作者: A.D. Irving , T. Dewson
DOI: 10.1016/S0167-2789(96)00248-5
关键词: Multigrid method 、 Mathematics 、 Differential algebraic equation 、 Method of characteristics 、 Nonlinear system 、 Numerical partial differential equations 、 Collocation method 、 Stochastic partial differential equation 、 Mathematical analysis 、 Exponential integrator
摘要: Abstract A new method is described for extracting mixed linear-nonlinear coupled differential equations from multivariate discrete time series data. It assumed in the present work that solution of ordinary can be represented as a Volterra functional expansion. tractable hierarchy moment generated by operating on suitably truncated The facilitates calculation coefficients equations. In order to demonstrate method's ability accurately estimate governing equations, it applied data derived numerical Lorenz with additive noise. then used construct dynamic global mid- and high-magnetic latitude ionospheric model where nonlinear phenomena such period doubling quenching occur. shown estimated inhomogeneous second-order equation foF2 peak plasma density forecast future behaviour set ionosonde stations which encompass earth. Finally, portfolio Japanese common stock prices. characterise observed wide class linear phenomena.