Abstract
Density functional theory (DFT) is one of the most widely used tools to solve the manybody Schrodinger equation. The core uncertainty inside DFT theory is the exchangecorrelation (XC) functional, the exact form of which is still unknown. Therefore, the essential part of DFT success is based on the progress in the development of XC approximations. Traditionally, they are built upon analytic solutions in low and highdensity limits and result from quantum Monte Carlo numerical calculations. However, there is no consistent and general scheme of XC interpolation and functional representation. Many different developed parametrizations mainly utilize a number of phenomenological rules to construct a specific XC functional. In contrast, the neural network (NN) approach can provide a general way to parametrize an XC functional without any a priori knowledge of its functional form. In this work, we develop NN XC functionals and prove their applicability to 3dimensional physical systems. We show that both the local density approximation (LDA) and generalized gradient approximation (GGA) are well reproduced by the NN approach. It is demonstrated that the local environment can be easily considered by changing only the number of neurons in the first layer of the NN. The developed NN XC functionals show good results when applied to systems that are not presented in the training/test data. The generalizability of the formulated NN XC framework leads us to believe that it could give superior results in comparison with traditional XC schemes provided training data from highlevel theories such as the quantum Monte Carlo and postHartreeFock methods.
Similar content being viewed by others
Introduction
Since its initial formulation, density functional theory (DFT)^{1,2} has become one of the main methods to solve the manybody Schrodinger equation. The source of the major ambiguity in DFT theory is the lack of a universal form of the exchangecorrelation (XC) functional. Therefore, the success of DFT is mostly determined by the improvement of XC representations. The seminal Monte Carlo (MC) simulations of Ceperley and Adler^{3} have resulted in a number of practical local density approximations (LDAs)^{4,5,6}. The next major progress was produced by the introduction of the generalized gradient approximation (GGA), which takes into account local gradients of electron density^{7,8,9}. This greatly enhanced the ability of DFT to describe systems with inhomogeneous electron densities, such as molecules and surfaces. A number of other functionals have been suggested with various degrees of complexity, extending DFT to a broad class of material properties^{10}. The seeking for better functionals is still a very active area of research^{11,12,13,14}.
Despite the obvious success of the LDA and GGA, their construction is a highly complicated process that includes many heuristics steps. Even the XC functional form itself is based on some assumptions about the local nature of XC interactions. A large number of analytical conditions to fulfill^{9} and a commitment to the local functional form make it difficult to incorporate new numerical results from highlevel quantum mechanics theories. For example, only numerical data from MC simulations of homogeneous electron gas are used to produce LDA^{6} and GGA^{9} functionals. Therefore, much of the progress in quantum MC^{15,16} and postHartreeFock^{17} methods is not adopted by XC approximations. One possible solution to overcome this problem is to use neural networks (NNs) for the interpolation of XC functionals. The NN provides a general way to represent any functional relationship^{18} and a highly flexible framework to embody both analytical and numerical training data.
There have been a number of studies in which NNs have been used to describe the potential energy surface to speed up electronic structure calculations^{19,20,21,22,23,24,25,26}, while fewer studies have been devoted to developing NN XC functionals^{27,28,29,30}. NN was firstly utilized to approximate XC potential by Tozer et al.^{31}. This study showed that NN approach is flexible and principally applicable to generate XC potentials to be used in the real physical systems DFT calculations.
In this work, we focus on constructing NN XC functionals for 3D systems without any feature engineering. We show that even rather simple NN architectures reproduce LDA and GGA XC potentials with satisfactory accuracy. To construct the LDA, potential pointtopoint mapping is used due to the local idea behind the LDA itself. GGA potential regiontopoint mapping is achieved by increasing the number of input neurons to take into account the local environment.
This paper is outlined as follows: In methods section, we describe the data generation process, NN topologies and hyperparameters. In results section, we show that the constructed NN XC potentials have the capability to reproduce LDA and GGA results for real physical systems. In conclusion section, we summarize our results and consider future research that could be done with the proposed framework, with a particular emphasis on incorporating data from highlevel methods such as quantum Monte Carlo and postHartreeFock.
Methods
To obtain the training data needed to build the NN XC potential, we carry out calculations of threedimensional (3D) electron gases in a framework of the KohnSham formalism^{1,2} for potentials of a simple harmonic oscillator (SHO). This potential takes the form:
where k_{i} is the spring constant and r = (x_{1}, x_{2}, x_{3}) is the radius vector. Parameter k_{i} pertains to a simple harmonic oscillator randomly sampled from intervals 0.01 ≤ k_{i} ≤ 0.9 Ha/\({{\rm{a}}}_{0}^{2}\), where a_{0} is 1 Bohr radius.
The XC potential is evaluated on a 40 × 40 × 40 a_{0} parallelepiped with 32 × 32 × 32 mesh, which corresponds to a spacing of 1.25 a_{0}. Three electrons are placed in this 3D space. For each chosen parameter, DFT calculations in real space were performed using the Octopus code^{32,33,34}. The parametrizations of the LDA exchangecorrelation functional used include the Slater exchange^{35,36} and PerdewZunger (modified) correlation part^{5}. The GGA parameters used include the exchange and correlation parts from^{9,37}. The total number of DFT calculations performed is 150 for each XC parametrization. The ranges of electron density represented by the WignerSeitz radius r_{s} are [1; 5 × 10^{6}] and [1; 10^{8}] for LDA and GGA, respectively. One calculation then provides a 32 × 32 × 32 dataset size for LDA and a 28 × 28 × 28 size for GGA. These yield overall dataset sizes of approximately 5.0 million and 3.3 million for LDA and GGA, respectively, but only 3.3 million samples are used for both functionals. 15% of all datasets are used for the test set.
The composition of Tensorflow^{38} and TFLearn^{39} is utilized for training the NN. The Adam algorithm with a learning rate of 0.001 is used for training, with MSE loss. The neural architecture used for the LDA is a fully connected network with one input neuron, one linear output neuron and two ReLU^{40} hidden layers with 1000 and 500 neurons (see Fig. 1). The only difference in the GGA case is that it takes into account the contribution of the local environment, which corresponds to an increase in the number of neurons in the first layer to 125 (5 × 5 × 5 cube). It should be noted that in principle the local environment could be taken into account explicitly by passing gradient and higher order derivatives of electron density. We delegate a computation of electron density derivatives to NN itself, making preprocessing step easy and convenient. Considered 5 × 5 × 5 input cube gives information to calculate gradient and Jacobian matrix with O(Δr^{2}) and O(Δr) precision correspondingly, where Δr is mesh spacing.
Results and Discussion
Multiple neural architectures, including different activation functions, numbers of layers and numbers of neurons, are tested. We choose robust and rather simple architectures, namely, the pointtopoint and regiontopoint architectures described above. We also test NNs with only one hidden layer with 1000 neurons, which have poor generalizability and result in a larger loss in comparison with the topology used. It should be noted that the topology chosen is heuristics, and it is possible that another number of hidden layers and neurons could give better results than those provided in this study. The search for the optimal NN topology is a large task that needs to be carried out in future studies.
Because the density range contains mostly small values, base10 logarithmic transformation is chosen for preprocessing the dataset. This scaling greatly improves the convergence speed of the backpropagation algorithm during the NN training cycle. No other preprocessing techniques and feature engineering are applied to the input data.
Training of the LDA NN is stopped after 1300 epochs. The NN LDA mean absolute error (MAE) is 0.03 mHa × Bohr^{3}. The mean minimum and maximum values of V_{xc} averaged over all configurations used for the LDA NN construction are 0.73 and 510 mHa × Bohr^{3}, respectively. Training of the GGA NN is terminated after 320 epochs. The NN GGA mean absolute error (MAE) is 0.156 mHa × Bohr^{3}. The mean minimum and maximum values of Vxc averaged over all configurations used for the GGA NN construction are 0 and 553 mHa × Bohr^{3}, respectively. The results of the training for both the LDA and GGA NNs are summarized in Table 1.
The results of applying an NN to the electron density from the test set are presented in Fig. 2. Both XC potentials provided by the LDA and GGA NNs are in good agreement with the numerical calculations made by Octopus. The special features of the V_{xc} form, for example, the “two humps”, are also well reproduced by them. For LDA it is also possible to provide comparison between \({V}_{xc}^{NN}({r}_{s})\) and analytical reference function V_{xc}(r_{s}). Plotted curves are indistinguishable in the range of r_{s} from 1 to 15, the absolute maximum difference is 1.7 mHa × Bohr^{3} (see Fig. S1 of supplementary materials).
To test the NN XCs for real physical systems, we consider Si in a diamond structure with 8 basis atoms in a cubic cell (lattice constant: 10.2 a_{0}). The numerical calculation is performed on a 10.2 × 10.2 × 10.2 a_{0} parallelepiped with a 14 × 14 × 14 mesh, which corresponds to a spacing of 0.729 a_{0}. The NN LDA and NN GGA MAEs are 0.6 and 18.1 mHa × Bohr^{3}, respectively. The NN LDA results are in excellent agreement with the reference data obtained from Octopus. The NN GGA describes V_{xc} reasonably well, except for regions with high density variations (see Fig. 3). We believe that this is because of the different spacings of the training/test data, which lead to the change in electron density gradients perceived by the NN XC.
The NN XC potentials are also tested on slices of 3D data in the plane of the benzene molecule. The numerical calculation is performed on a 14 × 14 × 14 a_{0} parallelepiped with a 44 × 44 × 44 mesh, which corresponds to a spacing of 0.318 a_{0}. The NN LDA and NN GGA MAEs are 1.36 and 18.3 mHa × Bohr^{3}, respectively. The overall form of the XC potential obtained by Octopus is similar to the form produced by both NNs. In the case of benzene, the MAE is substantially larger than the MAE obtained during the training/test procedure. This is due to the high electron density values obtained for benzene, which are not presented in the training/test dataset. Moreover, the mesh spacing in the benzene calculation is smaller than that in the dataset, which is important for NN GGA potential evaluation. Despite the lack of appropriate input data, both NN XC potentials show an impressive ability to generalize the XC functional (see Fig. 4).
The results obtained from the developed NN XC potentials show that NN LDA works well, provided that the electron density values are presented in the training/test dataset. Based on Table 2, the NN LDA MAE is rather small for silicon and becomes larger for benzene, in which the electron density contains values that are outside the training/test density range. It is not surprising that the NN GGA (see Table 3) demonstrates a large MAE for both systems considered, as the mesh spacing is not presented in the training procedure. Electron density spatial information not included explicitly in the training set leads to the fact that neural networks can take gradients into account in a way that cannot be properly generalized. Changing the mesh corrupts the learned representations and makes it not quite correct to apply the NN GGA to systems on meshes other than the one learned.
To construct meshindependent nonlocal functionals such as the GGA, it is necessary to use the coordinates of mesh points at which the electron density is defined. This additional information will help the NN evaluate meshindependent gradients or other higherorder variations of electron density and include them in the process of the interpolation of the exchangecorrelation potential. The type of local coordinate system to use remains the subject of further research.
We also implemented both NN LDA and NN GGA potentials in the OCTOPUS package. The xcenergy was taken from original LDA^{5,35,36} and GGA^{9,37} functionals. A behaviour during selfconsistent cycle and an ability to perform geometry optimization were tested. Lattice constant of Si was scanned in range from 9.6 to 10.95 Bohr with a step 0.15 Bohr. Bond length of a benzene molecule was changed in range from 2.37 to 2.90 Bohr with a step 0.053 Bohr. All performed selfconsistent calculations converged with density tolerance 0.5 × 10^{−5} Bohr^{−3}. In case of LDA dependencies of a total energy on a geometry parameter for both Si and the benzene molecule are in good agreement with results obtained for original xcpotentials (see Fig. 5). There is a notable shift to higher values for NN GGA dependencies of a total energy on a geometry parameter in comparison with one obtained from original GGA xcpotential. We attribute this differences to the fact that total xcenergy was not fully consistent with the NN xcpotential. In general, obtained results for selfconsistent calculations show that NN xcpotentials are converged and can be used in geometry optimization problems.
Conclusions
In this work, we developed, verified and tested a neuralnetwork approach to interpolating XC functionals. We show that the NN approach gives adequate results for both the LDA and GGA considered. When considering a model system based on three electrons placed in a parallelepiped box with an external potential of a 3D simple harmonic oscillator, the NN is in perfect agreement with numerical calculations from the test dataset. Moreover, the neural network shows a satisfactory error in real 3D physical systems not presented in the training/test data, such as silicon and benzene. Despite the lack of generalizability of the NN GGA for both systems, the analysis of those results led us to the idea of improving the NN approach by searching for a better type of coordinate system for proper representation of the input data. It is important to note that the neural network topology used in this work is very simple and has good potential for further improvements.
The main advantage of the NN approach in comparison with other interpolation techniques for XC functionals is its flexibility to incorporate exchangecorrelation data from different sources, such as postHartreeFock and quantum Monte Carlo. It is possible that application of the NN to interpolate highlevel XC quantum data could eliminate many heuristics used in the traditional construction of XC functionals.
Data availability
Optimized NN weights and code to use it are available at https://github.com/AlexanderFreeman/MLXC. Implementation of NN potentials into OCTOPUS package is available at https://github.com/AlexanderFreeman/octopus_nn.
References
Hohenberg, P. & Kohn, W. Inhomogeneous electron gas. Physical review 136, B864 (1964).
Kohn, W. & Sham, L. J. Selfconsistent equations including exchange and correlation effects. Physical review 140, A1133 (1965).
Ceperley, D. M. & Alder, B. Ground state of the electron gas by a stochastic method. Physical Review Letters 45, 566 (1980).
Vosko, S. H., Wilk, L. & Nusair, M. Accurate spindependent electron liquid correlation energies for local spin density calculations: a critical analysis. Canadian Journal of physics 58, 1200–1211 (1980).
Perdew, J. P. & Zunger, A. Selfinteraction correction to densityfunctional approximations for manyelectron systems. Phys. Rev. B 23, 5048–5079 (1981).
Perdew, J. P. & Wang, Y. Accurate and simple analytic representation of the electrongas correlation energy. Physical Review B 45, 13244 (1992).
Wang, Y. & Perdew, J. P. Spin scaling of the electrongas correlation energy in the highdensity limit. Physical Review B 43, 8911 (1991).
Perdew, J. P. et al. Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approximation for exchange and correlation. Physical review B 46, 6671 (1992).
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Physical review letters 77, 3865 (1996).
Mardirossian, N. & HeadGordon, M. Thirty years of density functional theory in computational chemistry: an overview and extensive assessment of 200 density functionals. Molecular Physics 115, 2315–2372 (2017).
Lani, G., Di Marino, S., Gerolin, A., van Leeuwen, R. & GoriGiorgi, P. The adiabatic strictlycorrelatedelectrons functional: kernel and exact properties. Physical Chemistry Chemical Physics 18, 21092–21101 (2016).
Maier, T. M., Haasler, M., Arbuznikov, A. V. & Kaupp, M. New approaches for the calibration of exchangeenergy densities in local hybrid functionals. Physical Chemistry Chemical Physics 18, 21133–21144 (2016).
MoriSánchez, P. & Cohen, A. J. The derivative discontinuity of the exchange–correlation functional. Physical Chemistry Chemical Physics 16, 14378–14387 (2014).
MoriSánchez, P. & Cohen, A. J. Exact density functional obtained via the levy constrained search. The journal of physical chemistry letters 9, 4910–4914 (2018).
Needs, R., Towler, M., Drummond, N. & Ríos, P. L. Continuum variational and diffusion quantum monte carlo calculations. Journal of Physics: Condensed Matter 22, 023201 (2009).
Kolorenč, J. & Mitas, L. Applications of quantum monte carlo methods in condensed systems. Reports on Progress in Physics 74, 026502 (2011).
Cremer, D. Møller–plesset perturbation theory: from small molecule methods to methods for thousands of atoms. Wiley Interdisciplinary Reviews: Computational Molecular Science 1, 509–530 (2011).
Cybenko, G. Approximation by superpositions of a sigmoidal function. Mathematics of control, signals and systems 2, 303–314 (1989).
Lorenz, S., Groß, A. & Scheffler, M. Representing highdimensional potentialenergy surfaces for reactions at surfaces by neural networks. Chemical Physics Letters 395, 210–215 (2004).
Behler, J. & Parrinello, M. Generalized neuralnetwork representation of highdimensional potentialenergy surfaces. Phys. Rev. Lett. 98, 146401 (2007).
Balabin, R. M. & Lomakina, E. I. Neural network approach to quantumchemistry data: Accurate prediction of density functional theory energies. The journal of chemical physics 131, 074104 (2009).
Handley, C. M. & Popelier, P. L. Potential energy surfaces fitted by artificial neural networks. The Journal of Physical Chemistry A 114, 3371–3383 (2010).
Behler, J. Neural network potentialenergy surfaces in chemistry: a tool for largescale simulations. Physical Chemistry Chemical Physics 13, 17930–17955 (2011).
Schütt, K. et al. Schnet: A continuousfilter convolutional neural network for modeling quantum interactions. In Advances in Neural Information Processing Systems, 991–1001 (2017).
Schütt, K. T., Arbabzadah, F., Chmiela, S., Müller, K. R. & Tkatchenko, A. Quantumchemical insights from deep tensor neural networks. Nature communications 8, 13890 (2017).
Xie, C., Zhu, X., Yarkony, D. R. & Guo, H. Permutation invariant polynomial neural network approach to fitting potential energy surfaces. iv. coupled diabatic potential energy matrices. The Journal of chemical physics 149, 144107 (2018).
Nagai, R., Akashi, R., Sasaki, S. & Tsuneyuki, S. Neuralnetwork kohnsham exchangecorrelation potential and its outoftraining transferability. The Journal of chemical physics 148, 241737 (2018).
Lei, X. & Medford, A. J. Design and analysis of machine learning exchangecorrelation functionals via rotationally invariant convolutional descriptors. Physical Review. Materials 3, 063801 (2019).
Nagai, R., Akashi, R. & Sugino, O. Completing density functional theory by machinelearning hidden messages from molecules. arXiv preprint arXiv:1903.00238 (2019).
Ramos, P. & Pavanello, M. Static correlation density functional theory. arXiv preprint arXiv:1906.06661 (2019).
Tozer, D. J., Ingamells, V. E. & Handy, N. C. Exchangecorrelation potentials. The Journal of chemical physics 105, 9200–9213 (1996).
Andrade, X. et al. Realspace grids and the octopus code as tools for the development of new simulation approaches for electronic systems. Physical Chemistry Chemical Physics 17, 31371–31396 (2015).
Andrade, X. et al. Timedependent densityfunctional theory in massively parallel computer architectures: the octopus project. Journal of Physics: Condensed Matter 24, 233202 (2012).
Andrade, X. & AspuruGuzik, A. Realspace density functional theory on graphical processing units: computational approach and comparison to gaussian basis set methods. Journal of chemical theory and computation 9, 4360–4373 (2013).
Dirac, P. A. M. Note on exchange phenomena in the thomas atom. Mathematical Proceedings of the Cambridge Philosophical Society 26, 376–385 (1930).
Bloch, M. F. Bemerkung zur elektronentheorie des ferromagnetismus und der elektrischen leitf¨ahigkeit. Zeitschrift für Physik 57, 545–555 (1929).
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple [phys. rev. lett. 77, 3865 (1996)]. Phys. Rev. Lett. 78, 1396–1396 (1997).
Abadi, M. et al. Tensorflow: A system for largescale machine learning. In 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI 16), 265–283 (2016).
Tang, Y. Tf. learn: Tensorflow’s highlevel module for distributed machine learning. arXiv preprint arXiv:1612.04251 (2016).
Nair, V. & Hinton, G. E. Rectified linear units improve restricted boltzmann machines. In Proceedings of the 27th international conference on machine learning (ICML10), 807–814 (2010).
Author information
Authors and Affiliations
Contributions
A.R. and P.Z. performed research. A.R. and P.Z. wrote the paper. P.Z and I.A. supervised the project. All authors reviewed the manuscript
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ryabov, A., Akhatov, I. & Zhilyaev, P. Neural network interpolation of exchangecorrelation functional. Sci Rep 10, 8000 (2020). https://doi.org/10.1038/s41598020646198
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598020646198
This article is cited by

Application of twocomponent neural network for exchangecorrelation functional interpolation
Scientific Reports (2022)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.