The error of Euler's method for floating-point arithmetic computations

We present an algorithm that enables us to find the optimal number of steps in Euler's method, in the sense of computational precision, while solving a Cauchy problem for a system of linear differential equations with constant coefficients. We include numerical examples of applications of this method for evaluating a solution to the Cauchy problem at a point and constructing solutions to systems of nonlinear ordinary differential equations.