Solving the nonlinear Schrodinger equation with an unsupervised neural network.

We solve the nonlinear Schrodinger equation with an unsupervised neural network with the optical axis position z and time t as inputs. The network outputs the real and imaginary components of the solution. Unsupervised training aims to minimize a non-negative energy function derived from the equation and the boundary conditions. The trained network is generalizing - a solution value is determined at any (z, t)-combination including those not considered during training. Solutions with normalized mean-squared errors of order 10-2, are obtained when the average energy is reduced to 10-2 from order 104. The NN method is universal and applies to other complex differential equations.

[1]  C Saloma,et al.  Solving N-body problems with neural networks. , 2001, Physical review letters.

[2]  W. Ketterle,et al.  Observation of Feshbach resonances in a Bose–Einstein condensate , 1998, Nature.

[3]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[4]  M. Saffman.  Self-Induced Dipole Force and Filamentation Instability of a Matter Wave , 1998 .

[5]  Tomaso A. Poggio,et al.  A Sparse Representation for Function Approximation , 1998, Neural Computation.

[6]  May Lim,et al.  Direct signal recovery from threshold crossings , 1998 .

[7]  J. Satsuma,et al.  B Initial Value Problems of One-Dimensional self-Modulation of Nonlinear Waves in Dispersive Media (Part V. Initial Value Problems) , 1975 .

[8]  M Soriano,et al.  Fluorescent image classification by major color histograms and a neural network. , 2001, Optics express.

[9]  Tommy W. S. Chow,et al.  A weight initialization method for improving training speed in feedforward neural network , 2000, Neurocomputing.

[10]  C. Greene,et al.  Effective potentials for dilute Bose-Einstein condensates , 1998 .

[11]  C Saloma,et al.  Improved classification robustness for noisy cell images represented as principal-component projections in a hybrid recognition system. , 1998, Applied optics.

[12]  Zhi-Quan Luo,et al.  On the Convergence of the LMS Algorithm with Adaptive Learning Rate for Linear Feedforward Networks , 1991, Neural Computation.

[13]  Jiménez,et al.  Neural network differential equation and plasma equilibrium solver. , 1995, Physical review letters.

[14]  Christopher Monterola,et al.  Characterizing the dynamics of constrained physical systems with an unsupervised neural network , 1998 .

[15]  G. Arfken.  Mathematical Methods for Physicists , 1967 .

[16]  Saad,et al.  Exact solution for on-line learning in multilayer neural networks. , 1995, Physical review letters.

[17]  G. Mourou,et al.  Nonlinear Optics in Relativistic Plasmas and Laser Wake Field Acceleration of Electrons , 1996, Science.