Continuation and the variational perturbation method (Q762893)

The author proposes a new approach to perturbation problems - which is named the continuation and variational perturbation method. The main idea is to construct recursively a sequence \(\{y(x,\epsilon_ i)\}\) corresponding to a sequence \(\{\epsilon_ i\}^{\infty}_{i=0}\) which tends to zero, with \(\epsilon_ 0\) for which the problem is satisfactorily solved. The sequence \(\{y(x,\epsilon_ i)\}\) is obtained from the equation of variations associated to the initial problem. The method is described in the simple cases: \(y'(x,\epsilon)=f(y,x,\epsilon)\), \(y(0,\epsilon)=a\), for which the variational equation is: \(s'(x,\epsilon_ i)=(\partial f/\partial y)_{\epsilon_ i}\quad s(x,\epsilon_ i)+(\partial f/\partial \epsilon)_{\epsilon_ i},\quad s(0,\epsilon_ i)=0;\) \(i=0,1,..\). with \(\epsilon_{iH}=\epsilon_ i-\Delta \epsilon_ i\). This method is discussed and the author himself remarks that, unfortunately, linear differential equations with variable coefficients and singular linear perturbation problems with variable coefficients are no good candidates for this new method because the variational equation in these cases are somewhat more complicated than the original equation. A list of non possibilities (Table I) and of possibilities (Table II) are then given. Numerical examples are elaborated with tables giving absolute relative error. Obviously the main interest of this method is that it is in fact a numerical technique and is well adapted for computer implementation, and so it is a good method for engineers. However the field of possibilities seems to be very particular - for instance for singular perturbations - but as the author says ''At this time the method is a plausible one. It needs to be put on a firm theoretical bases. In forthcoming papers we plan to establish its theoretical justification and systematically exploit its capabilities''.