Fortran vs. Python: An Unexpected Finding in Teaching Scientific Computing to Engineering Students

A professor who has taught an introductory scientific computing class for over 10 years using Python, questions whether Python is the optimal choice for beginners, especially in numerical linear algebra. He argues that a strongly-typed language like Fortran might offer a better learning experience. Python's flexibility, while convenient, leads to common student errors: confusion with the numpy library, indentation errors, and off-by-one errors stemming from 0-based indexing. In contrast, Fortran's strong typing and explicit loop structures help students better understand algorithms and reduce errors. The article uses the Jacobi method and QR decomposition as examples, comparing Python and Fortran implementations and arguing for Fortran's potential advantages in teaching, despite Python's broader ecosystem and versatility.