A Variational Approach to Modeling Slow Processes in Stochastic Dynamical Systems
Open accessThe DOI matches its Crossref record; 49 of the 50 references it cites were found in a registry or an index.
Abstract
The slow processes of metastable stochastic dynamical systems are difficult to access by direct numerical simulation due to the sampling problems. Here, we suggest an approach for modeling the slow parts of Markov processes by approximating the dominant eigenfunctions and eigenvalues of the propagator. To this end, a variational principle is derived that is based on the maximization of a Rayleigh coefficient. It is shown that this Rayleigh coefficient can be estimated from statistical observables that can be obtained from short distributed simulations starting from different parts of state space. The approach forms a basis for the development of adaptive and efficient computational algorithms for simulating and analyzing metastable Markov processes while avoiding the sampling problem. Since any stochastic process with finite memory can be transformed into a Markov process, the approach is applicable to a wide range of processes relevant for modeling complex real-world phenomena.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“It was recently discovered that the Rayleigh-Ritz variational principle—a fundamental concept in quantum chemistry—has an analog in molecular conformation dynamics, and that the eigenfunction approximation problem that appears in constructing MSMs can be cast as a generalized eigenproblem [14].”
From Markov state models of biomolecular conformational dynamics (2014) · cites this paper as [14]
The cited source says what the claim attributes to it.
In the citing paper
“This formulation implies that MSM-derived relaxation timescales are always underestimated except when linear combinations of the true eigenfunctions of the Markov operator are used as basis functions.”
From Markov state models of biomolecular conformational dynamics (2014) · cites this paper as [14]
The cited source says what the claim attributes to it.
In the citing paper
“Standard “crisp paritioning” MSMs based on clustering simulation data were shown to be a special case in which the basis set used in variational optimization is chosen to be a set of functions that are constant on the discrete MSM states [14].”
From Markov state models of biomolecular conformational dynamics (2014) · cites this paper as [14]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Reaction-rate theory: fifty years after Kramers
Hänggi, P., Talkner, P., & Borkovec, M. (1990) · Reviews of Modern Physics · doi:10.1103/revmodphys.62.251 · Cited by 6,154 · cited 1×
“71 03 v1 [ m at hph ] 2 9 N ov 2 01 2”
The Activated Complex in Chemical Reactions
Eyring, H. (1935) · The Journal of Chemical Physics · doi:10.1063/1.1749604 · Cited by 5,828 · cited 1×
“Examples are rate theories that describe the passage rate of a process across a surface that separates metastable states [15, 20], pathway-based theories and methods that describe the transition dynamics of a system from a subset A to a subset B of state space [14, 3, 25] and network-based approaches that attempt to c…”
BackgroundNew Developments in Molecular Orbital Theory
Roothaan, C. C. J. (1951) · Reviews of Modern Physics · doi:10.1103/revmodphys.23.69 · Cited by 5,519
Szabó, A. (1982) · Cited by 2,620
Random perturbations of dynamical systems
(1997) · Translations of Mathematical Monographs · doi:10.1090/mmono/161/03 · Cited by 2,584
T RANSITION P ATH S AMPLING : Throwing Ropes Over Rough Mountain Passes, in the Dark
Bolhuis, P. G., Chandler, D., Dellago, C., & Geissler, P. L. (2002) · Annual Review of Physical Chemistry · doi:10.1146/annurev.physchem.53.082301.113146 · Cited by 1,972
Atomic-level characterization of the structural dynamics of proteins
Shaw, D. E., Maragakis, P., Lindorff-Larsen, K., Piana, S., Dror, R. O., Eastwood, M. P., Bank, J. A., Jumper, J. M., Salmon, J. K., Shan, Y., et al. (2010) · Science · doi:10.1126/science.1187409 · Cited by 1,923 · cited 1×
“An extreme example is the atomistic simulation of solvated biomolecules which would require the propagation of a system with 10(4) − 10(6) particles for 10(7) − 10(10) time-steps, a task that is intractable or hardly tractable even with specialpurpose supercomputers [39].”
Random Perturbations of Dynamical Systems
Freidlin, M. I., & Wentzell, A. D. (1984) · Grundlehren der mathematischen Wissenschaften · doi:10.1007/978-1-4684-0176-9 · Cited by 1,622 · cited 1×
“In these cases, the mathematical analysis can be based on large deviation estimates and variational principles [16, 14].”
BackgroundSǐnanoğlu, O., & DeRocco, A. G. (1967) · doi:10.1119/1.1973756 · Cited by 898 · cited 2×
“For this, the formalism introduced above is used to formulate the Rayleigh variational principle used in quantum mechanics [43] for Markov processes.”
BackgroundMethodConstructing the equilibrium ensemble of folding pathways from short off-equilibrium simulations
Noé, F., Schütte, C., Vanden-Eijnden, E., Reich, L., & Weikl, T. R. (2009) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.0905466106 · Cited by 820
Self-citation
Research works printed but not linked: 4
- W. E, 2004, Berlin, 38
- Huisinga W., 2006, Berlin, 167
- Schütte C., 2009, EMS Publishing House, 297
- D. J. Wales, Energy Landscapes , Cambridge University Press, Cambridge, UK, 2003.
Cited by
Items 201–300 of 324
Learning-based robust model predictive control with data-driven Koopman operators
Wang, M., Lou, X., & Cui, B. (2023) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-023-01834-5 · Cited by 11
“Extended dynamical mode decomposition (EDMD) [11] and the variational approach of conformation dynamics [12] extend the DMD method using nonlinear measurements.”
MethodA Note on Learning Rare Events in Molecular Dynamics using LSTM and TransformerPreprint
Zeng, W., Cao, S., Huang, X., & Yao, Y. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.06573 · Cited by 24
Sparse feature map-based Markov models for nonlinear fluid flows
Jayaraman, B., Chen, L., Whitman, J., & Chowdhary, G. V. (2019) · Computers & Fluids · doi:10.1016/j.compfluid.2019.104252 · Cited by 13
Wang, X., Dong, Y., Yang, J., Liu, Z.-P., & Lu, J.-S. (2024) · Frontiers of Environmental Science and Engineering · doi:10.1007/s11783-024-1814-5 · Cited by 10
Enhanced Sampling with Machine Learning: A ReviewPreprint
Mehdi, S., Smith, Z., Herron, L., Zou, Z., & Tiwary, P. (2023) · Annual review of physical chemistry (Print) · doi:10.48550/arxiv.2306.09111 · Cited by 10
“A well-established technique for estimating the leading eigenfunctions of the transfer operator is the variational approach for conformation (VAC) dynamics (59, 60, 36).”
MethodMolecular Latent Space Simulators for Distributed and Multimolecular Trajectories
Jones, M. S., McDargh, Z. A., Wiewiora, R., Izaguirre, J. A., Xu, H., & Ferguson, A. L. (2023) · The Journal of Physical Chemistry A · doi:10.1021/acs.jpca.3c01362 · Cited by 14
Directed kinetic transition network model
Zhou, H.-Y., Wang, F., Bennett, D. I. G., & Tao, P. (2019) · The Journal of Chemical Physics · doi:10.1063/1.5110896 · Cited by 8
Brunton, S. L., & Kutz, J. N. (2019) · Cambridge University Press eBooks · doi:10.1017/9781108380690.011 · Cited by 4
Husic, B. E., Schlueter-Kuck, K., & Dabiri, J. O. (2019) · PLoS ONE · doi:10.1371/journal.pone.0212442 · Cited by 11
Brunton, S. L., & Kutz, J. N. (2019) · Cambridge University Press eBooks · doi:10.1017/9781108380690.009 · Cited by 8
A Control Theory of Predictability in Latent World Models
You, H., Zhang, Y.-G., Ran, M.-H., Yang, Z., Zhang, Z., Xue, W., Song, J., Tian, X.-M., & Guo, Y.-K. (2026) · arXiv.org · doi:10.48550/arxiv.2607.10362 · Cited by 9
“…linear Koopman operator underlies a large body of data-driven modeling [Brunton et al., 2016, Mezi´c, 2005, Lusch et al., 2018, Mardt et al., 2018, Noé and Nüske, 2013, Otto and Rowley, 2019], with recent attention to rigorous spectral computation [Colbrook et al., 2023, Colbrook and Townsend,…”
Dynamics Harmonic Analysis of Robotic Systems: Application in Data-Driven Koopman Modelling
Apraez, D. F. O., Kostic, V. R., Turrisi, G., Novelli, P., Mastalli, C., Semini, C., & Pontil, M. (2023) · Conference on Learning for Dynamics & Control · doi:10.48550/arxiv.2312.07457 · Cited by 9
“Related work For a data-driven approximation of Koopman operators No´e and Nuske (2013), Lusch et al. (2018), and Kostic et al. (2023) introduce symmetry-agnostic deep learning algorithms, which can be adapted to exploit DHA, as proposed in sec.”
MethodZhang, W., & Schütte, C. (2023) · PAMM · doi:10.1002/pamm.202300189 · Cited by 13
Exploring glassy dynamics with Markov state models from graph dynamical neural networks
Soltani, S., Sinclair, C. W., & Rottler, J. (2022) · Physical review. E · doi:10.1103/physreve.106.025308 · Cited by 12
Fourier and Wavelet Transforms
Brunton, S. L., & Kutz, J. N. (2019) · Cambridge University Press eBooks · doi:10.1017/9781108380690.003 · Cited by 11
Mittal, S., & Shukla, D. (2018) · Journal of Physical Chemistry B · doi:10.1021/acs.jpcb.8b07076 · Cited by 9
Enhanced Sampling for Efficient Learning of Coarse-Grained Machine Learning Potentials
Chen, W., Görlich, F., Fuchs, P., & Zavadlav, J. (2025) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.5c01712 · Cited by 9
Variability of SST through Koopman Modes
Navarra, A., Tribbia, J., Klus, S., & Sánchez, P. L. (2024) · Journal of Climate · doi:10.1175/jcli-d-23-0335.1 · Cited by 10
Goyal, P., Yıldız, S., & Benner, P. (2023) · Machine Learning: Science and Technology · doi:10.1088/2632-2153/adb9b5 · Cited by 8
“Recent efforts have focused on learning linear representations for systems with a continuous spectrum, though the eigenvalues of these linear representations are determined by a nonlinear operator (see [43]).”
Folding-upon-binding pathways of an intrinsically disordered protein from a deep Markov state model
Sisk, T., & Robustelli, P. J. (2023) · bioRxiv · doi:10.1101/2023.07.21.550103 · Cited by 8
Approximating Matrix Eigenvalues by Subspace Iteration with Repeated Random Sparsification
Greene, S. M., Webber, R. J., Berkelbach, T. C., & Weare, J. (2022) · SIAM Journal on Scientific Computing · doi:10.1137/21m1422513 · Cited by 5
“That scheme is an early instance of an algorithm often referred to as the Variational Approach to Conformal dynamics (VAC) [41, 61, 35].”
BackgroundMethodError Bounds for Dynamical Spectral Estimation
Webber, R. J., Thiede, E. H., Dow, D., Dinner, A. R., & Weare, J. (2021) · SIAM Journal on Mathematics of Data Science · doi:10.1137/20m1335984 · Cited by 10
Pattis, J. G., & May, E. R. (2020) · Structure · doi:10.1016/j.str.2020.03.002 · Cited by 10
Narayan, B., Herbert, C., Yuan, Y., Rodriguez, B. J., Brooks, B. R., & Buchete, N. (2018) · The Journal of Chemical Physics · doi:10.1063/1.5027580 · Cited by 8
Sparse Convolution-based Markov Models for Nonlinear Fluid Flows
Jayaraman, B., Chen, L., Whitman, J., & Chowdhary, G. V. (2018) · doi:10.48550/arxiv.1803.08222 · Cited by 8
Sadiq, S. K., Muñiz Chicharro, A., Friedrich, P., & Wade, R. C. (2021) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.1c00673 · Cited by 8
Sadiq, S. K., Chicharro, A. M., Friedrich, P., & Wade, R. C. (2021) · doi:10.1101/2021.06.22.449380 · Cited by 8
tICA-Metadynamics for Identifying Slow Dynamics in Membrane Permeation
Oh, M., da Hora, G. C. A., & Swanson, J. M. J. (2023) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.3c00526 · Cited by 14
Choice of damping coefficient in Langevin dynamics
Skeel, R. D., & Hartmann, C. (2021) · The European Physical Journal B · doi:10.1140/epjb/s10051-021-00182-z · Cited by 11
7 Data-driven methods for reduced-order modeling
Brunton, S. L., & Kutz, J. N. (2020) · De Gruyter eBooks · doi:10.1515/9783110671490-007 · Cited by 13
“DMD is a representation based on linear observables [109], which has been extended to nonlinear observables in the extended DMD (eDMD) [144] and the variational approach of conformation dynamics [99, 100].”
BackgroundTanaka, S. (2016) · The Journal of Chemical Physics · doi:10.1063/1.4942861 · Cited by 6
Machine Learning Driven Advances in Molecular Dynamics of Bulk and Interfacial Aqueous Systems
Wang, R., Meraz, V. J., & Tiwary, P. (2026) · Chemical Reviews · doi:10.1021/acs.chemrev.5c00708 · Cited by 12
How deep is your network? Deep vs. shallow learning of transfer operatorsPreprint
Tabish, M., Leimkuhler, B., & Klus, S. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.19930 · Cited by 8
“The network is trained by optimizing the VAMP-2 score [16], a loss function based on the variational principle for transfer operators [17].”
Prabhakar, P. R., Ray, D., & Andricioaei, I. (2025) · The Journal of Chemical Physics · doi:10.1063/5.0276623 · Cited by 6
Efficient Parametric SVD of Koopman Operator for Stochastic Dynamical SystemsPreprint
Jeong, M., Ryu, J. J., Yun, S.-Y., & Wornell, G. W. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.07222 · Cited by 3
“For the case of reversible processes, Noé and Nüske [26] proposed a method to approximate eigenfunctions from time-series data, followed by the extended DMD (EDMD) [41].”
Wang, D.-D., & Tiwary, P. (2024) · arXiv.org · doi:10.48550/arxiv.2406.14839 · Cited by 6
“24 This is unlike commonly used dynamics-based dimension reduction methods such as tICA, 35,36 which focus on learning a given number of slow processes.”
Kabir, K. L., Akhter, N., & Shehu, A. (2019) · Journal of Bioinformatics and Computational Biology · doi:10.1142/s0219720019400146 · Cited by 9
Effective equations for reaction coordinates in polymer transport
Baldovin, M., Cecconi, F., & Vulpiani, A. (2020) · Journal of Statistical Mechanics Theory and Experiment · doi:10.1088/1742-5468/ab5368 · Cited by 6
Automatic Feature Selection in Markov State Models Using Genetic Algorithm
Chen, Q., Feng, J., Mittal, S., & Shukla, D. (2018) · The Journal of Computational Science Education · doi:10.22369/issn.2153-4136/9/2/2 · Cited by 6
“The theoretical upper limit of GMRQ score is 6 in all cases[32–34], due to the fact that the number of MSM timescales is chosen to be 5 in the MSM settings.”
BackgroundMethodInfluentialNote: Variational encoding of protein dynamics benefits from maximizing latent autocorrelation
Wayment-Steele, H. K., & Pande, V. S. (2018) · The Journal of Chemical Physics · doi:10.1063/1.5043303 · Cited by 10
Markov State Models to Elucidate Ligand Binding Mechanism
Ge, Y., & Voelz, V. A. (2021) · Methods in molecular biology · doi:10.1007/978-1-0716-1209-5_14 · Cited by 16
Markov State Model Approach to Simulate Self-Assembly
Trubiano, A., & Hagan, M. F. (2024) · Physical Review X · doi:10.1103/physrevx.14.041063 · Cited by 7
Characterizing Structural and Kinetic Ensembles of Intrinsically Disordered Proteins Using Writhe
(2025) · doi:10.1021/acs.jctc.5c01133 · Cited by 5
“15,16,20 Canonical correlation analysis (CCA) can be viewed as a dimensionality reduction that relates two sets of variables (or data sets) by finding orthogonal transformations (or linear combinations) of each that maximize their correlation.”
Error Breakdown and Sensitivity Analysis of Dynamical Quantities in Markov State Models
Tuchkov, Y., Evans, L., Hanson, S. M., & Thiede, E. H. (2025) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.5c01143 · Cited by 5
Pérez-Lemus, G. R., Xu, Y., Jin, Y., Rico, P. F. Z., & Pablo, J. (2024) · The Journal of Chemical Physics · doi:10.1063/5.0237399 · Cited by 9
Wang, D., & Tiwary, P. (2024) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.4c00919 · Cited by 11
Zhang, W., & Schütte, C. (2025) · Multiscale Modeling and Simulation · doi:10.1137/24m1658917 · Cited by 3
“…class of these data-driven methods are based on the theory of transfer operators [53, 54] or Koopman operators [11], in which the dynamics of a underlying system is analyzed using the eigenvalues and the corresponding eigenfunctions of the transfer (or Koopman) operator associated to the system.”
An Information Bottleneck Approach for Markov Model Construction
Wang, D.-D., Qiu, Y., Beyerle, E., Huang, X.-H., & Tiwary, P. (2024) · arXiv.org · doi:10.48550/arxiv.2404.02856 · Cited by 5
“These metrics encompass various model aspects, including enhancing the metastability of macrostates in the kinetic model 7,15 , optimizing approximations of the principle eigenmodes of dynamics 16 , and maximizing the model’s capability to capture the leading slowest dynamics 10,17 .”
BackgroundKabir, K. L., Ma, B., Nussinov, R., & Shehu, A. (2022) · Biomolecules · doi:10.3390/biom12071011 · Cited by 6
“One looks for an exponential decay to the system equilibrium; the implied timescales plot needs to exhibit convergence within a few steps [58].”
BackgroundTransferring chemical and energetic knowledge between molecular systems with machine learning
Heydari, S., Raniolo, S., Livi, L., & Limongelli, V. (2023) · Communications Chemistry · doi:10.1038/s42004-022-00790-5 · Cited by 8
“Then, we performed a Time-lagged Independent Component Analysis (TICA) [36, 39] using the three couples of φ and ψ dihedrals in tri-alanine to construct two optimized CVs having the following construction:”
MethodGu, H.-L., Wang, W., Cao, S.-Q., Unarta, I., Yao, Y., Sheong, F., & Huang, X.-H. (2021) · bioRxiv · doi:10.1101/2021.08.04.455071 · Cited by 5
The Dynamic Conformational Landscapes of the Protein Methyltransferase SETD8Preprint
Chen, S., Wiewiora, R. P., Meng, F.-W., Babault, N., Ma, A., Yu, W.-Y., Qian, K., Hu, H., Zou, H., Wang, J., et al. (2018) · bioRxiv · doi:10.1101/438994 · Cited by 5
Abramyan, A. M., Quick, M., Xue, C., Javitch, J. A., & Shi, L. (2018) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.8b00175 · Cited by 4
Characterizing structural and kinetic ensembles of intrinsically disordered proteins using writhe
Sisk, T. R., Olsson, S., & Robustelli, P. (2025) · doi:10.1101/2025.04.26.650781 · Cited by 5
Advancements in characterization of protein dynamics with machine learning
Madani, M., & Tarakanova, A. (2026) · npj Soft Matter · doi:10.1038/s44431-026-00025-w · Cited by 6
“Taken together, these developments illustrate a coherent progression: from linear spectral methods (VAC/TICA) to generalized variational frameworks like VAMPNet, and onward to information ‑ theoretic and deep generative encoders such as PIB that target predictive, low ‑ dimensional kinetics.”
InfluentialLearning Graphons From Data: Random Walks, Transfer Operators, and Spectral Clustering
Klus, S., & Bramburger, J. J. (2026) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2026.3682885 · Cited by 4
“This means that it almost surely holds that”
Robust Data-Driven Attitude Control for Satellite Using Koopman Operator Theory
Chen, A., Bu, S.-C., Cui, L., Zhang, W., Li, J.-T., & Wang, Z.-H. (2024) · 2024 IEEE 13th Data Driven Control and Learning Systems Conference (DDCLS) · doi:10.1109/ddcls61622.2024.10606625 · Cited by 4
Representation of Protein Dynamics Disentangled by Time-structure-based Prior
Ishizone, T., Matsunaga, Y., Fuchigami, S., & Nakamura, K. (2023) · bioRxiv · doi:10.1101/2023.09.13.557264 · Cited by 4
Coarse-Graining and Forecasting Atomic Material Simulations with Descriptors
Swinburne, T. D. (2023) · Physical Review Letters · doi:10.1103/physrevlett.131.236101 · Cited by 9
Stochastic Lag Time Parameterization for Markov State Models of Protein Dynamics.
Gong, S., He, X., Meng, Q., Ma, Z.-M., Shao, B., Wang, T., & Liu, T.-Y. (2022) · Journal of Physical Chemistry B · doi:10.1021/acs.jpcb.2c03711 · Cited by 5
Machine Learning in Molecular Dynamics Simulations of Biomolecular Systems
Kolloff, C., & Olsson, S. (2022) · doi:10.48550/arxiv.2205.03135 · Cited by 4
“However, there have been approaches introduced that make use of kernel methods to deal with non-linear data manifolds [Noé and Nüske, 2013].”
MethodMarkov State Models of protein-protein encounters
Olsson, S. (2021) · doi:10.48550/arxiv.2105.02767 · Cited by 4
“VAC uses that the Eigenvalues of Pτ are bounded and the Eigenfunctions form an orthonormal basis.”
MethodInfluentialSingular Value Decomposition (SVD)
Brunton, S. L., & Kutz, J. N. (2019) · Cambridge University Press eBooks · doi:10.1017/9781108380690.002 · Cited by 74
Modelling Intrinsically Disordered Protein Dynamics as Networks of Transient Secondary StructurePreprint
Wayment-Steele, H. K., Hernández, C. X., & Pande, V. (2018) · bioRxiv · doi:10.1101/377564 · Cited by 4
Maximum margin clustering for state decomposition of metastable systems
Wu, H. (2015) · Neurocomputing · doi:10.1016/j.neucom.2014.12.093 · Cited by 4
Functional Regulation of Aquaporin Dynamics by Lipid Bilayer Composition
Nguyen, A. T. P., Weigle, A. T., & Shukla, D. (2023) · doi:10.1101/2023.07.20.549977 · Cited by 21
Song, J., Sun, Q., Skirtach, A. G., & Parakhonskiy, B. V. (2025) · Advances in Colloid and Interface Science · doi:10.1016/j.cis.2025.103732 · Cited by 8
Data-driven approximation of transfer operators for mean-field stochastic differential equations
Ioannou, E., Klus, S., & dos Reis, G. (2025) · arXiv.org · doi:10.48550/arxiv.2509.09891 · Cited by 3
“While transfer operators associated with conventional SDEs and numerical methods to learn these operators from data are well-understood, see, e.g., [27, 28, 21], there is a need to generalize these approaches to mean-field SDEs.”
Data-driven system identification using quadratic embeddings of nonlinear dynamics
Klus, S., & N’konzi, J.-P. N. (2025) · Physica D Nonlinear Phenomena · doi:10.1016/j.physd.2025.134839 · Cited by 3
“An open question—regarding not only QENDy and SINDy, but also methods for approximating the Koopman operator such as Extended Dynamic Mode Decomposition (EDMD) and its extensions [31, 5, 6]—is the choice of suitable basis functions.”
On the Correspondence between Gaussian Processes and Geometric Harmonics
Dietrich, F., Bello-Rivas, J., & Kevrekidis, I. (2021) · arXiv.org · doi:10.48550/arxiv.2110.02296 · Cited by 3
A Weak Characterization of Slow Variables in Stochastic Dynamical Systems
Bittracher, A., & Schütte, C. (2020) · Studies in systems, decision and control · doi:10.1007/978-3-030-51264-4_6 · Cited by 3
“This is a sensible and commonly made assumption [34, 32, 31, 26], as it holds true for many difference classes of chemical and molecular reactions.”
BackgroundZhang, M., Wang, Y., Keller, B. G., & Wu, H. (2026) · The Journal of Chemical Physics · doi:10.1063/5.0335247 · Cited by 1
“…as a combined ensemble average over both the phase space and the path space 33 : Here, µ ( ω ) is the conditional probability density of observing a path ω given its starting point at x 0 , m is the total number of paths collected, x are the starting and end points of the k th path sample, 1 S i…”
Accurate and robust analysis of molecular kinetics with random features
Sprink, H., Zhu, Y., Mey, A. S. J. S., & Nüske, F. (2025) · MPG.PuRe (Max Planck Society) · doi:10.5281/zenodo.17950457
Self-citationAdaptive tensor train metadynamics for high-dimensional free energy exploration
Strand, N. E., Yang, S.-Y., Khoo, Y., & Dinner, A. (2026) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.6c00487 · Cited by 2
Liao, J., Gao, W., Xu, K., Han, W., Li, Y., & Zhou, M. (2026) · npj Soft Matter · doi:10.1038/s44431-025-00017-2 · Cited by 3
Building multiscale Markov state models by systematic mapping of temporal communities
Nitskansky, N., Clein, K., & Raveh, B. (2025) · Bioinformatics · doi:10.1093/bioinformatics/btaf585 · Cited by 1
Learning Collective Variables from BioEmu with Time-Lagged GenerationPreprint
Park, S., Seong, K., Yang, S., Gómez‐Bombarelli, R., & Ahn, S. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.07390 · Cited by 2
Leveraging Transformer Models to Capture Multi-Scale Dynamics in Biomolecules by Nano-GPT
Zeng, W., Zhang, L., & Yao, Y. (2025) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.5c00180 · Cited by 1
“In this regard, extracting the long-term pattern from short simulations is critical to address the challenges in predicting biomolecular dynamics 3–5 .”
SOMMD: an R package for the analysis of molecular dynamics simulations using self-organizing map
Motta, S., Callea, L., Mulla, S. I., Davoudkhani, H., Bonati, L., & Pandini, A. (2025) · Bioinformatics · doi:10.1093/bioinformatics/btaf308 · Cited by 3
“…for simulation data include principal component analysis (PCA), time-lagged independent component analysis (TICA) (Molgedey and Schuster 1994, No � e and N € uske 2013), and self-organizing maps (SOMs) (Kohonen 1990, Shao et al. 2007, Fraccalvieri et al. 2011, Kohonen 2013, Bouvier et al. 2015).”
Ahdab, D. E., Lagardère, L., Hobaika, Z., Inizan, T. J., Célerse, F., Gresh, N., Maroun, R., & Piquemal, J.-P. (2025) · bioRxiv · doi:10.1101/2024.08.28.610081 · Cited by 2
Jones, M. S., Shmilovich, K., & Ferguson, A. L. (2024) · The Journal of Physical Chemistry A · doi:10.1021/acs.jpca.4c05389 · Cited by 2
Conformational Fluctuations in β2-Microglubulin Using Markov State Modeling and Molecular Dynamics.
Ghorbani, M., Brooks, B., & Klauda, J. B. (2023) · Journal of Physical Chemistry B · doi:10.1021/acs.jpcb.3c02473 · Cited by 2
George, A., Purnaprajna, M., & Athri, P. (2020) · PeerJ Physical Chemistry · doi:10.7717/peerj-pchem.9 · Cited by 3
“In our study, we explore a method for improving MSM efficiency, as measured by the established Generalized Matrix Rayleigh Quotient (GMRQ) scoring system ( Noé & Nuske, 2013 ), by combining both Laplacian scoring (to provide a heuristic initialization) and GAs.”
BackgroundMethodInfluentialBrunton, S. L., & Kutz, J. N. (2019) · Cambridge University Press eBooks · doi:10.1017/9781108380690.010 · Cited by 3
Romano, P., & Guenza, M. (2017) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.7b00261 · Cited by 3
Numerical approximation of the Koopman-von Neumann equation: Operator learning and quantum computing
Klus, S., Nüske, F., & Gelß, P. (2026) · Physical review. E · doi:10.1103/23fl-9hw8 · Cited by 1
“One of the most frequently used methods for approximating transfer operators is extended dynamic mode decomposition and its various extensions (and predecessors, which have been developed independently of each other), see [27, 28, 29, 18, 30, 31, 32].”
Self-citationZhang, Z., Shao, X.-G., & Cai, W.-S. (2026) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.6c00310 · Cited by 1
Random Functions as Data Compressors for Machine Learning of Molecular Processes
Debnath, J., & Hummer, G. (2025) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.5c01638 · Cited by 2
An attention-based approach for Koopman modeling and predictive control of nonlinear systems.
Wang, M., Lou, X., Cui, B., & Suykens, J. A. K. (2025) · ISA transactions · doi:10.1016/j.isatra.2025.04.011 · Cited by 1
Karasawa, N., Mitsutake, A., & Takano, H. (2024) · Journal of Chemical Physics · doi:10.1063/5.0211504 · Cited by 1
Weigle, A. T., & Shukla, D. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.06.12.598708 · Cited by 2
“Grid search MSMs were evaluated based off their VAMP-2 score and the absolute difference between the eigenvalues of the second and third timescale dominant processes (t 2 -t 3 ) (60, 61).”
MethodTime-Delayed Koopman Network-Based Model Predictive Control for the FRIB RFQ
Wan, J., Zhao, S.-L., Chang, W., & Hao, Y. (2024) · doi:10.48550/arxiv.2401.11031 · Cited by 1
“Numerical methods, such as Dynamic Mode Decomposition (DMD) [20], extended DMD (eDMD) [21], and variational approach for conformation dynamics (VAC) [22], are often used to approximate the Koopman operator.”
MethodImproving Estimation of the Koopman Operator with Kolmogorov-Smirnov Indicator Functions.
Ngo, V. A., Lin, Y., & Perez, D. (2023) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.3c00632 · Cited by 1
Estimation of the Koopman Generator by Newton's Extrapolation
Sechi, R., Sikorski, A., & Weber, M. (2021) · Multiscale Modeling and Simulation · doi:10.1137/20m1333006 · Cited by 3
A Markov state modeling analysis of sliding dynamics of a 2D model
Teruzzi, M., Pellegrini, F., Laio, A., & Tosatti, E. (2017) · The Journal of Chemical Physics · doi:10.1063/1.4986458 · Cited by 2
Data-Driven Model Reduction and Transfer Operator Approximation
Klus, S., Nüske, F., Koltai, P., Wu, H., Kevrekidis, I., Schütte, C., & Noé, F. (2018) · Journal of Nonlinear Science · doi:10.1007/s00332-017-9437-7 · Cited by 263
“The variational approach of conformation dynamics (VAC) (Noé and Nüske 2013; Nüske et al. 2014, 2016) is a generalization of the frequently used Markov state modeling framework that allows arbitrary basis functions and is similar to the variational approach in quantum mechanics (Nüske et al.”
BackgroundMethodInfluentialSelf-citationXu, Y.-C., Valva, C., Gao, B., & Ishikawa, I. (2026) · doi:10.48550/arxiv.2610.39580
“Variational principles guide the choice of the approximation [44, 48, 49, 69], residual and invariance criteria assess its accuracy [18, 19, 25], and the resulting a posteriori error bounds have been minimized to select dictionary and kernel parameters [19].”
Elucidating Lithium-Ion Dynamics in Highly Concentrated Electrolytes with Markov State Models.
Wang, T., & Yan, T.-Y. (2026) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.6c01341
LiveCoMS Roundtable: Discussing Markov state models after two decades of development v1.0Preprint
Zuckerman, D. M., Bowman, G. R., Caflisch, A., Dinner, A. R., Fadhluddin, A., Giangreco, G., Huang, X., Keller, B. G., Klus, S., Nüske, F., et al. (2026) · PubMed Central · doi:10.48550/arxiv.2610.24144
“In the following decade, the powerful variational principle known as VAMP (Variational Approach to Markov Processes) was developed to select models that optimally capture molecular kinetics [154, 234].”
Self-citationBilevel optimization for data-driven learning of Koopman embeddings using kernel-based autoencoders
N'konzi, J. N., Nüske, F., & Klus, S. (2026) · doi:10.48550/arxiv.2610.12370
“This includes in particular the extended dynamic mode decomposition (EDMD) algo-rithm 6,7 and the variational approach to conformation dynamics (VAC) 10,11 .”
InfluentialSelf-citation
Metadata
- Authors
- Noé, F., & Nüske, F.
- Venue
- Multiscale Modeling & Simulation
- Published
- 2013-01
- Volume
- 11
- Issue
- 2
- Pages
- 635-655
- ISSN
- 1540-3459 (ISSN-L)
- 1540-3467
- OpenAlex
- W2089842282
- Cited by
- 327
- References
- 50
- Citations by year
- 2026: 17 · 2025: 23 · 2024: 28 · 2023: 27 · 2022: 28 · 2021: 30 · 2020: 23 · 2019: 49 · 2018: 34 · 2017: 23 · 2016: 12 · 2015: 13 · 2014: 8 · 2013: 8
- FWCI
- 11.91
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Multiscale Modeling and Simulation (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Deutsche Forschungsgemeinschaft
- Freie Universität Berlin
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Markov Chains and Monte Carlo Methods
- Physical Sciences › Mathematics › Mathematical Physics › Stochastic processes and statistical mechanics
- Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Model Reduction and Neural Networks
- Keywords
- slow processes, variational principle, Rayleigh coefficient, sampling problems, distributed simulation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.