb, Marilyn Vazquez
Summary: This paper proposes a new approach to estimate appearance models directly from images without considering individual pixels. It introduces algebraic expressions that relate local image statistics to spatially coherent regions. Two algorithms, one using a least squares formulation and the other based on eigenvector computation, are presented for estimating appearance models. Experimental results demonstrate the effectiveness of these methods for image segmentation.
SIAM JOURNAL ON IMAGING SCIENCES
(2022)
Article
Chemistry, Analytical
Thijs Devos, Matteo Kirchner, Jan Croes, Wim Desmet, Frank Naets
Summary: The study presents a novel approach for estimating and sensor selection, capable of stabilizing the estimator Riccati equation for unobservable and non-linear system models, as well as proposing a sensor selection framework based on SVD. This method not only improves estimation performance, but also avoids the need for costly test campaigns.
Article
Mathematics
Nicholas Makumi, Romanus Odhiambo Otieno, George Otieno Orwa, Alexis Habineza
Summary: In the context of stratified sampling, a nonparametric regression technique was developed to estimate finite population quantiles in model-based frameworks. The proposed estimator was found to have good asymptotic behavior under certain conditions.
JOURNAL OF MATHEMATICS
(2022)
Article
Chemistry, Physical
Luigi Sbailo, Manuel Dibak, Frank Noe
Summary: The proposed method uses generative neural networks to connect metastable regions directly, propose new configurations in the Markov chain, and optimize the acceptance probability of large jumps between modes in the configuration space. It effectively increases the convergence speed of systems with multiple metastable states.
JOURNAL OF CHEMICAL PHYSICS
(2021)
Article
Mathematics, Applied
Marvin Luecke, Feliks Nueske
Summary: This study focuses on extracting information about dynamical systems from simulation data through modeling the Koopman operator semigroup. Recent work has been centered on deriving data-efficient representations of the Koopman operator in low-rank tensor formats and applying this to approximate the generator. The method presents consistency and complexity analysis, extensions to practical settings, and demonstrations of its applicability to benchmark numerical examples.
JOURNAL OF NONLINEAR SCIENCE
(2022)
Article
Physics, Multidisciplinary
Stefan Klus, Feliks Nueske, Sebastian Peitz
Summary: Koopman operator theory has wide applications in various research areas and this paper demonstrates the use of data-driven methods to analyze quantum physics problems and solve the Schrödinger equation, opening up a new avenue for research.
JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL
(2022)
Article
Multidisciplinary Sciences
Kalyan S. Chakrabarti, Simon Olsson, Supriya Pratihar, Karin Giller, Kerstin Overkamp, Ko On Lee, Vytautas Gapsys, Kyoung-Seok Ryu, Bert L. de Groot, Frank Noe, Stefan Becker, Donghan Lee, Thomas R. Weikl, Christian Griesinger
Summary: The study presents a theoretical and experimental framework to investigate protein binding mechanisms on sub-millisecond timescales. Using nuclear magnetic resonance and molecular dynamics simulations, the authors find that the binding mechanism between ubiquitin and the SH3 domain is based on conformational selection.
NATURE COMMUNICATIONS
(2022)
Article
Chemistry, Physical
Felix Musil, Iryna Zaporozhets, Frank Noe, Cecilia Clementi, Venkat Kapil
Summary: This research develops a method for accurately calculating vibrational spectra of molecular systems using a reduced computational cost path-integral formulation. By leveraging advances in machine-learned coarse-graining and a simple temperature elevation scheme, significant computational savings and improved accuracy are achieved compared to more expensive reference approaches. This method has the potential for routine calculations of vibrational spectra for a wide range of molecular systems with an explicit treatment of the quantum nature of nuclei.
JOURNAL OF CHEMICAL PHYSICS
(2022)
Article
Biochemistry & Molecular Biology
Tim Hempel, Simon Olsson, Frank Noe
Summary: With recent advances in structural biology, scalable molecular dynamics methods are required for large biomolecular systems. Current approaches focus on global state modeling, but are not applicable to large-scale systems. To address this, we propose using a set of coupled models to describe the local structure of molecular systems. Markov field models, including various models, are evaluated for their use in computational molecular biology.
CURRENT OPINION IN STRUCTURAL BIOLOGY
(2022)
Article
Mathematics, Applied
Feliks Nueske, Sebastian Peitz, Friedrich Philipp, Manuel Schaller, Karl Worthmann
Summary: In this paper, probabilistic bounds for the approximation error and the prediction error are derived for dynamical (control) systems using the Koopman operator. The analysis is extended to (stochastic) nonlinear control-affine systems. A previously proposed approach is proven to be effective in avoiding the curse of dimensionality. The effectiveness of the approach is demonstrated through comparisons with state-of-the-art techniques.
JOURNAL OF NONLINEAR SCIENCE
(2023)
Article
Multidisciplinary Sciences
Andreas Mardt, Tim Hempel, Cecilia Clementi, Frank Noe
Summary: This study addresses the challenge of modeling the dynamics of large molecular systems by introducing a method that simultaneously decomposes and models the system, providing an effective summary of the complex dynamics. While the issue of learning the dynamical coupling between subsystems still remains, it is a significant step towards learning Ising models of large molecular complexes from simulation data.
NATURE COMMUNICATIONS
(2022)
Article
Chemistry, Multidisciplinary
Wangfei Yang, Clark Templeton, David Rosenberger, Andreas Bittracher, Feliks Nueske, Frank Noe, Cecilia Clementi
Summary: The aim of molecular coarse-graining approaches is to simulate the physical properties of a molecular system more efficiently using a lower-resolution model. In this article, we argue that accurate coarse-grained models in soft matter contexts should accurately capture rare-event transitions to reproduce the system's long-time dynamics. We propose a bottom-up coarse-graining scheme that preserves the relevant slow degrees of freedom, and demonstrate its effectiveness in three systems of increasing complexity.
ACS CENTRAL SCIENCE
(2023)
Article
Chemistry, Physical
Feliks Nueske, Stefan Klus
Summary: The use of random Fourier features as a stochastic approximation method allows for more efficient estimation of slow kinetic processes.
JOURNAL OF CHEMICAL PHYSICS
(2023)
Article
Chemistry, Physical
Z. Schaetzle, P. B. Szabo, M. Mezera, J. Hermann, F. Noe
Summary: Computing accurate and efficient approximations to solve the Schrödinger equation in computational chemistry has been a challenge for decades. Quantum Monte Carlo methods, with their highly parallel and scalable algorithm, show promise in achieving high accuracy in a variety of molecular systems. The use of machine-learned parametrizations, relying on neural networks as universal function approximators, has further improved the accuracy of these methods. The development of software libraries like DEEPQMC aims to provide a common framework for future investigations and make this field accessible to practitioners from both the quantum chemistry and machine learning communities.
JOURNAL OF CHEMICAL PHYSICS
(2023)
Review
Chemistry, Multidisciplinary
Jan Hermann, James Spencer, Kenny Choo, Antonio Mezzacapo, W. M. C. Foulkes, David Pfau, Giuseppe Carleo, Frank Noe
Summary: Deep learning methods have surpassed human capabilities in pattern recognition and data processing, and have become increasingly important in scientific discovery. In molecular science, a key application of machine learning is to learn potential energy surfaces or force fields from ab initio solutions of the electronic Schrodinger equation obtained with various quantum chemistry methods. This review discusses a complementary approach that uses machine learning to directly solve quantum chemistry problems from first principles, focusing on quantum Monte Carlo methods with neural-network ansatzes to solve the electronic Schrodinger equation.
NATURE REVIEWS CHEMISTRY
(2023)
Article
Physics, Multidisciplinary
Paolo A. Erdman, Alberto Rolandi, Paolo Abiuso, Marti Perarnau-Llobet, Frank Noe
Summary: The full optimization of a quantum heat engine requires trade-offs between power, efficiency, and fluctuations. A general framework is proposed to identify Pareto-optimal cycles that balance these objectives. Reinforcement learning is used to find the Pareto front of a quantum dot-based engine, revealing abrupt changes in optimal cycles when switching between optimizing two and three objectives. Analytical results accurately describe different regions of the Pareto front in fast- and slow-driving regimes.
PHYSICAL REVIEW RESEARCH
(2023)
Article
Physics, Multidisciplinary
Manuel Dibak, Leon Klein, Andreas Kraemer, Frank Noe
Summary: Boltzmann generators solve the sampling problem in many-body physics by combining a normalizing flow and a statistical reweighting method. Temperature steerable flows (TSFs) are proposed to generate a family of probability densities parametrized by a choosable temperature parameter, allowing for sampling of a physical system across multiple thermodynamic states.
PHYSICAL REVIEW RESEARCH
(2022)
Article
Computer Science, Artificial Intelligence
Soren Ager Meldgaard, Jonas Koehler, Henrik Lund Mortensen, Mads-Peter Christiansen, Frank Noe, Bjork Hammer
Summary: This study proposes a reinforcement learning approach for generating molecules in chemical space and predicting their stability using quantum chemistry. By combining imitation learning and reinforcement learning, the sample efficiency is improved, and low energy molecules are generated under different stoichiometries conditions.
MACHINE LEARNING-SCIENCE AND TECHNOLOGY
(2022)