Hardware engineers may think they are more serious, but how many hardware engineers take the time to verify the numerics of the finite element, finite differenc

Hardware engineers may think they are more serious, but how many hardware engineers take the time to verify the numerics of the finite element, finite difference, or spectral method codes they rely upon for computational electromagnetics, fluid dynamics, quantum or classical mechanics modeling that they often are fully reliant upon to do their jobs. As someone who has spent a career writing tools for internal teams at hardware shops, or at software companies that cater to hardware engineers, I have seen far more rigor in my technical computing software colleagues than I have seen from my hardware engineering colleagues. Most of my hardware engineering colleagues have had no idea what version control is, don’t know or care how floating point calculations can wreak havoc on their models, and can’t craft anything more complex than a straight line Matlab script. Software written by hardware engineers is often atrociously bad, has no tests, and can only run on the machine where it was written. I know I can’t even consider ever asking most of my hardware design engineering colleagues to touch a C compiler or validate the numerics of their models. They just as often prefer to rely upon their magical point and click GUIs or back of the envelope spherical cow based scripts and spreadsheets.