A model for stability of the semi-implicit backward differentiation formulas (Q2277775)

The semi-implicit backward differentiation method for the Cauchy problem \(y'=f(x,y)\), \(y(a)=y_ a\) is considered. This paper extends the works of \textit{R. W. Klopfenstein} [RCA Rev. 32, 447-462 (1971)] and \textit{F. T. Krogh} and the author [ACM Trans. Math. Software 10, 45-57 (1984; Zbl 0555.65051)] which investigated a class of explicit algorithms involving a predictor followed by m applications of a derivative evaluation and a Newton-like process to compute the correction for the model test system \(y'=Jy\), \(y\in {\mathbb{R}}^ s\), where J is the Jacobian of the differential equation. The impact on stability of error in the matrix over a broad range of linear constant coefficient equations is modeled. The scalar case \(y'=Jy\) is discussed in detail to obtain valuable guidelines for practical codes. A technique is developed to enable the examination of families of \(\lambda\) values.