Using a neural network based method to solve the vibrational Schrödinger equation for H2O

A neural network (NN) algorithm is used to solve the vibrational Schrodinger equation for a molecule. Previous NN methods computed one level at a time, optimized all of the parameters using non-linear optimization methods, and were tested only on model potentials. Our approach combines non-linear optimization of neuron parameters with a linear matrix method. This improves dimensionality scaling and permits computing many levels. We use composite, flexible shape, radial basis function neurons. The algorithm avoids the calculation of integrals and of a potential energy function. We demonstrate that only a few dozen neurons are needed to compute five levels of water from a small set of potential points.

[1]  Michihiko Sugawara,et al.  NUMERICAL SOLUTION OF THE SCHRODINGER EQUATION BY A MICROGENETIC ALGORITHM , 2000 .

[2]  Kurt Hornik,et al.  Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks , 1990, Neural Networks.

[3]  Seung E. Choi,et al.  Highly excited vibrational eigenstates of nonlinear triatomic molecules. Application to H2O , 1992 .

[4]  T. Carrington,et al.  New ideas for using contracted basis functions with a Lanczos eigensolver for computing vibrational spectra of molecules with four or more atoms , 2002 .

[5]  Andrew R. Barron,et al.  Universal approximation bounds for superpositions of a sigmoidal function , 1993, IEEE Trans. Inf. Theory.

[6]  J. Tennyson,et al.  All the vibrational bound states of H+3 , 1990 .

[7]  Weitao Yang,et al.  The collocation method for bound solutions of the Schrödinger equation , 1988 .

[8]  T. Carrington,et al.  Calculation of triatomic vibrational eigenstates: Product or contracted basis sets, Lanczos or conventional eigensolvers? What is the most efficient combination? , 1994 .

[9]  Hua Guo,et al.  Discrete energy representation and generalized propagation of physical systems , 1998 .

[10]  P. Jensen.  The potential energy surface for the electronic ground state of the water molecule determined from experimental data using a variational approach , 1989 .

[11]  M Khauss,et al.  Potential Energy Surfaces , 1970 .

[12]  C. Chui,et al.  Approximation by ridge functions and neural networks with one hidden layer , 1992 .

[13]  D. Luckhaus,et al.  6D vibrational quantum dynamics: Generalized coordinate discrete variable representation and (a)diabatic contraction , 2000 .

[14]  T. Carrington,et al.  Using simultaneous diagonalization and trace minimization to make an efficient and simple multidimensional basis for solving the vibrational Schrodinger equation. , 2006, The Journal of chemical physics.

[15]  Vera Kurková,et al.  Kolmogorov's theorem and multilayer neural networks , 1992, Neural Networks.

[16]  John C. Light,et al.  Theoretical Methods for Rovibrational States of Floppy Molecules , 1989 .

[17]  Oriol Vendrell,et al.  Full-dimensional (15-dimensional) quantum-dynamical simulation of the protonated water dimer. I. Hamiltonian setup and analysis of the ground vibrational state. , 2007, The Journal of chemical physics.

[18]  Ken-ichi Funahashi,et al.  On the approximate realization of continuous mappings by neural networks , 1989, Neural Networks.

[19]  M. Mladenović.  Discrete variable approaches to tetratomic molecules: part II: application to H2O2 and H2CO. , 2002, Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy.

[20]  T. Carrington,et al.  The Utility of Constraining Basis Function Indices When Using the Lanczos Algorithm to Calculate Vibrational Energy Levels , 2001 .

[21]  B. Shore.  Comparison of matrix methods applied to the radial Schrödinger eigenvalue equation: The Morse potential , 1973 .

[22]  T. Carrington,et al.  Variational quantum approaches for computing vibrational energies of polyatomic molecules , 2008 .

[23]  Charles A. Micchelli,et al.  Dimension-independent bounds on the degree of approximation by neural networks , 1994, IBM J. Res. Dev..

[24]  N. Handy,et al.  The geometry and forcefield of acetylene , 2002 .

[25]  T. Carrington,et al.  Vibrational energy levels of CH5(+). , 2008, The Journal of chemical physics.

[26]  J. Light,et al.  Efficient distributed Gaussian basis for rovibrational spectroscopy calculations , 2000 .

[27]  Michihiko Sugawara,et al.  Numerical solution of the Schrödinger equation by neural network and genetic algorithm , 2001 .

[28]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[29]  Joel M Bowman,et al.  Full-dimensional quantum calculations of ground-state tunneling splitting of malonaldehyde using an accurate ab initio potential energy surface. , 2008, The Journal of chemical physics.

[30]  J. Light,et al.  On distributed Gaussian bases for simple model multidimensional vibrational problems , 1986 .

[31]  Shu-Cherng Fang,et al.  Relaxed conditions for radial-basis function networks to be universal approximators , 2003, Neural Networks.

[32]  Juan Ignacio Mulero-Martínez,et al.  Functions bandlimited in frequency are free of the curse of dimensionality , 2007 .

[33]  Bill Poirier,et al.  USING WAVELETS TO EXTEND QUANTUM DYNAMICS CALCULATIONS TO TEN OR MORE DEGREES OF FREEDOM , 2003 .

[34]  J. Light,et al.  Quasirandom distributed Gaussian bases for bound problems , 2001 .

[35]  Ah Chung Tsoi,et al.  Universal Approximation Using Feedforward Neural Networks: A Survey of Some Existing Methods, and Some New Results , 1998, Neural Networks.

[36]  D. Fotiadis,et al.  Artificial neural network methods in quantum mechanics , 1997, quant-ph/9705029.

[37]  Michael Elad,et al.  The Generalized Eigenvalue Problem for Nonsquare Pencils Using a Minimal Perturbation Approach , 2005, SIAM J. Matrix Anal. Appl..

[38]  Michael J Davis,et al.  Semiclassical Gaussian basis set method for molecular vibrational wave functions , 1979 .

[39]  Ernst Hairer,et al.  Solving Ordinary Differential Equations I: Nonstiff Problems , 2009 .

[40]  J. Bowman,et al.  A truncation/recoupling method for basis set calculations of eigenvalues and eigenvectors , 1991 .

[41]  C. Micchelli,et al.  Approximation by superposition of sigmoidal and radial basis functions , 1992 .

[42]  Kurt Hornik,et al.  Approximation capabilities of multilayer feedforward networks , 1991, Neural Networks.

[43]  T. Carrington,et al.  A general discrete variable method to calculate vibrational energy levels of three‐ and four‐atom molecules , 1993 .

[44]  T. Carrington,et al.  How to choose one-dimensional basis functions so that a very efficient multidimensional basis may be extracted from a direct product of the one-dimensional functions: energy levels of coupled systems with as many as 16 coordinates. , 2005, The Journal of chemical physics.