Active learning for saddle point calculation. (English) Zbl 1503.62066
Summary: The saddle point (SP) calculation is a grand challenge for computationally intensive energy function in computational chemistry area, where the saddle point may represent the transition state. The traditional methods need to evaluate the gradients of the energy function at a very large number of locations. To reduce the number of expensive computations of the true gradients, we propose an active learning framework consisting of a statistical surrogate model, Gaussian process regression (GPR) for the energy function, and a single-walker dynamics method, gentle accent dynamics (GAD), for the saddle-type transition states. SP is detected by the GAD applied to the GPR surrogate for the gradient vector and the Hessian matrix. Our key ingredient for efficiency improvements is an active learning method which sequentially designs the most informative locations and takes evaluations of the original model at these locations to train GPR. We formulate this active learning task as the optimal experimental design problem and propose a very efficient sample-based sub-optimal criterion to construct the optimal locations. We show that the new method significantly decreases the required number of energy or force evaluations of the original model.
MSC:
| 62K05 | Optimal statistical designs |
| 62G08 | Nonparametric regression and quantile regression |
| 62L05 | Sequential statistical design |
| 65D15 | Algorithms for approximation of functions |
References:
| [1] | Adler, R.J.: The Geometry of Random Fields. Classics in Applied Mathematics. Society for Industrial and Applied Mathematics (SIAM, 3600 Market Street, Floor 6, Philadelphia, PA 19104) (1981) · Zbl 0478.60059 |
| [2] | Branduardi, D.; Gervasio, FL; Parrinello, M., From a to b in free energy space, J. Chem. Phys., 126, 5 (2007) · doi:10.1063/1.2432340 |
| [3] | Chen, X.; Jørgensen, MS; Li, J.; Hammer, B., Atomic energies from a convolutional neural network, J. Chem. Theory Comput., 14, 7, 3933-3942 (2018) · doi:10.1021/acs.jctc.8b00149 |
| [4] | Cover, TM; Thomas, JA, Elements of Information Theory (2006), New York: Wiley-Interscience, New York · Zbl 1140.94001 |
| [5] | Crippen, GM; Scheraga, HA, Minimization of polypeptide energy: XI. The method of gentlest ascent, Arch. Biochem. Biophys., 144, 2, 462-466 (1971) · doi:10.1016/0003-9861(71)90349-3 |
| [6] | Denzel, A.; Kastner, J., Gaussian process regression for transition state search, J. Chem. Theory Comput., 14, 11, 5777-5786 (2018) · doi:10.1021/acs.jctc.8b00708 |
| [7] | Gao, W.; Leng, J.; Zhou, X., An iterative minimization formulation for saddle point search, SIAM J. Numer. Anal., 53, 4, 1786-1805 (2015) · Zbl 1321.65097 · doi:10.1137/130930339 |
| [8] | Gao, W.; Leng, J.; Zhou, X., Iterative minimization algorithm for efficient calculations of transition states, J. Comput. Phys., 309, 69-87 (2016) · Zbl 1351.82037 · doi:10.1016/j.jcp.2015.12.056 |
| [9] | Gu, S.; Zhou, X., Multiscale gentlest ascent dynamics for saddle point in effective dynamics of slow-fast system, Commun. Math. Sci., 15, 2279-2302 (2017) · Zbl 1453.60129 · doi:10.4310/CMS.2017.v15.n8.a7 |
| [10] | Gu, S.; Zhou, X., Simplified gentlest ascent dynamics for saddle points in non-gradient systems, Chaos, 28 (2018) · Zbl 1408.37050 · doi:10.1063/1.5046819 |
| [11] | Guan, Y.; Yang, S.; Zhang, D., Construction of reactive potential energy surfaces with Gaussian process regression: active data selection, Mol. Phys., 116, 7-8, 823-834 (2018) · doi:10.1080/00268976.2017.1407460 |
| [12] | Giuseppe, R., Dilek, H., Allen, G.: Active learning for automatic speech recognition. In: 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 4, pp. IV-3904-IV-3907 (2002) |
| [13] | Henkelman, G.; Jónsson, H., A dimer method for finding saddle points on high dimensional potential surfaces using only first derivatives, J. Chem. Phys., 111, 15, 7010-7022 (1999) · doi:10.1063/1.480097 |
| [14] | Huan, X.; Marzouk, YM, Simulation-based optimal Bayesian experimental design for nonlinear systems, J. Comput. Phys., 232, 1, 288-317 (2013) · doi:10.1016/j.jcp.2012.08.013 |
| [15] | Jónsson, H., Mills, G., Jacobsen, K.W.: Nudged elastic band method for finding minimum energy paths of transitions. Citeseer (1998) |
| [16] | Khorshidi, A.; Peterson, A., Amp: A modular approach to machine learning in atomistic simulations, Comput. Phys. Commun., 207, 310-324 (2016) · doi:10.1016/j.cpc.2016.05.010 |
| [17] | Koistinen, O.; Dagbjartsdóttir, FB; Ásgeirsson, V.; Vehtari, A.; Jónsson, H., Nudged elastic band calculations accelerated with Gaussian process regression, J. Chem. Phys., 147 (2017) · doi:10.1063/1.4986787 |
| [18] | Li, Q.; Lin, B.; Ren, W., Computing committor functions for the study of rare events using deep learning, J. Chem. Phys., 151, 5 (2019) · doi:10.1063/1.5110439 |
| [19] | Lin, Q.; Zhang, L.; Zhang, Y.; Jiang, B., Searching configurations in uncertainty space: Active learning of high-dimensional neural network reactive potentials, J. Chem. Theory Comput., 17, 5, 2691-2701 (2021) · doi:10.1021/acs.jctc.1c00166 |
| [20] | Lin, Q.; Zhang, Y.; Zhao, B.; Jiang, B., Automatically growing global reactive neural network potential energy surfaces: A trajectory-free active learning strategy, J. Chem. Phys., 152, 15 (2020) · doi:10.1063/5.0004944 |
| [21] | Maragliano, L.; Fischer, A.; Vanden-Eijnden, E.; Ciccotti, G., String method in collective variables: minimum free energy paths and isocommittor surfaces, J. Chem. Phys., 125, 2, 24106 (2006) · doi:10.1063/1.2212942 |
| [22] | Mitchell, T., Jones, R., Ghani, R., Rilo, E.: Active learning for information extraction with multiple view feature sets. In: Proceedings of the ECML-2004 Workshop on Adaptive Text Extraction and Mining (ATEM-2003) (2003) |
| [23] | Mousseau, N.; Barkema, GT, Traveling through potential energy surfaces of disordered materials: the activation-relaxation technique, Phys. Rev. E, 57, 2419-2424 (1998) · doi:10.1103/PhysRevE.57.2419 |
| [24] | Olsson, F.: A literature survey of active machine learning in the context of natural language processing. Technical Report (2009) |
| [25] | Pan, AC; Sezer, D.; Roux, B., Finding transition pathways using the string method with swarms of trajectories, J. Phys. Chem. B, 112, 11, 3432-3440 (2008) · doi:10.1021/jp0777059 |
| [26] | Pechukas, P., Transition state theory, Annu. Rev. Phys. Chem., 32, 1, 159-177 (1981) · doi:10.1146/annurev.pc.32.100181.001111 |
| [27] | Phillips, JC; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, RD; Kale, L.; Schulten, K., Scalable molecular dynamics with namd, J. Comput. Chem., 26, 16, 1781-1802 (2005) · doi:10.1002/jcc.20289 |
| [28] | Rasmussen, C.E.: Gaussian processes in machine learning. In: Summer School on Machine Learning, pp. 63-71. Springer (2003) · Zbl 1120.68436 |
| [29] | Ren, W.; Vanden-Eijnden, E., A climbing string method for saddle point search, J. Chem. Phys., 138 (2013) · doi:10.1063/1.4798344 |
| [30] | Ren, W.; Vanden-Eijnden, E.; Maragakis, P.; Weinan, E., Transition pathways in complex systems: Application of the finite-temperature string method to the alanine dipeptide, J. Chem. Phys., 123, 13 (2005) · doi:10.1063/1.2013256 |
| [31] | Riccardi, G.; Hakkani-Tur, D., Active learning: theory and applications to automatic speech recognition, IEEE Trans. Speech and Audio Process., 13, 4, 504-511 (2005) · doi:10.1109/TSA.2005.848882 |
| [32] | Samanta, A.; Wein, E., Atomistic simulations of rare events using gentlest ascent dynamics, J. Chem. Phys., 136 (2012) · doi:10.1063/1.3692803 |
| [33] | Settles, B.: Active Learning. Morgan & Claypool (2013) · Zbl 1270.68006 |
| [34] | Shane, C., Roman, G., Cynthia, L.: BASC: Applying Bayesian optimization to the search for global minima on potential energy surfaces. In: International Conference on Machine Learning, pp. 898-907. PMLR (2016) |
| [35] | Smith, JS; Nebgen, B.; Lubbers, N.; Isayev, O.; Roitberg, AE, Less is more: Sampling chemical space with active learning, J. Chem. Phys., 148, 24 (2018) · doi:10.1063/1.5023802 |
| [36] | Solak, E., Murray-Smith, R., Leithead, W., Leith, D., Rasmussen, C.: Derivative observations in Gaussian process models of dynamic systems. In: Becker, S., Thrun, S., Obermayer, K. (eds), Advances in Neural Information Processing Systems, Vol. 15. MIT Press, Cambridge (2002) |
| [37] | Spall, JC, Multivariate stochastic approximation using a simultaneous perturbation gradient approximation, IEEE Trans. Automat. Contr., 37, 3, 332-341 (1992) · Zbl 0745.60110 · doi:10.1109/9.119632 |
| [38] | Spall, JC, Implementation of the simultaneous perturbation algorithm for stochastic optimization, IEEE Trans. Aero. Elec. Syst., 34, 3, 817-823 (1998) · doi:10.1109/7.705889 |
| [39] | Torres, JAG; Jennings, PC; Hansen, MH; Boes, JR; Bligaard, T., Low-scaling algorithm for nudged elastic band calculations using a surrogate machine learning model, Phys. Rev. Lett., 122 (2019) · doi:10.1103/PhysRevLett.122.156001 |
| [40] | Truhlar, DG; Garrett, BC; Klippenstein, SJ, Current status of transition-state theory, J. Phys. Chem., 100, 31, 12771-12800 (1996) · doi:10.1021/jp953748q |
| [41] | Uteva, E.; Graham, RS; Wilkinson, RD; Wheatley, RJ, Active learning in Gaussian process interpolation of potential energy surfaces, J. Chem. Phys., 149, 17 (2018) · doi:10.1063/1.5051772 |
| [42] | Wang, H.; Lin, G.; Li, J., Gaussian process surrogates for failure detection: a Bayesian experimental design approach, J. Comput. Phys., 313, 247-259 (2016) · Zbl 1349.62360 · doi:10.1016/j.jcp.2016.02.053 |
| [43] | Wang, H., Zhou, X.: Explicit estimation of derivatives from data and differential equations by Gaussian process regression. Int. J. Uncertain. Quan., 11(4) (2021) · Zbl 1498.62073 |
| [44] | Weinan, E.; Ren, W.; Vanden-Eijnden, E., String method for the study of rare events, Phys. Rev. B, 66, 5 (2002) · Zbl 1023.65004 · doi:10.1103/PhysRevB.66.052301 |
| [45] | Weinan, E.; Zhou, X., The gentlest ascent dynamics, Nonlinearity, 24, 6, 1831 (2011) · Zbl 1237.37052 · doi:10.1088/0951-7715/24/6/008 |
| [46] | Yu, B.; Zhang, L., Global optimization-based dimer method for finding saddle points, Discrete Cont. Dyn-B., 26, 1, 741 (2021) · Zbl 1468.37057 |
| [47] | Zhang, L.; Du, Q.; Zheng, Z., Optimization-based shrinking dimer method for finding transition states, SIAM J. Sci. Comput., 38, 1, A528-A544 (2016) · Zbl 1332.37062 · doi:10.1137/140972676 |
| [48] | Zhang, L.; Lin, D.; Wang, H.; Car, R.; Weinan, E., Active learning of uniformly accurate interatomic potentials for materials simulation, Phys. Rev. Mater., 3, 2 (2019) · doi:10.1103/PhysRevMaterials.3.023804 |
| [49] | Zhang, L.; Wang, H.; Weinan, E., Reinforced dynamics for enhanced sampling in large atomic and molecular systems, J. Chem. Phys., 148, 12 (2018) · doi:10.1063/1.5019675 |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
