The author outlines the events occurring during stellar evolution and the current problems that are being considered. He then describes how computers are being increasingly used in astrophysics, and how the numerical solution of differential equations is being replaced by the solution of difference equations which obey the appropriate conservation laws on a discrete grid. The large size and time-scale of most astrophysical problems often need the largest possible supercomputer available.
Like others users, astrophysicists need friendly software, higher level languages using parallelism, portability, at least local ability to test and debug algorithms on realistic problems, and extensive graphics. “Astrophysicists can rarely afford the luxury of computing in a ‘production’ mode where the same program is run and rerun with different sets of input data.” The author ends by giving a list of further reading.
This short paper may well be used in the introductory section of courses in numerical analysis. The author points out various deficiencies of the “commute to compute” procedure (e.g., the lack of truly standard software) that are common to many others users.