Locating special events when solving ODEs (Q1104465)

Computing the solution, \(y\in R^ n\), of the initial value problem in ordinary differential equations (ODEs), (1) \(y'=f(x,y)\), \(a\leq x\leq b\), \(y(a)=y_ a\) may be only part of a larger task. Possibly the most common such task is to find either a first point \(t_ 0>a\) or a set of points \(\{t_ s\}\), \(a<t_ 0\leq t_ 1\leq...\), such that one of the equations (2) \(g_ j(x,y(x),y'(x))=0\) \(j=1,2,...\), is satisfied at \(x=t_ s\). The \(g_ j\) are called ``event functions'', and event j is said to occur at \(t_ s\) when \(t_ s\) is a root of the jth event function. This note is concerned with problems that have the form either (3) \(g_ j(x,y,y')\equiv y_{k_ j}(x)-\alpha_ j\) or (4) \(g_ j(x,y,y')\equiv y'_{k_ j}(x),\) where \(y_{k_ j}(x)\) is a component of the solution vector y(x). Common problems such as (i) finding where a component of the solution assumes a given value, and (ii) finding where a component of the solution has an extremum, have the form (3) or (4) with a single event function. We allow several event functions of both forms at the same time, and so can solve more complicated problems such as (iii) tabulating values of a dependent variable \(y_ j\); (iv) determining the location of switching points or points of discontinuity defined in terms of linear functions of a dependent variable \(y_ j\); and (v) determining zeros of a general event function \(q(x,y)=0\) by adjoining a differential equation for q to the system (1).